Chemical Engineering August 2014

Michael Kleiber
Process Engineering
De Gruyter Graduate
Also of Interest
Process Technology: An Introduction
De Haan; 2015
ISBN 978-3-11-033671-9, e-ISBN 978-3-11-033672-6
Reactive and Membrane-Assisted Separations.

Lutze, Górak (Eds.); 2016
ISBN 978-3-11-030783-2, e-ISBN 978-3-11-030784-9
Chemical Reaction Technology.

Murzin; 2015
ISBN 978-3-11-033643-6, e-ISBN 978-3-11-033644-3
Advanced Process Engineering Control.

Agachi, Christea, Csavdári, Szilagyi; 2016
ISBN 978-3-11-030662-0, e-ISBN 978-3-11-030663-7
Corrosion for Engineers.

Zander, Dietzel, Atrens; 2016
ISBN 978-3-11-030666-8, e-ISBN 978-3-11-030687-3
Reviews in Chemical Engineering.

Luss, Brauner (Editors-in-Chief)
ISSN 0167-8299, e-ISSN 2191-0235
Michael Kleiber
Process
Engineering
|
Addressing the Gap between Studies
and Chemical Industry
Author
Dr.-Ing. Michael Kleiber
thyssenkrupp Industrial Solutions AG
Friedrich-Uhde-Str. 2
65812 Bad Soden
Germany
[email protected]
ISBN 978-3-11-031209-6
e-ISBN (PDF) 978-3-11-031211-9
e-ISBN (EPUB) 978-3-11-038310-2
Library of Congress Cataloging-in-Publication Data

A CIP catalog record for this book has been applied for at the Library of Congress.
Bibliographic information published by the Deutsche Nationalbibliothek

The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie;
detailed bibliographic data are available on the Internet at http://dnb.dnb.de.
© 2016 Walter de Gruyter GmbH, Berlin/Boston

Cover image: thyssenkrupp Industrial Solutions AG
Typesetting: PTP-Berlin, Protago-TEX-Production GmbH, Berlin
Printing and binding: CPI books GmbH, Leck
♾ Printed on acid-free paper
Printed in Germany
www.degruyter.com
|
For Claudia and our little Timon
Preface
You need only 10 % of the things you learn at the university.
The problem is: You don’t know which 10 %!
(University Sapience)
There are so many good textbooks on process engineering. When you start writing a
new one, you must wonder what might be its identifying feature, making it different
from other textbooks. In fact, I think I have got one.
After having worked in industry for twenty years in process simulation, design,
and development, the author is by no means in a position to give proper answers
to any problem. On the contrary, any new process has its own characteristic prob-
lems, and more or less you start from scratch. But experience and a good network
help to develop an appropriate strategy to get a solution, and to be able to distin-
guish between important and less important knowledge in process engineering. In the
academic world, there is a tendency for over-emphasizing theoretical concepts with-
out integration of the application aspects. For instance, many phase equilibrium and
physical property data specialists have never simulated a distillation column, which
would give a certain feeling for the importance of an activity coefficient at infinite di-
lution. On the other hand, practitioners have a tendency to believe in a solution which
worked once, disregarding that this might have been related to conditions which are
not always available. The corresponding pieces of software are simply trusted in any
case, while nobody can explain what they are based on and what their limitations are.
Bridging the gap between university and industry is the utmost concern of this
book. The intention is not to write a textbook for beginners in process engineering,
but to help the reader to be prepared with the most essential pieces of knowledge in
practical applications. It tries to answer the so-called silly questions, things that many
students have learnt at the university without understanding their implications.
The target of this book is not to generate specialists but to make the reader do
something reasonable and keep the overview. It is not a textbook which gives thorough
explanations for any topic listed in the book; for this purpose, some 400 pages are
by far not enough. In fact, long mathematical and scientific derivations are avoided,
and other existing textbooks are referred to where the reader can acquire an in-depth
knowledge if needed. Instead, we try to explain the meaning of the topics and formulas
so that the reader gets a feeling for the relationship and the interpretation. It should
enable the reader to take part in discussions and to know where it is worth increasing
his knowledge with further literature, and to distinguish between important and less
important topics.
To give an example: the author has often been asked to explain what an activity
coefficient is. People always expect at most two sentences, in the usual manner of engi-
neers. This is simply not possible. Good textbooks need several pages to explain phase
equilibria of pure substances, the difference between mixtures and pure substances,
VIII | Preface
the meaning of the Gibbs energy, and the concepts of excess and partial molar proper-
ties. Certainly, this explanation is a requirement for thermodynamists, but it is not the
way for application engineers to understand why he must use activity coefficients (or
an equivalent concept) for nonideal mixtures and how to get them. Instead, a “recipe”
for the usage of a model is required, and in many projects it is important to explain the
necessity of a proper evaluation of the model parameters, and to avoid that the project
manager gets the impression that it is just an accuracy fad.
The author is fully aware that the text reflects his own opinion. For example, equa-
tions of state are currently favored by most scientific authors. Nevertheless, the author
wants the reader to be capable, not perfect and at state of the art in each area. For this
purpose, activity coefficients are the more pragmatic approach and taken as the stan-
dard in this book.
The author is grateful to Jürgen Gmehling, Michael Benje, and Hans-Heinrich
Hogrefe, who eliminated many errors and misprints in my draft. Thanks also to
Hristina and Olaf Stegmann, who helped me a lot to write reasonable texts in Chap-
ters 1 and 11.
I would like the reader, probably a process engineer in his startup phase, to have
a better idea about the 10 % knowledge which will help him in his professional life.
And of course: always have fun in your job!
And remember:
There is always much more to learn than can ever be taught!

(Peter Ustinov)
A fool with a tool is still a fool.

(Grady Booch)
Contents
Preface | VII
1 Engineering projects | 1
1.1 Process engineering activities | 1
1.2 Realization of a plant | 8
1.3 Cost estimation | 18
2 Thermodynamic models in process simulation | 25

2.1 Phase equilibria | 27
2.2 φ-φ-approach | 32
2.3 γ-φ-approach | 41
2.3.1 Activity coefficients | 41
2.3.2 Vapor pressure and liquid density | 48
2.3.3 Association | 53
2.4 Electrolytes | 55
2.5 Liquid-liquid equilibria | 58
2.6 Solid-liquid equilibria | 61
2.7 φ-φ-approach with g E mixing rules | 63
2.8 Enthalpy calculations | 65
2.9 Model choice and data management | 68
2.10 Binary parameter estimation | 72
2.11 Model changes | 75
2.12 Transport properties | 75
3 Working on a process | 79
3.1 Flowsheet setup | 80
3.2 Heat integration options | 90
3.3 Batch processes | 97
3.4 Equipment design | 102
3.5 Troubleshooting | 103
4 Heat exchangers | 107

4.1 Something general | 107
4.2 Shell-and-tube heat exchangers | 109
4.3 Heat exchangers without phase change | 110
4.4 Condensers | 120
4.5 Evaporators | 122
4.6 Plate heat exchangers | 130
4.7 Double pipes | 132
X | Contents
4.8 Air coolers | 133

4.9 Fouling | 134
4.10 Vibrations | 136
5 Distillation and absorption | 139

5.1 Thermodynamics of distillation and absorption columns | 142
5.2 Packed columns | 144
5.3 Tray columns | 154
5.4 Comparison between packed and tray columns | 171
5.5 Distillation column control | 173
5.6 Constructive issues in column design | 179
5.7 Separation of azeotropic systems | 181
5.8 Rate-based approach | 184
5.9 Dividing wall columns | 186
5.10 Batch distillation | 188
6 Two liquid phases | 191

6.1 Liquid-liquid separators | 191
6.2 Extraction | 194
6.2.1 Mixer-settler arrangement | 197
6.2.2 Extraction columns | 198
6.2.3 Centrifugal extractors | 201
7 Alternative separation processes | 203

7.1 Membrane separations | 203
7.2 Adsorption | 210
7.3 Crystallization | 217
8 Fluid flow engines | 223

8.1 Pumps | 223
8.2 Compressors | 231
8.3 Jet pumps | 240
8.4 Vacuum generation | 242
9 Vessels and separators | 247
10 Chemical reactions | 255

10.1 Reaction basics | 255
10.2 Reactors | 262
11 Mechanical strength and material choice | 269

Contents | XI
12 Piping and measurement | 275

12.1 Pressure drop calculation | 275
12.1.1 Single-phase flow-through pipes | 275
12.1.2 Pressure drops in special piping elements | 280
12.1.3 Pressure drop calculation for compressible fluids | 280
12.1.4 Two-phase pressure drop | 283
12.2 Pipe specification | 289
12.3 Valves | 291
12.3.1 Isolation valves | 291
12.3.2 Control valves | 293
12.4 Measurement devices | 295
13 Utilities and waste streams | 299

13.1 Steam and condensate | 299
13.2 Heat transfer oil | 306
13.3 Cooling media | 306
13.4 Exhaust air treatment | 308
13.4.1 Condensation | 311
13.4.2 Combustion | 313
13.4.3 Absorption | 319
13.4.4 Biological exhaust air treatment | 320
13.4.5 Exhaust air treatment with membranes | 321
13.4.6 Adsorption processes | 322
13.5 Waste water treatment | 323
14 Process safety | 327

14.1 HAZOP procedure | 330
14.2 Pressure relief | 336
14.2.1 Introduction | 336
14.2.2 Mass flow to be discharged | 340
14.2.3 Fire case | 342
14.2.4 Actuation cases | 347
14.2.5 Safety valve peculiarities | 353
14.2.6 Maximum relief amount | 358
14.3 Explosions | 366
Glossary | 373
List of Symbols | 379
Bibliography | 383
XII | Contents
A Some numbers to remember | 393
B Pressure drop coefficients | 397
Index | 401
1 Engineering projects
An engineering project is a huge and complex task for usually several hundred people.
Coming from the university and just having finished one’s studies, one has usually no
clue on what is going on beyond the own desk. In fact, the construction of a chemical
plant is often compared to the erection of the pyramids in ancient Egypt. While the
weight of a chemical plant is much lower, its complexity is by far greater, and the
project can usually be completed in approx. three years instead of twenty. The target
for a beginner must be to become a increasingly larger cog in the machine. First, an
overview on the particular phases and activities must be obtained.
1.1 Process engineering activities
A plant always belongs to somebody. His target is to quickly earn money by producing
the substance the plant is designed for. At the beginning, a feasibility study has to be
done. A market analysis is performed, which hopefully shows that it is worth starting a
more detailed project. For a new process, it has to be checked whether it is possible to
overcome the technical difficulties. The legal situation with patents and licenses has
to be clarified, and possible locations for the plant are compared, whereby it is often
necessary to consider different energy prices or transport costs for raw materials and
products. A realistic production capacity and an impression of investment (CAPEX)
and operation costs (OPEX) must be available before starting a project (Chapter 1.3).
For the production capacity, it must be taken into account that no plant is in operation
all the time; usually, 8000 h per year are scheduled, giving approx. 90 % availability.
A corresponding overcapacity must be provided in the design.
It is important to know that an engineering project is not a sequential process,
where e.g. first the reactors are planned and finished, then the product purification,
and so on. This would actually be impossible, because due to recycle streams in the
process a complete engineering design of a single part of the plant could never be
achieved. Instead, all parts of the plant are worked out simultaneously, with increas-
ing accuracy and degree of detailing. The advantage is that possible bottlenecks and
difficulties are detected as early as possible, the interconnections are identified early,
and an appropriate number of project participants can be assigned to work on the
various parts of the plant. Certainly, this is not the way we are used to in everyday life,
and there is often doubt as to whether it makes sense to perform a design of a piece of
equipment while it is clear that the input streams are just preliminary and will change
several times during the project. Nevertheless, as mentioned, it is most important to
get an overview on the process as soon as possible. And with today’s tools, the design
from the previous phase is usually an ideal starting point when the preconditions have
been subject of change.
2 | 1 Engineering projects
The process engineer should know what the follow-up activities of his calcula-
tions are. The first phase in a project is the conceptual design, where the first mass
and energy balances are prepared, often based on lab trials and estimations.
The mass and energy balance is a key issue for all the following activities up to the
phase of detail engineering. A change in the mass balance has often a major impact on
all other participants of the project, so it is desirable to make it as exact as possible,
and to update it as soon as it makes sense. There is a certain misunderstanding as
to what a mass and energy balance really is. The term “process simulation” is very
common, and is also used here, but hardly applies. In fact, what the process does in
the steady state for a given set of inlet conditions is calculated, i.e. the streams and the
operating conditions of the particular pieces of equipment. Sometimes, the purpose
is in fact to find out how the plant or the equipment behaves, at least how it reacts,
and what the sensitivities are. However, in most cases, its purpose is to generate the
data for the design of the equipment, applying conservative cases concerning process
conditions or impurities. The exact process conditions that would enable the process
engineer to really “simulate” the plant are usually not known, at least not in the Con-
ceptual Design phase.
Despite these often occurring misunderstandings, “process simulation” is nowa-
days well acknowledged as a useful tool which requires a well-trained process engi-
neer who has a profound knowledge of the process itself, its thermodynamics (Chap-
ter 2), the various pieces of equipment and their peculiarities, and the simulation
experience, in order to achieve convergence in the simulation flowsheet, which of-
ten turns out to be complex. Nowadays, some well-established commercial (ASPEN,
HYSYS, ChemCAD, PRO/II, ProSim) and inhouse process simulators (Chemasim at
BASF, VTPlan at Bayer) are available, performing calculations that would have been
considered to be absolutely impossible 30 years ago. The genuine process simulation
showing the actual plant behavior with respect to the design of the equipment, the
startup-behavior, and the process control is called dynamic simulation. Nowadays,
its application becomes more and more popular, and conventional process simula-
tion can be used as a starting point for the dynamic version.
Sometimes, single process steps remain unknown and are represented in the mass
balance by simple split blocks. At least, there must be a concept of how to overcome
this lack of knowledge and what the effort might be. At the beginning of the basic
engineering these points should be completely clarified, and a full mass and energy
balance must be available. How this is done is the subject of Chapters 2 and 3. It is
desirable that pilot plant activities take place to confirm the mass balance and to make
sure of the influence of the recycle streams. The main purpose of such an activity is to
see whether all components are regarded and whether none of them accumulates in
the process.
The particular pieces of equipment are preliminarily designed according to the
current knowledge so that it becomes clear what the critical pieces of equipment are,
either because of their size or because of possible delivery limitations. As well, it must
be considered whether the plant can be operated at reduced or increased capacity,

which might be necessary for a certain period of time. Useful tools are the process
flow diagrams (PFD), where the whole process is visualized, including the main con-
trol loops (Figure 1.1). A PFD is a document to understand the process, operation data
for the important streams and blocks are usually included. The counterpart of the PFD
is process description, which describes the PFD in written form. It should not be exces-
sively detailed, as its main purpose is to enable the reader to understand the essentials
of the process. At the end of the conceptual design phase, equipment and operation
costs and hence the feasibility and their basis are better defined, often with respect to
a possible location.
Fig. 1.1: Example for the detailing in a PFD.
In a so-called HAZID (HAZard IDentification) the main issues concerning the safety
of the process are first discussed and listed, often with first recommendations. At a
later stage, the so-called HAZOP will take place, where all relevant safety issues are
discussed (Chapter 14.1). Finally, lists of utilities, raw products, auxiliary substances
(e.g. catalysts) and emissions (exhaust air, waste water, solid and organic wastes) are
issued. In the conceptual design phase, the design of the equipment can be done in
a preliminary way using rules of thumb. A first optimization of the process should be
performed. In process development, optimization is rarely a mathematical problem,
where an objective function is defined and somehow minimized. Process simulator
programs offer such a function; however, the author’s experience is that in most cases
process optimization cannot be translated into an objective function, as many soft fac-
tors have to be regarded (e.g. danger of fouling, increasing complexity, material issues,
ease of startup etc.). Equipment costs can be estimated by a process engineer as long
as only dimension changes are involved; however, it takes a specialist if the type or
the material of the equipment changes. The number of team members in the concep-
tual design phase is comparatively low, as the process engineering tasks are usually
complicated but of a limited extent. The complexity of the process development is
encountered by an iterative procedure, where many options are tested to achieve a
stepwise progress towards an improved process. There is no clear workflow plan; in-
stead, the creativity of the project members is decisive [1]. Nevertheless, it is desirable
to compose a comprehensible documentation to save the process knowledge which
was gained during the assessment of the various options. As there is no special struc-
ture available for this purpose, the documentation is done with a final report.
A successful and systematic way of optimizing a plant is the so-called value en-
gineering procedure. It starts with a brainstorming session, where all ideas of the
process team are formulated, collected, and clustered, either old ones as well as com-
pletely new ideas. In the following, these ideas are distributed back to the members
of the team. In a standardized procedure, the impact of an idea on CAPEX and OPEX
is carefully and comprehensibly evaluated, and the team can finally decide whether
the particular ideas are adopted or not.
In the basic engineering phase, the focus of the engineering moves from the pro-
cess design parameters like operating temperatures and pressures, flowrates or com-
positions to the geometric dimensions of the process equipment, design temperatures
and pressures (Chapter 11) and materials as parameters for the mechanical strength,
and the plant layout [2].
First, the design basis must be specified. The design basis is a document which
fixes the constraints of the project, e.g. formal things like capacity, operating hours
per year, the apportionment of the plant into units, a general description of the plant,
consumption figures, and the targeted quality of the product. Discretionary decisions
during the project should be avoided as much as possible. The ranges of meteorolog-
ical data are fixed, i.e. for the barometric pressure, the temperature, the air humidity,
and other data such as possible rainfall and their frequency, wind data, data on solar
radiation, sea water temperature in coastal regions, tidal and geotechnical data such
as the ground carrying capacity or the frequency and strength of earthquakes. The
minimum and maximum conditions of the utilities (steam, cooling water, deminer-
alized water, process water, brine, instrument air, nitrogen for inertization, natural
gas, electric current, etc.) are set so that the engineer can choose the design cases
with the most unfavorable conditions. The compositions of the raw materials are de-
fined as well as the ones of the waste streams. The constraints for construction and
design are set, such as the allowed tube lengths and shell diameters of heat exchang-
ers, the fouling factors, and the overdesign to be chosen. The engineering standards
and guidelines to be applied are listed for the various activities. Often, the client com-
pany has its own standards, which are explained in the so-called typicals, where the
arrangement of standard equipment is illustrated (Figure 1.2). The philosophy of back-
ups should be defined, e.g. for pumps. If a pump fails, it is possible that the whole
plant must be shut down. A spare pump which is already installed can solve this
problem. In cases which are less urgent it might be sufficient to have a spare pump
in the storehouse. Also, a preventive maintenance strategy is often applied, where
devices are maintained or even exchanged after a certain time when experience in-
dicates that a failure becomes probable. Any deviation from the design basis must be
reported to the client. The design basis is continuously updated until the engineering
is finished.
I
P
FY
IA
300605 B
FV
300605 B
3" 3"
VBC125A 3"/2" FC 3"/2" VBC125A

VBC125A
VBC125A
1"
1"
3"
Fig. 1.2: Example of a typical
VRC125AT of a control valve arrangement.
In the basic engineering, the process equipment is designed both with respect to its
function in the process and to the mechanical strength requirements. For each equip-
ment, the first issue of the data sheet is released, containing the data derived from
the process, i.e. the process parameters, the dimensions, the nozzles, the materials of
construction, and a specification of the insulation [2]. From that point on, the piece of
equipment is more or less decoupled from the process itself; the responsibility is taken
by the specialists for construction or machinery, who work out the specifications from
the process, especially the mechanical strength.
Besides the equipment, specifications are also prepared for the measurement and
control devices, the piping, and the safety valves. Detailed lists for equipment, utility
consumers, electrical consumers, emissions, control equipment, and instrumentation
are compiled. In collaboration between process and instrumentation engineers, the
interlocks are defined. An interlock is an automatic action of the process control sys-
tem due to safety considerations or equipment protection. Examples are the switch-off
of a pump when the liquid level in the vessel downstream falls below the limit (LSLL,
level switch low low) to avoid cavitation or the stop of feed flow and steam to a column
after a pressure high-high switch (PSHH) had actuated.
The most important work of the process engineer in the basic engineering is the
setup of the piping and instrumentation diagrams (PID), which are the most elaborate
documents of the process. While the PFD contains only process relevant lines and
equipment, the PID shows other equipment, e.g. auxiliary lines for startup, safety de-
vices, and valves, as well. The instrumentation engineers play a decisive role to fix the
concept for measurement and control, which comprises a large part of the PID work.
The PID is continuously updated in the detail engineering and should finally con-
tain the following information [2]:
– all equipment and machinery, including design data and installed spare parts
(e.g. substitute pumps);
– all drives;
– all piping and fittings, nozzles;
– design information for the piping;
– all instruments, control devices and control loops, interlocks with the correspond-
ing signal flow;
– all check valves, safety valves, level gauges, drain lines;
– dimensions and operating data of equipment and machinery, materials of con-
struction, elevations;
– the battery limits, i.e. an illustration of the agreement of the scope of the project;
– the tie-in information, i.e. how the new plant is linked to existing equipment and
utilities.
Figure 1.3 shows the elaborate PID based on the PFD shown in Figure 1.1. Note that
only one column is depicted, because of the increased detailing the second column
is on a separate PID. As a rule of thumb, at most 2–3 pieces of equipment can be
represented on one PID sheet. Thus, it is normal that a major plant documentation
comprises 100–150 PID sheets. One of the most important tasks of the project manage-
ment is the coordination of the various teams and the time schedule for the provision
of the particular pieces of information on the PID. There should always be a “master”
PID, where all the changes are supplemented manually. In fact, there is a guideline
for these changes:
– red color: supplements;
– blue color: omissions;
– green color: comments.
It is a useful agreement to indicate who has made the various changes. At certain
stages, the PIDs are frozen, and a new issue is printed out. This happens several times
during a project, and minor changes are even made during commissioning. The final
PID set is called the “as built” status.
The basic engineering should contain all the information in a way that the de-
tail engineering can be performed without difficulty, and possibly without detailed
knowledge about the process itself. This is an important requirement, as the detail
Fig. 1.3: Example of the detailing in a PID.
engineering is not necessarily carried out by the same company as the basic engineer-
ing. One target of the basic engineering is usually a cost estimation of ± 30 %, based on
budget price offers for the most important equipment and on scaled prices for stan-
dard equipment. While in the basic engineering only the essential information like
dimensions, operating conditions or materials is prepared, in the detail engineering
the complete specification of the whole static equipment and machinery is generated,
which enables manufacturers to submit bids.
In the detail engineering the documents of the basic engineering are more elabo-
rated and finally fixed, so that the following activities can take place [2]:
– preparation of bid invitations for equipment, materials, civil work, and construc-
tion;
– selection of manufacturers and vendors;
– quality assurance operations on vendors and manufacturers (“expediting”);
– planning of the transport of plant equipment;
– execution of civil work and plant construction;
– commissioning.
The information from the equipment manufacturers are considered and implemented
in the documentation. In this phase, layout, piping construction, process, electrical
and other types of engineers work closely together.
In the detail engineering, inconsistencies become less and less permissible. In [1]
an example is described: the process engineer fixes the necessary pipe diameter with
respect to the requirements of the process. Due to reasons of cost and accuracy, the
instrument engineer chooses a smaller diameter for the flow measurement. This gives
a temporary inconsistency, which is tolerable. However, a change is necessary, either
the change of the pipe or the instrument diameter or a reducing adapter. At the end of
the project, this change must have been performed; otherwise, money will be wasted
due to the ordering of wrong materials, not to mention the possible time delay for the
project.
1.2 Realization of a plant
The engineer’s work is a struggle between the pencil and the eraser. If the pencil wins, there is a
chance of getting things finished at some point.
(Dimitar Borisov)
With the ongoing project, the plant layout¹ becomes more and more a high-priority
item. Plant layout is a procedure which involves knowledge of the space requirements
for the facilities and also involves their proper arrangement so that continuous and
steady movement of the products takes place [3].
During recent decades, there has been great progress with regard to tools. Previ-
ously the documents were produced with drawing ink, using special pens with differ-
ent line widths. Erasing a mistake was a risky procedure; instead of an eraser, a razor
blade was used to remove the ink². The drawings were archived on microfilm. Before
the computer evolution began, the only way to convey a 3D impression were isometric
drawings. While this seems possible for piping illustrations, one can hardly imagine
that this was ever appropriate for equipment drawings, especially if the drawing is
subject to changes.
Three-dimensional (3D) objects can not be effectively described in a two-dimen-
sional (2D) space [4]. At least two views of the object are required (see below). From
the 1980s on, layout models made of plastics were constructed. They were very useful,
as they conveyed a good impression about the final appearance of the plant [2], and
they still count as objects worth being shown on guided visitor tours in an engineer-
ing company (Figure 1.4). They were set up according to the 2D drawings to verify the
concept or to solve special problems concerning piping. Plastic models gave an im-
mediate overall impression, but from the engineering point of view, they had severe
disadvantages:
1 The author is grateful to Mrs. Hristina Stegmann, who gave the main part of the input to this section.
2 Of course, razor blades do not distinguish between the ink and the skin of the user. The blood losses
of the author during his studies, usually spread on the drawing, were incredible.
Fig. 1.4: Example of a detailed plastic model.
– it was practically impossible to implement major changes;

– complete representation of the whole plant was a huge effort;
– the accuracy of such a model was limited.
More details can be found in [4].
In the era of scale models, there were many anecdotal stories of fingers being glued together,
inhalation of noxious fumes from plastic solvents, lacerations from cutting tools, and dissolved
fingerprints. Today, the 3D CAD modellers suffer from carpal tunnel syndrome and e-mail over-
load.
(found in [4])
With increasing computer capacity, 3D documents have become standard, and their
precision is amazing. They can hardly be distinguished from a photograph after the
construction is finished. A number of programs are on the market, usually with a
quite considerable license fee. While the elementary functions can be operated quite
rapidly, it takes a few months until an engineer can claim to use such a system prop-
erly. The programs are able to show the plant with any level of detail from any point of
view. The required documents can be created automatically, i.e. plot plans, equipment
arrangement drawings, piping isometrics, and piping layout drawings. Other docu-
ments such as line lists or equipment lists can be consolidated. Creating a 3D model
can be as time-consuming as manual drawings, but the saving of time and work occurs
downstream in the workflow [4]. Figure 1.5 shows an example.
A possible drawback of 3D models is an overconfidence in them, just because of
their precision and the amazing visualization technology. It is still the engineer whose
abilities are crucial for the quality of the model. The program does not prevent him
from making mistakes.
Nowadays, virtual reality (VR) programs offer the next option: a pair of data gog-
gles enables the user to go through the plant and check, for instance, its operability.
Currently, this technology is not standard, but with time, it will become more and more
accessible and less expensive. Connected to a dynamic process simulation, VR is sup-
posed to be a valuable training tool for the staff, which can be exposed to a large
range of training scenarios including hazardous situations. It increases the process
and safety procedure knowledge, improves plant reliability, and lowers the accident
rate [5]. Whether just 3D or VR is applied will probably be a matter of the cost-benefit
ratio. More information can be found in [6].
Fig. 1.5: Example of a 3D representation.

Courtesy of AVEVA GmbH.
The layout work already begins in the proposal phase of a project. A first concept must
be set up in order to give a first idea about the appearance of the plant and to confirm
that the proposed area on the site is sufficient.
During the conceptual phase, precise information about the equipment is usually
not known, unless reliable data from a reference plant is available. Only the size of
key equipment such as reactors, large tanks, and silos can at least be estimated ac-
cording to their capacity. Therefore, the particular pieces of process equipment are
often represented by placeholders with an estimated preliminary size. Their position
inside the battery limits is defined with regard to numerous safety and service require-
ments, considering the natural process flow for an efficient arrangement. When new
information is generated, the layout plan can be continuously updated. With the first
issues of the PIDs, it becomes more and more elaborate, and the adjustments to be
made become less significant – and it makes sense to begin with the 3D piping work.
There are a number of 2D documents which are often relevant. The genuine docu-
ments containing the engineering information are drawings which show the top and
the side views. Isometric drawings are often produced to obtain a better imagination
of the plant. The three axes of space are not perpendicular to each other but angled
towards the viewer with 60° angles between them. The lengths are distorted; they
appear foreshortened by a certain factor. Isometric drawings are more useful in archi-
tecture. In engineering, they are only produced for illustration and usually not used
as documents containing the information. The relevant 2D documents are:
– Overall plot plan:
Shows an overview of a complete plant including the battery limits (Figure 1.6).
– Area plot plan:
Shows the overview of a part of a plant, e.g. a unit.
– Overall plot plan, isometric view:
Shows the view of the complete plant from two opposite directions. It is not really
necessary, but gives a good projection of the plant. Nowadays, there a tools which
can easily provide this.
– Equipment arrangement drawing:
Turns the focus to the particular pieces of equipment. The general rule is that each
piece of equipment must be visible from two sides to have a clear definition; usu-
ally the top view for defining the ground and the side view to define the tangent
line and the center line elevation are required.
– Iso-view of the equipment arrangement drawing:
Again, the iso-view does not provide information for the engineer, but it is easy to
produce and gives a good ide of it.
– Pipe iso (piping isometric drawing):
Isometric representation of a pipe with the coordinates for beginning and end and
the bends (Figure 1.9). It is automatically generated for all lines; it is used for or-
dering the material, the manufacturing, and the fitting into the plant. It is also
useful for process technology if a pipe must be carefully calculated, e.g. for an
exact pressure-drop calculation. Inlet and outlet lines of safety valves are well-
known examples.
– Piping layout drawings:
Piping layout drawings are equipment arrangement drawings which show not
only equipment and supporting structures but also the pipe lines available in
the represented area. As this kind of drawing is often overloaded with informa-
tion, it is being increasingly replaced by the direct use of the 3D model. Nowadays,
viewer programs are available where engineers can use the 3D model without spe-
cial knowledge about design.
Fig. 1.6: Example for an overall plot plan.
In basic engineering, the main dimensions of the particular pieces of equipment are
outlined. The goal is that they are as exact as possible, so that other activities like
piping and static calculations for the steel structure can get started in a reasonable
way. Other details of equipment such as nozzles are still omitted.
Also, the safety distances between the units or the pieces of equipment are worked
out, considering that service and maintenance concepts are set up as well as construc-
tion procedures (e.g. necessity of cranes). Taking into account that later on people will
spend time in the plant, the concepts for fire and explosion protection, the location of
safety showers and escape routes and the removal of flammable liquids in case of an
accident have to be clarified.
For layout, there are a number of basic principles that should be followed, among
them:
– Follow the process flow to keep the pipe lengths short. For example, if the con-
denser of a distillation column is located on the 2nd floor, put the reflux drum
on the 1st floor and the corresponding pump on the floor. Utility units should be
placed near the corresponding tie-in.
– Consider the minimum safety distances. E.g., compressors or units containing a

compressor such as cooling units need a distance of 9 m to other pieces of equip-
ment, depending on the engineering standard used.
– Estimate the space requirement of the piping around an apparatus and keep it
free! At the beginning, there is no information about this, and it is often underes-
timated. The information is usually generated when it is too late for relocating the
apparatus. It is not a solution to keep some space left in a conservative way; the
waste of space and the increased length of the pipes are expensive. Nevertheless,
some space for unforeseens should always be considered. Insufficient space for
pipe lines and instruments mostly leads to poor operability and personnel safety
issues. The estimation of the space requirement can be done best according to a
reference plant design with a similar capacity. If the reference plant has a signifi-
cantly different capacity, one should be aware that there is no linear relationship
between size and capacity. The process specialists should be able to make a good
first guess. The package units are usually the most challenging tasks. The layout
information is provided by the vendor, who is chosen at a late stage of the project.
Different vendors might use different technologies, and the space demand can
vary considerably.
– First create a concept for all escape and service routes and stick to it! For example,
provide continuous corridors on each level, and in the same way on each level so
that one can orientate oneself even if there is, for example, heavy smoke.
– Peculiarities concerning maintenance and service must be identified and taken
into account. A BEU heat exchanger (Chapter 4, Figure 4.7) will probably be regu-
larly dismantled for cleaning. There must be enough space for this operation, and
the dismantled tube-bundle should show to the road and not to the pipe rack.
Especially, reboilers which can be dismantled are a delicate issue. As well, dip-
pipes (Chapter 9) are useful for directing liquid flow, but the layout engineers have
to consider reserved space for dismantling above the apparatus.
– Any machines with movable parts like pumps and compressors need regular
maintenance, which is convenient to be done on the ground floor. Furthermore,
pumps and compressors cause vibrations, which can be controlled in the best
way on the floor. For pumps, the location on the ground floor (Figure 1.7) makes
sense anyway, as the maximum NPSH value is generated (Chapter 8.1).
– Finally, a sense for symmetry is useful. The vessels should be arranged in a
straight line, as well as the outlet nozzles of the pumps (Figure 1.7). The distances
are even numbers, e.g. 2 m or 1.5 m.
– Interconnecting pipes are collected on a pipe rack.
Fig. 1.7: Typical pump arrangement.
A comparably new trend is modularization, which means that units can be manufac-
tured and assembled in a frame (Figure 1.8). This is useful if only a short time slot for
construction and assembly is available. If it takes a long time to get the permission for
the construction of the plant, the project can be shortened if at least the modules are
ready. Besides the time savings, some other aspects of modularisation are:
– Smaller space demand, but the module frames might be so narrow that piping,
maintenance and operation might become difficult. In plant engineering, “nar-
row” is equivalent to “dangerous”, so one has to take care that the safety concept
is still fulfilled.
– Lower costs, as assembly does not need to be performed at the site with a large
personnel staff
– Higher quality, as qualified people do the assembly in their own workshop
– Good option for standardization and optimization, but there are different types of
transport limitations (e.g. container sizes), and the module must fit in any case.
Fig. 1.8: Fully modularized process blocks.
Fire protection is one of the most important concepts of a chemical plant, as its target
is to protect human life and resources, and namely, in this order. It has three compo-
nents:
– Fire prevention:
Fire prevention comprises the education of the staff, the working procedures for
fire and explosion prevention and emergency case procedures. Layout takes care
that the safety distances are kept and that the access to fire engines is ensured [2].
– Passive fire protection:
The main target of passive fire protection is to avoid the spreading of the fire across
the plant. This can be achieved by the use of fire resistant walls and floors and the
fire protection of equipment supports and steel structures.
– Active fire protection:
Active fire protection is the system for the delivery and distribution of fire-fighting
water or, alternatively, the foam generation system.
The piping engineering is an iterative procedure, as it depends on more or less all other
disciplines. While the first approaches are based on assumptions, the details become
more and more available, and at the end of the detail engineering the line routing is
fixed, i.e. all the bends and lengths are specified in isometric drawings (Figure 1.9) so
that each pipe can be manufactured. The process engineer must then check whether
the exact line routing corresponds to the assumptions made in the basic engineer-
ing, e.g. the outlet lines of safety valves (Chapter 14.2) must not generate significantly
more pressure drop then calculated before, and the pump specification (Chapter 8.1)
has to be rechecked. In piping engineering, it is an important procedure to classify all
components (pipes, fittings, flanges, valves, sealings, insulation, etc.) with respect to
the media and the maximum operating conditions. A piping class represents such a
set of conditions. In the PID, the corresponding information is included in the iden-
tification code of each line. A piping list is composed which indicates the material
requirements for the piping, which is closely related to the costs. Extensive mechan-
ical strength calculations take place. The necessary wall thickness can be evaluated
(Chapters 11 and 12), and elasticity calculations are performed to make sure that the
stress resulting from temperature changes during operation is tolerable.
Further documents which are compiled for the construction are:

– underground coordination plan (foundations, underground lines, pits, chan-
nels);
– civil info plan (steel and concrete structures, paving);
– escape and rescue plan (escape routes, eye-showers, alarm buttons, fire and gas
detection);
– room data list (surfaces, air condition);
– load data list
The load data list is one of the most important documents, compiling the weights
of the particular pieces of equipment, which are decisive for the design of the sup-
porting structures; failures can lead to serious accidents or at least to significant
delays.
Fig. 1.9: Example of an isometric drawing of a line.
For the commissioning, an operating manual is prepared where all activities in

startup, operation, and shutdown are described in detail, broken down to the po-
sitions of the valves at the various activities. Also, it is checked whether the recom-
mendations from the HAZOP study have been considered. An important tool for this
purpose is the so-called cause & effect matrix, which connects failures to the actions
of the interlocks and gives an overview on the interlock structure.
The electrical demand of the process is specified, concerning voltage levels, loca-
tions of the distribution stations, and cable dimensions with respect to the electrical
consumer list.
Finally, the process control system is set up. It illustrates the operation of the
plant. In a central station, all information is gathered and made available for the plant
staff. The data are usually stored so that they can be used for later analyses.
The consistency of the documents becomes more and more important for the qual-
ity of the engineering, whereas the typical process engineering issues like design,
process performance, and economy are regarded as finished. In fact, changes at a late
stage of a project are always associated with considerable cost and should be avoided.
It is amazing that even in Germany there is no chair for

“administrative process engineering”.
(heard on the evening meeting of the UNIFAC Consortium)
All engineering activities are strongly affected by organization structures in the com-
pany and in the project. Currently, the most common organization form is matrix
project management (Figure 1.10). Each project has a number of engineers coming
from the various disciplines (here: process, engineering, procurement, construction;
additional disciplines may be piping, instrumentation, electrical, layout etc.). Nor-
mally, they report to the head of their division, and for the time of the project they are
also members of the particular project teams, where they report to the project man-
ager. The project manager decides which activities have to be executed and when the
due date is, while the heads of the division delegate appropriate people to the project
and supervise their technical decisions. The project manager is directly responsible
for the project and reports to the board of his company; for the customer he is the main
partner in discussions. He ensures the compliance with budget costs, time schedules,
and quality level [2].
Executive Board
Project Process Engineering Procurement Construction

management division division division division
division
Project A
Project manager
Project responsibility
Project B
Project manager
Members of
project teams
Project C
Project manager
Project D
Project manager
Functional responsibility
Fig. 1.10: Matrix organization [2].

© Wiley-VCH Verlag GmbH & Co. KGaA. Reproduced with permission.
Even after the successful startup of the plant, the engineering work is not yet finished
[1], as further projects (e.g. maintenance, revamping, debottlenecking) will take place
during the whole life cycle of the plant. For such activities, an accurate documentation
of the plant as it was created during basic and detail engineering is valuable. Not only
the current state (“as-built”) is needed, but also the history, which reveals the reasons
and presumptions for the design of the plant, e.g. the various operation cases relevant
for the dimensioning of a heat exchanger. Documentation is, apart from the content
itself, an art of its own, usually related with various software compatibility problems,
as software changes with time. Nowadays, it is clear that a process documentation
which has to be updated continuously cannot be handed over in a paper version but as
files with common software. The data exchange is often difficult if different softwares
are applied, even for different revisions and configurations of the same software.
1.3 Cost estimation
Cost estimation is certainly one of the most nontransparent things which engineers
encounter at the beginning, as it is not really studied at the university. Nevertheless,
it is one of the most important parts of any project.
Costs are divided into investment (CAPEX) and operation costs (OPEX); the latter
can be further divided into fixed costs (e.g. salaries of the staff) and variable costs,
which are proportional to the production output (e.g. raw materials, utilities). On the
other hand, there are revenues due to the sale of the product. Having a heat and mass
balance and the specific prices available, one can perform an estimate of operation
costs and revenues. This first approach is certainly inaccurate, but on its basis one
can decide whether it is worth continuing with the more labor-consuming investment
cost estimation. The revenues must be significantly higher than the operation costs;
earning money is the purpose of building a plant, and the payback of the investment
costs including interests must take place before anything in fact is earned. During the
course of the project, the estimate of operation costs and revenues is continuously
updated.
The estimation of the investment costs requires a lot of experience and knowl-
edge about similar or even previous projects of the same kind. In the early stage of the
project, until cost estimates for each item are available, the estimation of the invest-
ment costs is based on the costs of the major equipment. Additionally, costs for bulk
material (piping, instrumentation etc.) and construction are considered by percent-
ages of the major equipment or the total engineering costs; these percentages depend
on the various constraints of the project (buildings to be erected, use of expensive ma-
terials). E.g., the piping costs are reported to be 20 … 40 % of the total engineering [2].
Moreover, the requirements of the engineering company have to be considered; the
project costs are therefore supplemented by the costs for engineering and procure-
ment, license fees, and a profit margin.
The estimation of the equipment costs should be based on costs known from previ-
ous projects. Especially the prices for the materials must be considered appropriately.
The dimensions can be considered by the six-tenth-law:
C1 M
P(C1 ) = P(C0 ) ⋅ ( ) , (1.1)
C0
where C is the capacity of the equipment and P is its price. For the exponent M, M =
0.6–0.7 ≈ 10 6
is often a good approach. This refers to the fact that the volume V of
an apparatus is proportional to the capacity, and the surface, which determines the
necessary amount of material, is proportional to V 2/3 . More elaborated values for the
exponent M and orders of magnitudes for prices are given in [7].
Equation (1.1) also illustrates the economy-of-scale-effect. While the revenues of
the plant operation increase linearly with the capacity of the plant, the investment
costs both for equipment and the whole plant increase by far more slowly. The larger
the capacity is, the relatively smaller the investment costs are.
Once the costs for the equipment are calculated, the costs for the whole project
can be estimated by means of factors referring to the costs of the equipment. Great
experience is necessary to assess their values. From the literature [7], the following
factors in Table 1.1 can be taken as a rule of thumb for processes with fluids. The item
“contingency” accounts for uncertainties in the process or in project execution. Each
of the values listed in Table 1.1 has a certain range, depending on the type of plant.
The values can reliably be determined if a plant of the same type or at least a similar
one has been analyzed before. The factors become smaller if expensive materials are
Table 1.1: Factors for evaluation of the capital costs.
Item Factor
Main equipment 1.0
Equipment stationing 0.4
Piping 0.7
Instrumentation 0.2
Electrical 0.1
Utilities 0.5
Offsites 0.2
Buildings 0.2
Site preparation 0.1
Total equipment costs 3.4
Engineering 1.0
Contingency 0.4
Total fixed capital costs 4.8
Working capital (inventories) 0.7
Total capital costs 5.5
used; in this case, costs for activities like engineering stay the same, but the costs for
the main equipment rises so that the percentage becomes lower.
Capital costs for a chemical plant are often in the range 50 million to 1 billion €,
and a possible investor must be convinced that he will get his money back within a
manageable time period. For this purpose, the revenues from the sale of the products
must be higher than the operating costs. The best overview is achieved by referring to
1 t of the main product, as Table 1.2 with its fictitious example shows. The calculation
of the production costs is not only used for the decision whether a project is feasible
or not but also later for the cost control of the process.
Table 1.2: Example for the structure of the operation costs.
Capital costs 200 000 000 $

Capacity 200 000 t/a
25 t/h
Operating time p.a. 8 000 h/a
Product sales 1 500 $/t
Item Consumption Unit Price Unit Costs ($/t)

Raw materials
Raw material A 40.0 t/h 400 $/t 640.00
Raw material B 25.0 t/h 100 $/t 100.00
Ammonia 5.0 t/h 600 $/t 120.00
Utilities
Cooling water 10 000.0 m3 /h 0.03 $/m3 12.00
Steam 200.0 t/h 20 $/t 160.00
Electrical energy 3 000.0 kW 0.1 $/kWh 12.00
Effluents
Waste water 3.0 m3 /h 10 $/m3 1.20
Combustible waste 0.5 m3 /h 40 $/m3 0.80
Coproduct for sale 5 t/h −150 $/t −30.00
Variable costs 1016.00
Personnel 40 70 000 $/a 14.00
Overhead 15 % of personnel 2.10
Maintenance, insurance 2 % of capital costs 20.00
Capital costs 10 years depreciation 100.00
Fixed costs 136.10
Production costs 1 152.10
Product sales 1 500.00
Margin 347.90
One should be aware that Table 1.2 is only an example. Depending on the product,
the cost structures can of course differ, and finally, it is much more detailed. It can
include site-related costs like license fees, fire brigade, staff canteen, site bus, site
streets, staff association, and, of course, taxes. Nevertheless, some typical numbers
are worth discussing. It is typical that raw materials make up a large part of the produc-
tion costs. Without a reasonable value creation from the raw materials to the product,
there is hardly a chance of making the process feasible. This value creation should
be checked first; especially, the amount of product per t of raw material should be
checked, which is easily possible even without a detailed mass balance³. Often, fluc-
tuations of the prices of product and raw materials are decisive even for the feasibility
of the process. In comparison, utilities play a minor role; however, the optimization of
energy consumption is often the only thing which can be influenced by good process
engineering. The consumption of 1 MW el. per 100 000 t/a product capacity is a typical
value, which can be exceeded if the process contains one or more compressors. Energy
costs differ greatly according to the site location; on the Arabian peninsula or in the
United States steam costs are easily below 10 $/t, whereas in Europe or Asia 30 $/t are
realistic.
The numbers for product revenue should not be based on single values on a daily
basis. Instead, the market and its flows must be understood. It must be clear on what
basis the number is founded and how the forecast had been performed. Also, one
of the main tasks of a process engineer is to explain what the numbers in Table 1.2
are based on. Coproducts which can be sold are often generated; in contrast to this,
by-products are simply losses and have to be disposed of. An additional distribution
channel must be established for coproducts and furthermore, from the process point
of view an additional quality surveillance is necessary. The cost structure should not
indicate that the sale of the coproducts would tip the scales to make the project fea-
sible. From the fixed costs items, usually only the depreciation of the plant plays a
decisive role. Rules of thumb indicate the following ranges for the operation costs:
– capital costs: 15–30 %;
– energy costs: 10–40 %;
– raw materials: 30–90 %;
– Salaries: 5–25 %.
The ratio between fixed and variable costs determines whether there is an economy of
scale. The variable costs are proportional to the production rate; they do not vary in a
larger plant. The fixed costs stay constant (e.g. personnel costs) or increase less than
proportional to the capacity (investment costs; see Equation (1.1)). The more the fixed
costs determine the cost structure, the larger is the effect of the economy of scale.
A good compilation of cost estimation for chemical plants can be found in [8].
3 An example is given on p. 88.

There are a number of characteristic values to assess the feasibility of a project.

The most widely used one is the so-called net present value [9, 10]. For its evaluation,
the period where the project produces positive or negative cash flow is divided into
time segments, in most cases years. A discount rate is considered, which takes into
account that the investment takes place at the beginning of the project, whereas the
revenues and further expenses occur later. With this discount rate, all cash flow refers
to the time when the project starts. Its value is chosen in a way that it represents the
assumed risk of the project. For the cash flow evaluation, the following constituent
parts are considered:
Cash flow = (revenues − fixed costs − variable costs) ⋅ (1 − tax rate)

(1.2)
+ Depreciation ⋅ tax rate − investment costs ,
where the revenues are calculated as price multiplied by quantity of the product. Fixed
and variable costs depend directly on the economic and operating assumptions of the
process. Integrated over the lifetime of the project, the Net Present Value (NPV) gives
T
(Cash Flow)i
NPV = ∑ , (1.3)
i=0
(1 + r)i
with
T lifetime of the project in years;
i number of the year;
r discount rate.
20
18
% Rate of return
16
14
Investment costs
Degree of utilization
12 Raw material costs
Product sales
Utility costs
10
–20 –10 0 10 20
% Variation
Fig. 1.11: Example of a sensitivity spider chart.

The NPV maximizes the value for the company performing the project. However, it has
also some disadvantageous properties. It depends strongly on the scale of the project,
and therefore it is difficult to compare different scenarios. It is more sensitive to the
annual revenues than to the investment costs. Constraints like availability of the initial
investment and the market situation must be set [10]. Directly related to the NPV is the
internal rate of return (IRR), which is defined as the interest rate which gives a zero
NPV in Equation (1.3). The IRR is independent of the scale of the project. From the
mathematical point of view it is not easy to handle, as multiple solutions often exist.
It should not be used as the economically decisive criterion without a critical analysis
[10]. There are a number of other methods available which are well explained in [8].
A very simple criterion which is often used for the assessment of a smaller project
(e.g. a revamp for heat integration) is the static payback period. It is calculated how
much time it will take until the investment costs for a project are covered by the rev-
enues. The static payback period should be less than 3 years to make a project feasible.
The so-called sensitivity spider chart is a useful tool to evaluate how the economic
situation varies if one of the assumptions varies. This way, one gets a feeling about the
most decisive assumptions which have to be continuously watched and updated. An
example is shown in Figure 1.11.
2 Thermodynamic models in process simulation
Without reliable physical properties, a process simulator is just an expensive
random number generator.
(A. Harvey and A. Laesecke, 2002)
The understanding of physical properties of fluid substances and their phase equilib-
ria is one of the main keys for a successful process engineering. Although there are
more exciting problems to solve in process engineering, almost every larger project
starts with the clarification of the physical property situation. Without having a sound
understanding of this, a reasonable process simulation cannot be carried out. Mainly,
the following properties are important for process engineering:
– thermodynamic properties:
– density;
– enthalpy;
– phase equilibrium;
– transport properties:
– viscosity;
– thermal conductivity;
– surface tension;
– diffusion coefficients.
For the mass and energy balance, only the thermodynamic properties are required,
whereas the transport properties play a major role for the equipment design, e.g. for
columns, heat exchangers, or pumps. For the thermodynamic properties, the calcula-
tions of the phase equilibria play the key role for the whole process simulation, as they
determine the separation steps in the process, which often cause 60–80 % of the total
costs of the process. In process simulators, a model has to be chosen which decides
in which way the phase equilibria are determined; also, it has an influence on the
evaluations of densities and enthalpies. It is not necessary but advantageous if one
model could be used for the whole process; each model change holds the risk that in-
consistencies are introduced (Chapter 2.11). Two kinds of models can be distinguished:
the equation-of-state models (φ-φ-approach) and the activity coefficient models (γ-φ-
approach). The difference and the advantages and disadvantages between these two
approaches should be known for a reasonable choice of the model. In the following
section, the essentials of the most important models are introduced and discussed
without any thermodynamic framework. For more details of these models, see [11].
The importance of the particular physical properties have been rated in Table 2.1
[12], where the relationships between accuracy of the particular properties and the
influence on the investment costs are listed. Certainly, this table should not be taken
as the absolute scientific truth but as an examplary case study. The outstanding item
26 | 2 Thermodynamic models in process simulation
is the large influence of the activity coefficient¹. The author would agree upon its im-
portance; however, in fact the costs are driven by the separation factors
p si γ i
α ij = . (2.1)
p sj γ j
The importance of its accuracy strongly depends on the case. When the separation fac-
tors are far away from unity, the influence on the accuracies of the activity coefficent
and vapor pressures is limited. When the separation factors are close to unity, their in-
fluence is incredibly high, and, moreover, they can even decide whether a separation
is possible at all.
The heat of vaporization is clearly proportional to the reboiler duty in a distillation
and therefore to the size of the reboiler; thus, the proportionality is quite in line with
the experience of the author. In comparison to the heat of vaporization, the influence
of specific heat capacity is smaller. Nevertheless, single pieces of equipment can be
strongly influenced (e.g. liquid-liquid heat exchangers), and errors occur pretty often
(Chapter 2.8). The transport properties, thermal conductivity and viscosity, have an
influence on the heat transfer coefficient in heat exchangers. The author would guess
that the influence of the viscosity is greater for large viscosities. Moreover, errors in
the viscosity occur quite frequently, especially for mixtures, whereas thermal conduc-
tivities do not vary too much for the particular liquids, with the exception of water
and glycols. Vapor viscosities and thermal conductivities are usually not measured
but estimated anyway.
The accuracy of estimations of physical properties is often of great interest. This
question is not easy to answer, as most of the authors are suspected to claim a higher
accuracy than they should. However, it is hard to imagine how objective criteria can
be set up. Clearly, the average deviation of the fit to the data available is an indica-
tor, but certainly the fit to unknown data will be worse. Other authors try to leave out
Table 2.1: Example of a relationship between physical property accuracy and investment costs [12].
Physical property % error % error capital cost

Thermal conductivity 20 % 13 %
Specific heat capacity 20 % 6%
Heat of vaporization 15 % 15 %
Activity coefficient 10 % 100 %
Diffusion coefficient 20 % 4%
Viscosity 50 % 10 %
Density 20 % 16 %
1 The activity coefficient γ will be explained in Chapter 2.1. In the context of this paragraph, it is de-
fined as a factor describing the deviations from Raoult’s law. It can be interpreted as a correction factor
for the concentration.
Table 2.2: Accuracy of physical property predictions [15].
Physical property Error expected Error desired

Heat of formation 2.5–4 kJ/mol 4 kJ/mol
Liquid heat capacity > 10 % 10 %
Liquid density > 2% 2%
Vapor pressure > 10 % 10 %
Normal boiling point 6K 3K
Transport properties > 10 % 10–20 %
Heat of vaporization 15 % 15 %
Limiting activity coefficient > 10 % 10 %
certain data sets in the fitting process [13], however, one does not know according to
which criteria they are chosen. Another method is to predict new data sets before they
are integrated into the database [14]. This method can not deliver large amounts of
examples, and, moreover, it is not reproducible, as after testing they will certainly be
added to the database.
A thorough examination has been done in Table 2.2. The conclusions are as fol-
lows [15]:
– the accuracy of the methods are not at industrial target level;
– experimental data for thermal and transport properties are limited;
– group contribution methods seem to have reached their potential, there is hardly
room for improvement. To meet the industrial demand, new approaches are nec-
essary.
2.1 Phase equilibria

Life is too short to worry about phase equilibria.
(Based loosely on a comment of Georgios Kontogeorgis on electrolytes.)
The knowledge and understanding of the various phase and chemical equilibria is
the key for a successful process simulation. Two-phase regions have a great impor-
tance in technical applications. Even for a one-component system phenomena occur
which need to be discussed thoroughly. Exemplarily, Figure 2.1 illustrates the isobaric
vapor-liquid equilibrium of water when it is heated from t1 = 50 °C to t2 = 150 °C at
atmospheric pressure.
In the two-phase region, vapor and liquid coexist at the same temperature and
pressure [11]. In this case, both liquid and vapor are called saturated. If the saturated
liquid is further heated, the temperature does not change. Instead of a temperature
rise, the liquid is vaporized. After the last drop of liquid vanishes, the temperature
rises again. Figure 2.1 clearly indicates that much more heat is consumed for the evap-
oration of water (enthalpy of vaporization, ∆hv ) than for the 100 K temperature eleva-
150
125
100
t/°C
75
50
25
0
0 500 1000 1500 2000 2500 3000
q/J/g
p
p
p
p p
V V V
L L L
All liquid First vapor Phase Last liquid All vapor

bubble equilibrium drop
Fig. 2.1: Temperature change of water at p = 1.013 bar with respect of the heat added [11].
tion. The state of a pure substance in the phase equilibrium region is not characterized
by temperature and pressure as in the one-phase region. Temperature and pressure
are related; the relationship
ps = f(T) (2.2)
is the vapor pressure curve. For the complete determination of the two-phase system
the vapor quality x
n󸀠󸀠
x= 󸀠 (2.3)
n + n󸀠󸀠
is necessary, where the superscripts 󸀠 and 󸀠󸀠 denote the saturated states of liquid and
vapor, respectively. x = 0 means a saturated liquid, x = 1 means a saturated vapor. In
the two-phase region, the specific volume v, the specific enthalpy h and the specific
entropy s can be written as
v = xv󸀠󸀠 + (1 − x)v󸀠 (2.4)

󸀠󸀠 󸀠
h = xh + (1 − x)h (2.5)
󸀠󸀠 󸀠
s = xs + (1 − x)s . (2.6)
Vapor pressure and enthalpy of vaporization of a pure substance are related by the
Clausius–Clapeyron equation
dps 󸀠󸀠
∆hv = T (v − v󸀠 ) . (2.7)
dT
For a multicomponent system, we must also consider that both phases have different
composition, and the distribution of each component is a central issue to be solved
by the thermodynamic model. As will be shown later, the behavior of a multicompo-
nent system is mainly described by binary subsystems. The best way to illustrate the
vapor-liquid equilibrium of a binary system is the pxy diagram at constant tempera-
ture. Figure 2.2 gives an example for the system ethanol/water.
80
Bubble point curve
70
Liquid
Tie-line
60
p/kPa
2-phase Vapor
region
50
40 Dew point curve

T = 343.15 K
30 Fig. 2.2: Example for a pxy diagram [11].
0 0.5 1 © Wiley-VCH Verlag GmbH & Co. KGaA.
x1, y1/mol/mol Reproduced with permission.
In the upper part of this diagram, there is the liquid region above the bubble point
curve. Below the dew point curve, there is the vapor region. Between bubble and dew
point curve, there is the two-phase region. For a specific pressure, a horizontal tie-line
connects the corresponding liquid and vapor points in phase equilibrium, referring
to their concentration on the abscissa. These phase equilibrium diagrams can be ob-
tained by correlating phase equilibrium data. There are several experimental options;
the most popular ones are to measure both the vapor and the liquid concentration
and the temperature at constant pressure, or to measure the bubble point at constant
temperature for certain liquid concentrations. The latter option should be preferred
in most cases; isothermal data are much more useful for the adjustment of model pa-
rameters [16]. Bubble and dew point line meet at the ordinates at x = 0 and x = 1,
indicating the corresponding pure component vapor pressures. If the temperature is
greater than the critical temperature of one of the components, the vapor pressure of
60
177.59 K 166.48 K
50
144.26 K
40 133.15 K
p/bar
30
20 110.93 K
10
Fig. 2.3: pxy diagram with one component
0 becoming supercritical [11].
0 0.5 1 © Wiley-VCH Verlag GmbH & Co. KGaA.
x1, y1/mol/mol Reproduced with permission.
this component does not exist, although it can take part in a phase equilibrium in the
dilute state. In Figure 2.3, the typical change in the shape of the pxy diagram is shown
for the system nitrogen/methane when nitrogen becomes supercritical.
Some other diagrams are useful for illustrating a binary vapor-liquid equilibrium.
The simple yx diagram shows the relationship between vapor and liquid concentra-
tion without indicating the pressure or temperature. For a better overview, the bisect-
ing line is usually depicted (Figure 2.4, first column). Also a Txy diagram is possible
for isobaric phase equilibria, as can be seen in the right column of Figure 2.4. In this
diagram, the upper line is the dew point curve, and the lower line is the bubble point
curve. Figure 2.4 also shows the various kinds of binary vapor-liquid equilibria.
The upper row shows an ideal mixture obeying Raoult’s law (Equation (2.40)).
There are no interaction forces between the molecules. The phase equilibrium is just
determined by the vapor pressures of both components. Typical examples for ideal
systems are benzene/toluene or n-hexane/n-heptane. In the pxy diagram, the boiling
point curve is a straight line. In the Txy diagram, usually no straight lines occur. The
activity coefficients (Chapter 2.3.1) are all equal to 1, i.e. their logarithm is 0.
Row 2 shows a system having small nonidealities. Molecules of the same kind
“prefer” to be together instead of mixing with molecules of a different kind. As a result,
greater pressure is built up than for the ideal mixture. The activity coefficients are
greater than 1. A typical example for such a system is methanol/water.
When the activity coefficients become larger, the system can exhibit an azeotrope
with a pressure maximum in the pxy diagram and a temperature minimum in the
Txy diagram (row 3). At the azeotropic point, the liquid and vapor concentrations
are identical. Therefore, an azeotrope cannot be separated by simple distillation. The
knowledge of azeotropes is essential for any process development. A typical exam-
ple for a homogeneous azeotrope is water/1-propanol. The occurrence of azeotropes
is also related to the vapor pressures; the closer together the vapor pressures of the
components are, the more probable is the occurrence of an azeotrope [11] (p. 46).
1 4 140
0.8 380
3 120
0.6
p/kPa
100 370
ln γi
T/K
2
y1
0.4
80 360
0.2 1
60
0 0 350
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
(1) x1 x1 x1, y1 x1, y1
1 4 350
80
0.8 340
3
0.6 60
p/kPa
ln γi
T/K
2 330
y1
0.4
40
0.2 1 320
20
0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
(2) x1 x1 x1, y1 x1, y1
1 4 130 355
0.8 120
3
350
0.6
p/kPa
110
ln γi
T/K
2
y1
0.4
100 345
0.2 1
90
0 0 340
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
(3) x1 x1 x1, y1 x1, y1
1 4 30 345
0.8 25 340
3
335
0.6 20
p/kPa
ln γi
T/K
2 320
y1
0.4 15
315
0.2 1 10 310
0 0 5 305
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
(4) x1 x1 x1, y1 x1, y1
1 0 120 340
0.8 110
335
0.6
p/kPa
ln γi
T/K
–1 00
y1
0.4
330
0.2 90
0 –2 80 325
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
(5) x1 x1 x1, y1 x1, y1
Fig. 2.4: Kinds of binary vapor-liquid equilibria [11].

1 ideal mixture
2 small nonidealities
3 larger activity coeffcients, homogeneous azeotrope
4 heteroazeotrope
5 vapor pressures close to each other/strong negative deviations from Raoult’s law.

If the activity coefficients increase even more, the system exhibits a miscibility gap,
and the liquid splits into two phases. A heteroazeotrope occurs (row 4). A typical
example is water/n-butanol. Note that a miscibility gap and the occurrence of a het-
eroazeotrope is not necessarily coupled. If the vapor pressures largely differ, the
azeotrope does not occur, whereas the miscibility gap itself is not related to the vapor
pressures.
In addition, negative deviations from Raoult’s law can occur, where the activity
coefficients are lower than 1. This means that the molecules “like” each other and
prefer to be surrounded by molecules of a different kind. An example for such a system
is dichloromethane/2-butanone [11].
If vapor pressures are close to each other and the negative deviations are strong,
even an azeotrope with a pressure minimum in the pxy diagram and a temperature
maximum in the Txy diagram can occur (row 5). An example is acetone/chloroform.
It cannot happen that systems with negative deviations from Raoult’s law show a mis-
cibility gap.
In addition to these five types of phase equilibria there are others. For example,
methyl acetate/water shows a miscibility gap with a homoazeotrope [17], and ben-
zene/hexafluorobenzene has a double azeotrope, one with a pressure minimum and
one with a pressure maximum [18].
As mentioned, there are two different approaches for the description of phase
equilibria. In the following sections, they are explained on vapor-liquid equilibria.
2.2 φ-φ-approach
You cannot just learn thermodynamics. You must love it.

(Recommendation to desperate students)
Generally, an equation of state is a relationship between the pressure p, the abso-

lute temperature T, and the specific volume v. In a first step only pure substances are
regarded. It makes sense to focus on equations of state which are pressure-explicit.
Hence, the general form of an equation of state can be written as
p = f(T, v) . (2.8)
The simplest equation of state is the ideal gas equation

RT
p= . (2.9)
v
Equation (2.9) is exact for a gas where the molecules have no volume and do not
exert interaction forces on each other. It is a good approximation for gases at low
pressures. At increasing pressures, Equation (2.9) becomes increasingly inaccurate.
A well-known modification of Equation (2.9) is the virial equation:
RT B(T) C(T) D(T)

p= (1 + + 2 + 3 + ⋅⋅⋅) . (2.10)
v v v v
Equation (2.10) is the so-called Leiden form; correspondingly, there is also a volume-
explicit Berlin form expressed in form of a polynomial of the pressure p. The virial
coefficients B, C, D, … account for the deviations from the ideal gas equation. Al-
though it has a theoretical background, the virial equation is not suitable for practical
applications. Usually, it has to be truncated after the second term, as the third virial co-
efficient C and the ones following are hardly ever known. The consequence is that the
virial equation can only be used up to moderate densities (rule of thumb: ρ = 0.5ρc ).
The most widely used equations of state in technical applications are modifica-
tions of the van-der-Waals equation
RT a
p= − 2 (2.11)
v−b v
Invented in 1873 [19]², the van-der-Waals equation was the first equation of state valid
for both the vapor and the liquid phase that could at least qualitatively explain the
pvT behavior of a pure substance, illustrated by the pv diagram in Figure 2.5.
Melting curve
Crystallization curve
p Critical point
pc
Dew point
C B curve
ps (T)
A T>
Boiling Tc
point curve Tc
T<T
c
Fig. 2.5: The pvT behavior of a pure
substance [11].
© Wiley-VCH Verlag GmbH & Co. KGaA.
vc v Reproduced with permission.
2 Just like all authors, I give this citation. However, I will admit that I have not read it: The language
is Dutch, and even if I could speak it I would guess that the way it was used in the 19th century would
not be comprehensible for a nonnative.
The pv diagram for a pure substance is dominated by the large vapor-liquid equilib-
rium region, where the isotherms are horizontal, e.g. between points B and C, indi-
cating that the pressure during condensation or evaporation remains constant. At the
left-hand side of the vapor-liquid equilibrium region, the isotherms are very steep,
meaning that large pressures are needed to lower the specific volume. This is the re-
gion of the liquid phase, and the steep slope means that compressing a liquid does
not result in a major volume change. At the right-hand side there is the vapor region,
where it is easier to compress the substance. At low pressures, the isotherms obey to
the ideal gas law equation (2.9). At higher pressures, the ideal gas law becomes more
and more inaccurate. In the two-phase region, the boiling point line and the dew point
line are connected by horizontal isotherms as mentioned, giving the saturated vapor
volume v 󸀠󸀠 (e.g. point B) and the saturated liquid volume v󸀠 (e.g. point C). With increas-
ing temperature, v󸀠󸀠 and v󸀠 are getting closer to each other. At the critical point, vapor
and liquid become identical. Above the critical temperature (and, correspondingly,
the critical pressure), no phase equilibria between vapor and liquid exist.
Remarkably, one can get from the vapor to the liquid region without crossing the
two-phase region, i.e. a vapor can be gradually transformed into a liquid and vice
versa. E.g. starting from point C, the saturated liquid can be isochorically (i.e. at v =
const.) heated up to a temperature above the critical one, where simultaneously the
pressure is increased to a value above the critical pressure. Then, the substance can
be heated isobarically, and finally, it can be cooled down isochorically to point B. As a
result, a liquid has been turned into a vapor without ever having both phases existing
simultaneously.
When the typical application of an equation of state, the calculation of v at given
T and p, is performed, Equation (2.11) gives a third-degree polynomial:
RT 2 a ab
v3 − (b + )v + v− = 0. (2.12)
p p p
Therefore, the van-der-Waals equation (2.11) is called a cubic equation of state. The
advantage is that it can be solved analytically [11, 20]. One obtains either one real and
two complex or three real solutions. In the latter case, the largest solution corresponds
to a vapor specific volume, and the smallest one corresponds to a liquid volume. The
middle one has no physical meaning. In contrast to Equation (2.9) or the virial equa-
tion (2.10) truncated after the second term, the cubic equations can be used both for
the vapor and for the liquid phase. To decide whether the liquid or the vapor solution
is valid, the vapor pressure is needed.
Of course, an equation of state defined by a continuous function like the van-der-
Waals equation (2.11) cannot reproduce the dogleg at the transition from the one-phase
to the two-phase region. Thermodynamics points out that the equation of state can
evaluate the phase equilibrium using the Maxwell criterion (Figure 2.6).
Liquid and vapor phase are in equilibrium at p = ps if the hatched areas between
p = ps and the equation of state in Figure 2.6 are equal. Analytically, it can be deter-
35
v′ v″
p = ps
p / bar
10
0.5 1 1.5 2
–15
v / kmol/m3
–40
–65 Fig. 2.6: Application of the Maxwell criterion

(propylene, t = 20 °C, Peng–Robinson equa-
–90 tion).
mined by the equation

v󸀠󸀠
󸀠󸀠 󸀠
ps (v − v ) = ∫ p dv . (2.13)
v󸀠
The curvature between v󸀠 and v󸀠󸀠 has no physical meaning, including the negative
values obtained for the pressure. Equation (2.13) is equivalent to the formulation
f 󸀠 = f 󸀠󸀠 , (2.14)
where f is the so-called fugacity, which can be interpreted as a corrected pressure. For
a pure substance, the fugacity is the product of the pressure and the fugacity coeffi-
cient φ:
pφ󸀠 = pφ󸀠󸀠 , (2.15)
where the pressure cancels out.
However, while its theoretical value is unquestioned, the van-der-Waals equation
had limited success in practical applications, where quantitatively correct results are
required. In the second half of the 20th century, several modifications of the van-der-
Waals equation have been developed. The most successful and most widely-used ones
are the Soave–Redlich–Kwong equation (SRK or RKS)
RT a(T)
p= − , (2.16)
v − b v(v + b)
with
R2 Tc2
a(T) = 0.42748 α(T) (2.17)
pc
2
α(T) = [1 + (0.48 + 1.574 ω − 0.176 ω2 )(1 − Tr0.5 )] (2.18)
RTc
b = 0.08664 , (2.19)
pc
and the Peng–Robinson equation (PR)

RT a(T)
p= − , (2.20)
v − b v(v + b) + b(v − b)
where
R2 Tc2
a(T) = 0.45724 α(T) (2.21)
pc
2
α(T) = [1 + m(1 − Tr0.5 )] (2.22)
m = 0.37464 + 1.54226 ω − 0.26992 ω2 (2.23)
RTc
b = 0.0778 . (2.24)
pc
It is remarkable that both the Soave–Redlich–Kwong and the Peng–Robinson equa-
tion need only three substance-specific parameters, i.e. the critical temperature Tc ,
the critical pressure pc and the acentric factor ω, which is defined as
ps
ω = −1 − lg ( ) . (2.25)
pc T=0.7 Tc
Essentially, ω represents the vapor pressure at T = 0.7Tc . For most substances, this
is a temperature close to the normal boiling point. Thus, the critical point and one
specified point of the vapor pressure curve are the only substance-specific information
used. This characterization of a substance is called the three-parameter corresponding
states principle (Tc , pc , and ω). Equations of state using only this input information
are called generalized equations of state.
Equations of state valid for both the vapor and the liquid state can provide all
thermodynamic properties needed for process calculations. Its original well-known
purpose was to be a relationship between p, v, and T in the vapor phase. As seen,
cubic equations of state can also be used to calculate the specific volume in the liq-
uid phase, and as seen in Equations. (2.13) and (2.15), the vapor pressure for a given
temperature can be evaluated. From thermodynamics, an expression for the enthalpy
using a pressure-explicit equation of state can be derived: [11]:
T v
∂p
h(T, v) = h0 + ∫ cid
p dT + ∫ [T ( ) − p] dv + pv − RT . (2.26)
∂T v
T0 ∞
Inserting the equation of state into Equation (2.26), an expression is obtained which
can calculate the specific enthalpy of the substance at any state. The only additionally
required input is the specific isobaric heat capacity in the ideal gas state. The enthalpy
of vaporization for a temperature T is then easily determined by
∆h v = h(T, v󸀠󸀠 ) − h(T, v󸀠 ) . (2.27)
For the Peng–Robinson equation, the most important calculation equations for pure
substances are given in the following:
– Cubic equation for the determination of the volume:
RT 2 a 2bRT RT 2 ab
v3 + (b − ) v + ( − 3b2 − ) v + b3 + b − = 0. (2.28)
p p p p p
– Specific enthalpy [11]:
1 da v + (1 + √2)b
(h − hid (T, v)) = RT(Z − 1) − (a − T ) ln ( ), (2.29)
√8b dT v + (1 − √2)b
with
pv
Z= (2.30)
RT
and
da R2 Tc2 m
= −0.45724 [1 + m(1 − Tr0.5 )] . (2.31)
dT pc √ TTc
– Fugacity coefficient [21]:
bp a v + (1 + √2)b
ln φ = Z − 1 − ln (Z − )− ln ( ). (2.32)
RT √
2 2bRT v + (1 − √2)b
The success of this route is remarkable, taking into account that only three substance-
specific parameters are used (Tc , pc , ω), but limited. The three-parameter correspond-
ing states principle is especially valid for nonpolar compounds. The generalized cubic
equations of state have become standard tools in the oil and gas industries, where the
compounds involved are usually nonpolar. In these cases, the vapor densities and the
vapor pressures are reproduced very well, especially in the high-pressure region. This
can be explained by the fact that the critical point is necessarily reproduced exactly;
the closer the distance to the critical point is, the better the results will be. For polar
compounds (e.g. water, methanol, ethanol), the results are usually remarkably good
for the specific volume in the vapor phase, but the vapor pressure is not reproduced re-
liably enough for process calculations. The enthalpies of the vapor are usually a good,
but not exact, estimate. For the enthalpy of vaporization the same result as for the va-
por pressure is obtained; it is remarkably good for nonpolar substances, but hardly
reliable for polar ones. The enthalpies of a liquid are usually poor for both polar and
nonpolar compounds; the reason for this is explained in Chapter 2.8. The specific vol-
ume of the liquid phase is not even intended to be correct. In most cases, only the
order of magnitude is met. For technical applications, the liquid volume calculated
by cubic equations of state is by far not accurate enough. Instead of the correct liquid
volume, it should only be taken as an auxiliary quantity for the calculation of the vapor
pressure and the liquid enthalpy. In the result view of process simulators, the liquid
volume is usually overwritten by default with results from a liquid density correlation
(Chapter 2.3).
To overcome these limitations, the α-functions of the generalized equations
of state can be replaced by individual ones. An example is the PSRK (Predictive
Soave–Redlich–Kwong) equation, where the generalized α-function (2.18) has been
replaced by
2
α(T) = [1 + c1 (1 − Tr0.5 ) + c2 (1 − Tr0.5 )2 + c3 (1 − Tr0.5 )3 ] . (2.33)
The adjustable parameters c1 , c2 , c3 are usually fitted to vapor pressure data. Tr (re-
duced temperature) is an abbreviation for T/Tc . This way, polar components can be
described as well. In the volume-translated Peng–Robinson equation (VTPR) the α-
function of the Peng–Robinson equation (2.22) has been replaced by the flexible Twu-
α-function
N(M−1)
α(Tr ) = Tr exp [L(1 − TrMN )] , (2.34)
where the adjustable parameters L, M, N can again be fitted to vapor pressure data.
Also, data for cLp can simultaneously be adjusted so that the enthalpy of the liquid can
also be described by the equation [22]. With the volume translation, an attempt was
made to solve the remaining problem with the bad reproduction of the liquid density.
The approach is that the specific volume v in the original equation is replaced by a
term v + c, where c is a volume translation:
RT a(T)
p= − . (2.35)
v + c − b (v + c)(v + c + b) + b(v + c − b)
Therefore, it remains a cubic equation of state, but the results for the specific volumes
for both the vapor and the liquid phase are shifted by the value of c. While this is al-
most negligible for the vapor phase, it causes an improvement for the specific volume
of the liquid phase. The parameter c can be fitted to liquid density data or, if not avail-
able, calculated by a generalized function. However, an acceptable improvement is
restricted to low pressure data, and the densities of liquids can still not be reproduced
with the accuracy required in technical applications. Therefore, in spite of the use of
volume translations it is still recommended to make use of the option to overwrite it
with a liquid density correlation.
Besides the cubic ones, a lot of other equations of state are in use. Only a few of
them with a special importance for process engineering purposes can be mentioned
here.
In process engineering, cases occur where a much higher accuracy in the physical
properties is required. Examples are the power plant processes, the heat pump process
or the pressure drop calculation in a large pipeline. For the description of the complete
pvT behavior including the two-phase region several extensions of the virial equation
were suggested. All these extensions have been derived empirically and contain a large
number of parameters, which have to be fitted to experimental data. A large database
is necessary to obtain reliable parameters. One of the first approaches to equations of
state with higher accuracies has been made by Benedict, Webb, and Rubin (BWR) in
1940 [23, 24], who used 8 adjustable parameters. Their equation allows reliable cal-
culations of pvT data for nonpolar gases and liquids up to densities of about 1.8ρc .
Bender [25] extended the BWR equation to 20 parameters. With this large number
of parameters it became possible to describe the experimental data for certain sub-
stances over a large density range in an excellent way. In the last two decades, the
so-called technical high-precision equations of state have been developed [26–28].
Their significant improvement was possible due to progress in measurement tech-

niques and the development of mathematical algorithms for optimizing the structure
of equations of state. With respect to the accuracy of the calculated properties, their
extrapolation behavior and their reliability in regions where data are scarce, these
equations define the state-of-the-art representation of thermal and caloric properties
and their particular derivatives in the whole fluid range. There is also an important de-
mand to get reliable results for derived properties (cp , cv , speed of sound). Technical
high-precision equations of state are a remarkable compromise between keeping the
accuracy and gain in simplicity. Furthermore, these equations should enable the user
to extrapolate safely to the extreme conditions often encountered in industrial pro-
cesses. For example, in the LDPE³ process ethylene is compressed to approximately
3000 bar, and it is necessary for the simulation of the process and the design of the
equipment to have a reliable tool for the determination of the thermal and caloric prop-
erties. The complexity and limited availability is no longer an issue. Currently, there
are approximately 80 substances for which the data situation has justified the devel-
opment of a technical high-precision equation of state, e.g. water, methane, argon,
carbon dioxide, nitrogen, ethane, n-butane, isobutane, and ethylene. In the FLUID-
CAL software [29], these equations of state have been made applicable for users with-
out special knowledge. The most serious drawback of the high-precision equations of
state is that the concept can hardly be applied to mixtures. So far, only for natural gas
mixtures an equation of state of similar quality has been developed [30].
Table 2.3 illustrates the accuracy demand for technical equations of state.
The φ-φ-approach can be extended to mixtures. Thermodynamics says that the
equilibrium condition Equation (2.14) becomes
px i φ󸀠i = py i φ󸀠󸀠
i (2.36)
for each component involved. Again, the pressure cancels out. The cubic equations of
state can be transformed to mixture applications by mixing rules for the parameters
a and b. The most common ones are
a = ∑ ∑ z i z j (a ii a jj )0.5 (1 − k ij ) (2.37)
i j
b = ∑ zi bi (2.38)
i
Table 2.3: Accuracy demand for technical high-precision equations of state.
ρ(p, T) w ∗ (p, T) cp (p, T) ps (T) ρ󸀠 (T) ρ󸀠󸀠 (T)

p < 30 MPa 0.2 % 1–2 % 1–2 % 0.2 % 0.2 % 0.2 %
p > 30 MPa 0.5 % 2% 2%
3 low density polyethylene.

for both the Peng–Robinson and the Soave–Redlich–Kwong equation of state. As the
mixing rules (2.37) and (2.38) refer to both the vapor and the liquid phase, the neutral
variable z is used for both the vapor and the liquid mole fraction. k ij is an adjustable
binary interaction parameter. It is symmetric (k ij = k ji , k ii = k jj = 0) and has usually
small values (−0.1 < k ij < 0.1). Nevertheless, it has a significant influence on the cal-
culation of phase equilibria and cannot be neglected. The impact on the results for
the liquid and vapor volumes is comparably small. Using Equations (2.37) and (2.38),
the calculation routes for pure components can be applied to mixtures. For the phase
equilibrium calculations, Equation (2.36) looks very easy but is in fact a complicated
equation, as the determination of the fugacity coefficients ends up in long equations,
e.g. for the Peng–Robinson equation [21]
bi p
ln φ i = (z − 1) − ln [ (v − b)]
b RT
a 2 bi v + (1 + √2)b (2.39)
− ( ∑ z j a ij − ) ln [ ].
2√2bRT a j b v + (1 − √2)b
The mixing rules (2.37) and (2.38) have had considerable success as long as only non-
polar substances were involved. When polar compounds are regarded, poor results are
obtained. For an adequate description of systems with polar components using the φ-
φ-approach, the so-called g E mixing rules have to be applied (Figure 2.7). They will be
explained in Chapter 2.7, as understanding them is not possible without knowledge
of the γ-φ-approach.
120 120
100 100
333K 333K
80 80
p/kPa
p/kPa
60 323K 60 323K
40 40
20 20
308K 308K
0 0
0 0.5 1 0 0.5 1
x1, y1 / mol/mol x1, y1 / mol/mol
Fig. 2.7: Experimental and calculated VLE data for the system acetone (1)/water (2) using the
Peng–Robinson equation with k ij (left-hand side) and g E mixing rules (right-hand side) [11].
2.3 γ-φ-approach
2.3.1 Activity coefficients
The activity coefficient is the factor by which I have to miscalculate, to get the correct result al-
though I use the wrong equation.
(Student in his thermodynamics exam. He passed easily.)
The phase equilibrium condition (2.14) can also be elaborated in a different way [11],
using different approaches for the vapor and the liquid phase. The simplest solution
is Raoult’s law:
x i p si = y i p , (2.40)
where the bubble point curve in the pxy diagram is a straight line. Only few binary
systems obey this law, an example is the system benzene/toluene (Figure 2.4, upper
row).
If deviations from Equation (2.40) occur, the activity coefficient γ i is introduced,
which depends on the concentration as well as on the temperature. One can take the
activity coefficient as a factor which corrects the concentration. The equilibrium con-
dition, still not the final one, is then written as
x i γ i p si = y i p . (2.41)
This equilibrium condition is valid in the low pressure region, as the vapor phase is
regarded as an ideal gas. While it is still often used in academics, it has become more
and more outdated in process simulation applications, as the enthalpy calculations
should not be performed using the ideal gas law for the vapor phase (Chapter 2.8).
Instead, a real vapor phase is considered using an equation of state:
φ i (T, p, y i )
x i γ i p si Poyi = y i p pure , (2.42)
φi (T, p si )
with Poyi as the Poynting factor
vLi (p − p si )
Poyi = exp . (2.43)
RT
pure
φ i (T, p, y i ) accounts for the nonideality of the vapor phase, whereas φ i (T, p si ) refers
to the liquid fugacity coefficient at the vapor pressure. Fortunately, at p = p si the liq-
uid fugacity coefficient is equal to the vapor one (Equation (2.15)); thus, in contrast
to Equation (2.36), the equation of state does not need to be valid for both the vapor
and the liquid phase. Therefore, the virial equation (2.10) truncated after the second
term can also be applied. Nevertheless, usually generalized cubic equations of state
are used, as they are the most powerful tools for this purpose with easily accessible
input parameters (Tc , pc , and ω). The Poynting factor corrects the small error caused
by evaluating the fugacity of the pure liquid at its vapor pressure p si instead of the
system pressure p. It can usually be neglected.
The heart of Equation (2.42) is the activity coefficient γ i . The formal character of
the γ i is a correction of the molar concentration, which, however, hardly explains
anything. In fact, the activity coefficient accounts for the intermolecular interactions
between the molecules and the entropic effects. Figure 2.8 shows the typical isother-
mal concentration dependence of the activity coefficient.
4
γi
Fig. 2.8: Typical isothermal concentration

1 dependence of the activity coefficient [11].
0 0.5 1
x1 Reproduced with permission.
Its maximum values (respectively minimum values for systems with negative devia-
tions from Raoult’s law) occur when the component is extremely diluted; at the con-
centration x i → 0. This value is called the activity coefficient at infinite dilution (γ∞
i )
and is characteristic for the illustration of the nonideal behavior.
Nowadays, mainly three equations for the correlation of the γ i are in use; the Wil-
son [31], the NRTL⁴ [32], and the UNIQUAC⁵ equation [33]. They are all based on the
so-called principle of local compositions, which is explained by means of the Wilson
equation in the following paragraph:
In Figure 2.9 it can be seen that in both cells the concentrations of the molecules 1
and 2 are x1 = 37 and x2 = 47 . Due to the intermolecular forces, the local concentrations
can be different. In the left cell, there is a molecule of kind 1 in the center of the cell.
Around this molecule, the concentration of the molecules of kind 1 is x11 = 26 , and the
concentration of the molecules of kind 2 is x21 = 46 . Similarly, in the cell on the right
hand side with a molecule of kind 2 in the center the concentrations are x12 = x22 = 36 .
The concentrations around a molecule and the total concentrations are assumed to be
4 nonrandom two-liquid.
5 universal quasi-chemical.
2 1
1 1 1 2
1 2
2 2 2 2
2 1
Fig. 2.9: Sketch for the explanation of the Local Composition Models [11].
related by Boltzmann factors:

x ji x j exp (−λ ji /(RT))
= , (2.44)
x ii x i exp (−λ ii /(RT))
where the λ terms account for the intermolecular forces. Defining the local mole frac-
tion as
x ii v i
ξi = , (2.45)
∑j x ji v j
γ i can be introduced as a correction factor of the total concentration:
γ i = ξ i /x i . (2.46)
After several mathematical transformations [11] the following expression for the activ-
ity coefficients is obtained:
x k Λ ki
ln γ i = 1 − ln ∑ x j Λ ij − ∑ , (2.47)
j k
∑j x j Λ kj
with Λ as an abbreviation representing the intermolecular forces, which are tempera-

ture-dependent
T
Λ ij = exp (A ij + B ij /T + C ij ln + D ij T) (2.48)
K
The parameters A, B, C, D can be adjusted to experimental binary phase equilibrium
data. The structure of Equation (2.48) often leads to misunderstandings. The first pa-
rameters to be adjusted are the B ij . To account for the temperature dependence, the
A ij parameters can be additionally fitted if corresponding data are available, i.e. phase
equilibrium data at significantly different temperatures or h E data. For better illustra-
tion, Equation (2.48) should be rewritten as
B ij + A ij T + C ij T ln KT + D ij T 2
Λ ij = exp . (2.49)
T
The C ij or D ij parameters are used only if it is justified by the data situation, which is
not often the case. It should be noted that the specific volumina introduced in Equa-
tion (2.45) do not occur in the final equations (2.47) and (2.48). The term A ij is equal
to
vj
A ij = ln , (2.50)
vi
so the ratio of the volumina just represents the term for the linear temperature depen-
dence. In process simulation, it would be awkward anyway if the binary parameters
would depend on the pure component properties which could be subject to change any
time. A more detailed derivation of the Wilson equation can be found in [11]. The NRTL
and UNIQUAC equations are more complicated [11], but for application it is sufficient
to understand their structure:
– Wilson:
ln γ i = f(x i , Λ ij ) , (2.51)
with Λ ij as explained above.
– NRTL:
ln γ i = f(x i , α ij , τ ij ) , (2.52)
with τ ij as a temperature function describing the molecular interactions
T
τ ij = exp (A ij + B ij /T + C ij ln + D ij T) , (2.53)
K
analogous to Λ ij in the Wilson equation and α ij as an additional symmetric (α ij =
α ji ) adjustable parameter for complicated concentration dependences. Usually,
α ij is set to 0.3, a reasonable range of values is α ij = 0.2–0.5. In extreme cases,
this range of values can be exceeded or α ij can be made temperature-dependent.
The coefficients A ij , B ij , C ij , D ij , and α ij are called binary interaction parameters
(BIPs). They play the key role in the thermodynamic model.
– UNIQUAC:
ln γ i = f(x i , τ ij ) , (2.54)
with τ ij analogous to the NRTL equation.
The remarkable issue is that it takes only binary parameters to describe a multicom-
ponent system. Therefore, the effort for the model development is at least limited.
Without this simplifying tool, process simulation would still not be possible up to
now. A famous example from Gmehling [34] indicates that it would take approx.
37 years to carry out the measurements necessary for a model capable to describe a
10-component-system.
There has been a lot of discussion whether the binary interaction parameters have
a physical meaning or not. This is a question which cannot be answered simply with
“yes” or “no”. The tendency of the author is to say “no”. The local composition models
can be mathematically derived [11, 35], where the binary parameters have the charac-
ter of defined energies for removing single molecules from a cell. However, this energy
is not measured; instead, the model with all its assumptions is adjusted to measured
phase equilibrium data. The results obtained are not unique; the parameters strongly
intercorrelate. It can be seen that various completely different parameter pairs are pos-
sible to obtain more or less the same results. This means that the value of a single
parameter cannot have any physical significance without the other parameter in the
pair. Therefore, a single parameter B ij has no significant physical meaning. However,

it can be shown mathematically that any equation describing the activity coefficients
must obey the Gibbs–Duhem equation [11]:
∑ x i d ln γ i = 0 (2.55)
i
The established equations like Wilson, NRTL, or UNIQUAC and several other do so;
thus, they are in some way physically justified. They should be taken as empirical ap-
proaches to fulfill the Gibbs–Duhem equation, meaning that they are able to represent
the correct shape of the course of the activity coefficient as a function of concentration.
0.6
0.55
0.5
0.45
p/bar
0.4
0.35
0.3
0.25
0.2
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
xR134a′ yR134a /mol/mol
Fig. 2.10: The azeotropic system R134a–R218 at T = 220 K [36].
An azeotrope occurs when the vapor pressure curves intersect.

(Completely wrong but often successful explanation for the occurrence of azeotropes.)
Activity coefficients can be used to obtain an understanding how azeotropes occur.

Figure 2.10 shows the well-defined azeotrope of the system R134a(1)–R218(2) at T =
220 K. Neglecting the influence of the fugacity coefficients in Equation (2.42), one can
add up Equation (2.41) for both components in the following form:
p = x1 γ1 p s1 + x2 γ2 p s2 . (2.56)
This is the equation of the boiling point line, i.e. the upper one in Figure 2.10. As
an azeotropic system it shows a clear maximum at the azeotropic point at approx.
x1 = y1 = 0.3. Both pure component vapor pressures are lower than the azeotropic
pressure, while R218 is the light end with the higher vapor pressure. On the right hand
side of the diagram, it is expected that the addition of R218 to R134a increases the pres-
sure, as R218 is the low-boiling component. It is more surprising that the addition of
the high-boiler R134a to the low-boiler R218 on the left hand side of the diagram gives
a pressure increase as well. This is a necessary condition, as starting from both pure
components the maximum in the azeotropic pressure must be reached. Evaluating
Equation (2.56) at the pure R218, one gets for x1 → 0:
p = x1 γ∞
1 p s1 + (1 − x 1 ) ⋅ 1 ⋅ p s2 > p s2 (2.57)
or
γ∞
1 p s1
> 1. (2.58)
p s2
Equation (2.58) is the azeotropic condition: If the product of vapor pressure and activ-
ity coefficient at infinite dilution of the high-boiler is larger than the vapor pressure of
the low-boiler, an azeotrope with a pressure maximum will result.
Similarly, for systems with negative deviations from Raoult’s law one gets the fol-
lowing: if the product of vapor pressure and activity coefficient at infinite dilution of
the low-boiler is smaller than the vapor pressure of the high-boiler, an azeotrope with
a pressure minimum will result.
Solely by the continuity of the boiling point line, the occurrence of an azeotrope
can be predicted just by slopes of the boiling point line caused by the addition of traces
to the pure components. However, the first step for checking whether an azeotrope
occurs should be a look into the database. There are even special monographs where
azeotropic data are collected in a systematic way [37].
A question often asked is which of the three equations, Wilson, NRTL, and UNI-
QUAC, is the best one. In fact, it is not much use to compare large amounts of fitting
results; no clear advantage can be observed. A description of the characteristics might
be more helpful.
The Wilson equation cannot describe systems with a miscibility gap due to math-
ematical reasons [11]. For a system without a miscibility gap the Wilson equation is an
adequate model. The NRTL equation can describe a miscibility gap. The third param-
eter α is a valuable tool if complicated systems have to be correlated. The comparison
with Wilson and UNIQUAC is not fair in these cases, as these equations do not have
this opportunity. Nevertheless, sometimes the third parameter is perceived to be use-
ful and makes NRTL the favorite choice. In a project, all binary subsystems must be
described with the same model; if NRTL gives a significant improvement for one of the
important subsystems, the decision for NRTL is probable. Furthermore, NRTL has a
very popular extension for electrolytes (ELECNRTL), which is fully consistent with the
simple NRTL equation for conventional systems. Thus, if electrolytes occur or even if it
is possible that they might occur, it is usually a good choice to take NRTL. UNIQUAC is
clearly the equation with the strongest physical background. For application, it is not
so popular as it requires the van-der-Waals surfaces and volumina as additional pure
component parameters, which cannot be assigned for all components due to missing
subgroups (e.g. CO2 ). An advantage of UNIQUAC is that it has a combinatorial part,
which takes into account the behavior of molecules differing in size.
In comparison with the φ-φ-approach, the γ-φ-approach has the disadvantage
that supercritical components cannot be treated with this concept. According to Equa-
tion (2.42), as there is no reference point in the liquid phase for a pure supercritical
component. As a workaround, a supercritical component can be treated as a Henry
component, with its reference state at infinite dilution in a pure solvent. The phase
equilibrium condition for such a component is
x i H ij = y i pφ i (T, p, y i ) , (2.59)
with H ij as the so-called Henry coefficient of Henry component i in solvent j. Equa-

tion (2.59) can be applied for low concentrations of the Henry component in the liquid
phase (rule of thumb: x i < 0.03), otherwise, further corrections are necessary, clearly
favoring the φ-φ-approach. The Henry coefficient has the character of a vapor pres-
sure; its meaning becomes clear when Equation (2.59) is applied to a subcritical com-
ponent at low pressure (i.e. φ i (T, p, y i ) = 1) at infinite dilution of the Henry component
in the liquid phase. Compared with Equation (2.41), the Henry coefficient is equal to
H i = p si γ∞
i . (2.60)
The Henry coefficient is a temperature function; the temperature dependence is usu-

ally described using a function like
Hi B
ln = A + + CT + DT 2 , (2.61)
p0 T
where p0 is an arbitrary pressure unit, necessary to make the argument of the loga-
rithm dimensionless.
It should be noted that in contrast to the vapor pressure the Henry coefficient is
not necessarily a function monotonically rising with temperature. It can exhibit well-
defined maxima (e.g. oxygen in water [11, 35]).
For mixed solvents, a mixing rule like
H ij
H i ∑j x j ln p0
ln = (2.62)
p0 ∑j x j
can be applied. In process simulators, often even more complicated mixing rules are
used. It should be mentioned that the averaging in the mixing rules is only performed
with the solvents where Henry coefficients are available. The index j refers only to the
solvent components where a Henry coefficient is given. One must be careful, as these
solvents are not always representative for the whole solvent.
It is often remarked that mixing rules like Equation (2.62) are arbitrary and empir-
ical. From the physical point of view, the application of equations of state, especially
with g E mixing rules (Chapter 2.7), is more justified. Nevertheless, one must realize
that in process calculations the whole gas solubility calculation has the character of
an estimation, giving only a reasonable order of magnitude. The solubility of gases in
liquids usually takes a lot of time to reach equilibrium. For experimental setups, sev-
eral hours are scheduled to get a data point; much more than a gas has in a process
step. Therefore, even relatively large errors in the Henry coefficient are not relevant for
the target of the calculation. For example, if the correct solubility of a gas component
is 100 ppm, an overestimation of 20 % of the Henry coefficient in Equation (2.59) would
yield a solubility of 83 ppm, which is a fully acceptable result in a process calculation.
In principle, excess enthalpies (enthalpies of mixing, see Glossary) can also be
calculated if a correlation for the activity coefficients is available. h E depends on the
temperature derivatives of the activity coefficients:
∂ ln γ i
h E = −RT 2 ∑ x i ( ). (2.63)
i
∂T
However, the physical background of the equations for the activity coefficient is not
sufficient so that application of Equation (2.63) usually gives wrong results unless the
parameters are temperature-dependent and have been fitted to both phase equilib-
rium and excess enthalpy data.
The great advantage of the γ-φ-approach is that all properties are described by
individual equations that have no influence on each other. The phase equilibrium is
described only by the activity coefficients, which do not need to be varied if any other
property is changed. Independent and highly accurate correlations are available for
each property.
2.3.2 Vapor pressure and liquid density
In contrast to the φ-φ-approach, in the γ-φ-approach the liquid density, the vapor
pressure and the specific enthalpy are determined independently with separate corre-
lations, giving the opportunity of yielding better accuracies and having a more conve-
nient workflow, as changes in one property do not affect the others. While the enthalpy
calculation needs an own chapter (Chapter 2.8), the correlations and estimations for
the liquid density and the vapor pressure can be briefly discussed here.
The density of a pure substance or a mixture is a fundamental quantity in any
process calculation. The vapor density as a function of temperature and pressure is
determined by the equation of state chosen in Equation (2.42). The liquid density of
pure components is treated only as a function of temperature, the pressure effect on
the density can usually be neglected. Appropriate correlations and their coefficients
are the Rackett equation
A
ρL = D
(2.64)
B1+(1−T/C)
and the PPDS⁶ equation

ρL ρc
= + A(1 − Tr )0.35 + B(1 − Tr )2/3 + C(1 − Tr ) + D(1 − Tr )4/3 , (2.65)
kg/m3 kg/m3
which is usually slightly more accurate.

It is important to mention that the liquid density mixing rule is based on the spe-
cific volume, i.e. the reciprocal value of the density:
1 xi
=∑ . (2.66)
ρL i
ρ L,i
Thermodynamics says that there is a so-called excess volume (see Glossary), i.e. a sys-
tematic deviation from Equation (2.66). If the liquid density is calculated using an
equation of state, as it is the case in the φ-φ-approach (Chapter 2.2), this excess vol-
ume is accounted for automatically. Nevertheless, no advantage can be taken, as it
is usually not described quantitatively correct, and in most cases the density of the
pure substances is badly reproduced so that even with a correction Equation (2.66)
yields better results (p. 38). The maximum error caused by neglecting the excess vol-
ume can be quantified regarding the system exhibiting the largest excess volume: To
the knowledge of the author, it is in fact ethanol-water, and the maximum excess vol-
ume observed is approx. 3.5 % [11].
Liquid densities can be estimated e.g. with the COSTALD equation [38]. It can in
principle be written as
1
= v∗ ⋅ f (T, Tc , ω) , (2.67)
ρL
meaning that the liquid density can be estimated with a generalized function depend-
ing only on Tc and ω. Additionally, the parameter v∗ is used, which can be adjusted
to one or more data points. In this case, the procedure is very accurate; up to tem-
peratures below 0.95 Tc the error is usually below 2 %. If no data point is available,
one can use the critical volume for v∗ , which of course increases the risk but often
yields surprisingly good results. It should be mentioned that the COSTALD equation
has weaknesses if polar components are involved.
The vapor pressure is the most important quantity in thermodynamics. It is de-
cisive especially in the simulation of distillation columns. Furthermore, it is directly
related to the enthalpy of vaporization via the Clausius–Clapeyron equation (Chap-
ter 2.8). The vapor pressure is an exponential function of temperature, starting at the
triple point and ending at the critical point. It comprises several orders of magnitude,
therefore, a graphical representation usually represents only part of its characteris-
tics. Figure 2.11 shows two diagrams, with linear and logarithmic axes for the vapor
pressure of propylene. The linear diagram makes it impossible to identify even the
6 physical property data service

50 5
45
40 0
ln (Ps/bar)
35
30 –5
Ps/bar
25
20 –10
15
10 –15
5
0 –20
80 130 180 230 280 330 80 130 180 230 280 330
(a) T/K (b) T/K
Fig. 2.11: Typical vapor pressure plots as a function of temperature [11].

1
0.8
0.6
0.4
Δps/ps/%
0.2
0
–0.2
–0.4
–0.6
–0.8
150 200 250 300 350
T/K
Fig. 2.12: Deviation plot for the fit of a vapor pressure equation [11].
qualitative behavior at low temperatures, whereas in the logarithmic diagram only

the order of magnitude can be identified on the axis. At least, logarithmic diagrams
allow the comparison between vapor pressures of different substances, e.g. deciding
whether they intersect or not. On the other hand, many process calculations require
extremely accurate vapor pressure curves, e.g. the separation of isomers by distilla-
tion. For the visualization of a fit of a vapor pressure curve, special techniques must
be applied, the simple graphical comparison between experimental and calculated
data in a diagram is not possible. For this purpose, a deviation plot is useful (Fig-
ure 2.12).
Most companies use parameter databanks. When a parameter database is set up,
e.g. the one in a commercial process simulation program, it cannot be known in ad-
vance at which conditions exact vapor pressures will be required. For distillation ap-
plications, a high accuracy is often required, especially when components with similar
vapor pressures have to be separated in distillation columns. Simple vapor pressure
equations cannot be applied for the whole temperature range from the triple point to
the critical point. More capable vapor pressure correlations are needed, the most pop-
ular ones are the extended Antoine equation and the Wagner equation. In principle,
the extended Antoine equation is a collection of various useful terms:
ps B T T T G
ln =A+ + D + E ln + F ( ) , (2.68)
p0 T/K + C K K K
where p0 is an arbitrary reference pressure, usually the pressure unit. A very useful
correlation is the Wagner equation
ps 1
ln = [A(1 − Tr ) + B(1 − Tr )1.5 + C(1 − Tr )3 + D(1 − Tr )6 ] , (2.69)
pc Tr
where Tr = T/Tc (reduced temperature). It can correlate the whole vapor pressure
curve from the triple point to the critical point with an excellent accuracy. The Wag-
ner equation has been developed by identifying the most important terms in Equa-
tion (2.69) with a structural optimization method [39]. The extraordinary capability of
the equation has also been demonstrated by Moller [40], who showed that the Wagner
equation can reproduce the difficult term ∆h v /(z V − z L ) reasonably well. Eq. (2.69) is
called the 3-6-form, where the numbers refer to the exponents of the last two terms.
Some authors [41] prefer the 2.5-5-form, which is reported to be slightly more accu-
rate. For the application of the Wagner equation, accurate critical data are required.
As long as the experimental data points involved are far away from the critical point
(e.g. only points below atmospheric pressure), estimated critical data are usually suffi-
cient. As the critical point is automatically met due to the structure of the equation, the
Wagner equation extrapolates reasonably to higher temperatures, even if the critical
point is only estimated. However, like all vapor pressure equations it does not extrapo-
late reliably to lower temperatures. Sometimes, users of process simulation programs
calculate vapor pressures beyond the critical point, although it is physically meaning-
less. If the Wagner equation is applied above the critical temperature, it will yield a
mathematical error. Therefore, the simulation program must provide an extrapolation
function that continues the vapor pressure line with the same slope. For the particular
parameters of the Wagner equation, the following ranges of values are reasonable for
both forms:
A = −9 . . . −5
B = −10 . . . 10
(2.70)
C = −10 . . . 10
D = −20 . . . 20 .
If these ranges are exceeded, one should carefully check the critical data used and
the experimental data points for possible outliers. Coefficients for the Wagner equa-
tion can be found e.g. in [41–43]. For a vapor pressure correlation, average deviations
should be well below 0.5 %. Data points with correlation deviations larger than 1 %
should be rejected, as long as there are enough other values available. Exceptions can
be made for vapor pressures below 1 mbar, as the accuracy of the measurements is
lower in that range. The structure of the deviations should always be carefully inter-
preted. A guideline is given in [11].
Despite this high accuracy demand for vapor pressures, there is also a need for
good estimation methods. Often, a lot of components is involved in a distillation
process. Not all of these components are really important, however, one should know
whether they end up at the top or at the bottom of a distillation column. In many
cases, a measurement would even be not possible, as the effort for the isolation and
purification of these components might be too large. Estimation methods are mostly
applied to medium and low pressures for molecules with a certain complexity, as
small molecules usually have well-established vapor pressure equations, whereas
large molecules often have a volatility which is so low that a purification of the
substance for measurement by distillation is not possible. The estimation of vapor
pressures is one of most difficult problems in thermodynamics. Due to the exponen-
tial relationship between vapor pressure and temperature, a high accuracy must not
be expected. Deviations in the range of 5–10 % have to be tolerated. Thus, estimated
vapor pressure correlations should not be used for a main substance in a distillation
column to evaluate the final design, however, they can be very useful to decide about
the behavior of side components without additional measurements.
Different estimation methods are discussed in [11]. At least, one data point, usu-
ally the normal boiling point, should be known. Most methods use a vapor pressure
equation with two adjustable parameters; therefore, a second piece of information is
necessary. It can be generated by a group contribution method (e.g. Rarey method
[40]) or by a second data point, which can be either a genuine data point or a point
based on an estimation method itself, e.g. the critical point. An example for the latter
case is the application of the Hoffmann–Florin equation [44]
ps 1 T T
ln = α + β[ − 7.9151 ⋅ 10−3 + 2.6726 ⋅ 10−3 lg − 0.8625 ⋅ 10−6 ] , (2.71)
p0 T/K K K
with the adjustable parameters α and β. It has the advantage that it can easily be trans-
formed to the extended Antoine equation (2.68) by
A = α − 7.9151 ⋅ 10−3 β
B=β
D = −0.8625 ⋅ 10−6 β
(2.72)
2.6726 ⋅ 10−3
E= β
ln 10
C = F = G = 0.
If no single data point is available, one must estimate even the normal boiling point
[11, 45]. In this case, one can hardly rely on the results obtained, however, as long as
no better information is available, there is no choice. When a vapor pressure has to be
estimated, one should also have a look at vapor pressure curves of components which
have a similar structure. Just by defining a constant vapor pressure ratio between the
two components, one can at least obtain a reasonable vapor pressure curve.
Vapor pressures play the most decisive role in distillation if isomers have to be
separated by distillation. In this case, the separation factor depends only on the ra-
tio between the vapor pressures as the activity coefficients between isomers can be
set to 1 as an approximation⁷. If the vapor pressures of the isomers are close, a large
number of theoretical stages are necessary, and its determination is very sensitive to
the separation factor. In these cases, it is strongly recommended not to rely on data
from the literature, not even on good data. Instead, the vapor pressure of the isomers
should be measured as accurately as possible in the same apparatus and on the same
day to avoid any systematic measurement errors.
2.3.3 Association
Another advantage of the γ-φ-approach is that substances showing association in the

vapor phase can be described. These substances are the carboxylic acids like formic
acid, acetic acid⁸ or propionic acid, which form dimers in the vapor phase, and hydro-
gen fluoride, which forms hexamers. These substances are involved in many chemical
processes, and the deviations from the ideal gas law are significant even at low pres-
sures. For example, the compressibility factor of acetic acid at the normal boiling
point, which is expected to be close to 1, is ZNBP = 0.6. Up to now, no equation of
state valid for both the vapor and the liquid phase has been available in this case. The
corresponding equation of state does not need to be valid for both the vapor and the
liquid phase; with the γ-φ-approach, it is sufficient to cover only the vapor phase. The
formation of associates is treated as a chemical reaction in equilibrium. As an illustra-
tion, formic acid as a substance forming only dimers is regarded. The association can
be described with the law of mass actions:
z2
K2 = , (2.73)
z1 (p/p0 )
2
where K is the equilibrium constant for the reaction
2 HCOOH 󴀕󴀬 (HCOOH)2 .
z1 is the true concentration of the monomer, while z2 denotes the true concentration
of the dimers in the mixture. p0 is simply the pressure unit. The equilibrium constant
can be correlated by
B2
ln K2 = A2 + . (2.74)
T
7 which is, however, not always correct.

8 For acetic acid, an additional tetramer formation is often considered to obtain more accurate results.
The sum of true mole fractions z is equal to one
z2 + z1 = 1 (2.75)
Combining Equations (2.73) and (2.75), z1 can be determined by
√1 + 4K2 (p/p0 ) − 1
z1 = . (2.76)
2K2 (p/p0 )
Assuming that the ideal gas equation is valid, the specific volume can be determined
by
RT 1
v= . (2.77)
p z1 + 2z2
Figure 2.13 illustrates the difference to conventional equations of state like the cubic
ones. Case (a) represents the situation in an ideal gas phase; there are no forces
between the molecules. Case (b) represents a real vapor phase, where the molecules
attract or repulse each other by intermolecular forces. Vapor phases like this are
typically modeled with the equations of state described in Chapter 2.2 like the cubic
equations. Case (c) denotes the association in the vapor phase. The model (Equa-
tions (2.73)–(2.77)) takes into account that associates are formed, but it is still assumed
that no intermolecular forces are exerted. The nonideality in case (c) is only achieved
by the formation of associates. This approximation is sufficient for low pressures.
For higher pressures, a good model for substances showing vapor phase association
would have to take the intermolecular forces into account as well (case (d)). However,
an appropriate model for this situation has not yet been introduced.
For associating substances, the heat capacity of the vapor phase shows a well-
defined maximum (Figure 2.14). At low temperatures, the dimerization as the exother-
mic reaction is preferred in the equilibrium, and all molecules are dimerized. When
(a) (b)
(c) (d)
Fig. 2.13: Illustration of the modeling of vapor phases.

6 p = 1.013 bar
5
cp/J/gK
4
p = 6 bar
3
2 p = 0.332 bar
1
ideal, p = 0
0 Fig. 2.14: Specific isobaric heat capacity
350 450 550 650 750 of acetic acid vapor at different pressures [11].
T/K © Wiley-VCH Verlag GmbH & Co. KGaA.
–0.332 bar –1.013 bar Reproduced with permission.
the temperature rises, the dimers are split. For this endothermic reaction, energy is
required which is not used for increasing the temperature. Therefore, the heat capac-
ity increases drastically. With increasing temperature, the number of dimers which
can be split decreases. The heat capacity passes a maximum and comes down to the
normal value of the ideal gas. For the sizing of heat exchangers, this effect must be con-
sidered. The curvature is even more dramatic for HF with its hexamers, where the peak
of the maximum can be up to 40 times higher than the ideal gas heat capacity [46].
2.4 Electrolytes
Water is one of the strangest substances that occurs in chemical engineering. Its phys-
ical properties are not comparable to the behavior of other substances. With a low
molecular weight of 18 g/mol, its normal boiling point of tb = 100 °C is incredibly
high, as well as its critical point (tc = 374 °C, pc = 221 bar). The well-known maxi-
mum in the density of liquid water at t = 4 °C is technically not important, but the
course of the liquid density as a function of temperature is extraordinarily flat. Be-
tween 0 °C and 50 °C, the density decreases just by 1 %. For comparison, n-heptane
with a similar normal boiling point (98.4 °C) has a density decrease of more than 4 %
in the same temperature range. When water crystallizes, it expands significantly by
almost 10 %, whereas other substances reduce their density as expected. The specific
heat capacity of liquid water (approx. 4.2 J/(g K)) is about twice as large as for typical
organic substances, but the most remarkable physical property is the enthalpy of va-
porization. At t = 100 °C, it is approx. 2250 J/(g K), more than seven times larger than
for n-heptane as a typical organic substance.
The reason for this behavior is the strong polar character of the molecule. It is not
linear; the two bonds between oxygen and hydrogen form an angle of approx. 105°.
The oxygen atom is a strongly electronegative, it attracts the electrons of the bond
with their negative charges so that they concentrate at the oxygen atom, while the
hydrogen atoms become positively charged. Therefore, the water molecules arrange
in a kind of lattice.
Moreover, the strong polarity of the water molecule is the reason why it is an ex-
cellent solvent for many electrolytes. An electrolyte is a substance which conducts
electric current as a result of its dissociation into positively and negatively charged
ions in solutions or melts. Ions with a positive charge are called cations, ions with
negative charge anions, respectively. The most typical electrolytes are acids, bases,
and salts dissolved in a solvent, very often in water.
The total charge of an ion is a multiple of the elementary charge (e = 1.602 ⋅
10 −19 C), given by the number z. Examples are:
H3 O+ z=1
−
Cl z−1
Ca 2+
z=2
SO2−
4 z = −2 .
In a macroscopic solution, the sum of charges is always zero, since the solution is
always neutral. Otherwise, an electric current would occur.
For electrolyte solutions, the particular ions are formed by dissociation reactions
like
NaCl → Na+ + Cl−
CaSO4 → Ca2+ + SO2−
4
H3 PO4 + H2 O → H3 O+ + H2 PO−4
H2 PO−4 + H2 O → H3 O+ + HPO2−
4
+
4 + H2 O → H3 O + PO4 .
3−
HPO2−
The H+ ion cannot exist as a pure proton; it is always attached to a water molecule
H2 O, giving H3 O+ . Strong and weak electrolytes can be distinguished. While strong
electrolytes like HCl, H2 SO4 , HNO3 , NaCl, or NaOH dissociate almost completely, weak
electrolytes do so only to a small extent. Sometimes, their electrolyte character plays a
secondary role and can often be neglected. Examples are formic acid (HCOOH), acetic
acid (CH3 COOH), HF, H2 S, SO2 , NH3 , or CO2 [11].
The molecular structure of an electrolyte solution is significantly determined by
the electrostatic interactions between the charged ions (Coulomb-Coulomb interac-
tions) and by the long-range interactions of the charged ions and the dipole moments
of the solvent (Coulomb-dipole-interactions). Figure 2.15 illustrates the schematic dis-
tribution of water as a strongly polar solvent around a cation and an anion.
As mentioned, the oxygen atom in the water molecule has a negative partial
charge due to its high electronegativity. Therefore, the water molecule in the vicinity
of an ion is arranged in a way that the oxygen atom is directed towards the positively
(–)
(+) (+) (+)
(+)
(–)
(+) (–) (+) (+)

+ (–) (+) –
(–) (–)
(+) (+) (+)
(+)
(+)
(–) (+)
(+) (–)
(+)
Fig. 2.15: Structure of an aqueous electrolyte solution [11].

charged cations. Vice versa, the hydrogen atoms in the water molecule are partially
positively charged and oriented towards the anions. Around the ions, a shell of solvent
molecules is formed.
The corresponding procedure is called solvation. It is an exothermic process. On
the other hand, the dissociation of the electrolyte is an endothermic process, because
the ionic lattice has to be destroyed, which is connected to a need of energy. The
overall heat of solution is the sum of both contributions. It is usually dominated by
the solvation and therefore remains exothermic. However, there are many exceptions.
Electrolytes in their dissociated form are not volatile and remain completely in the liq-
uid phase. However, it can happen that liquid drops containing electrolytes are subject
to entrainment (Chapter 9).
The process simulation programs offer models for electrolyte solutions which de-
scribe the arrangement above. Not necessarily the best, but the most widely used one
is the electrolyte NRTL model, often also called Chen model [47, 48]. Its obvious ad-
vantage is the compatibility with the conventional NRTL activity coefficient model.
This is the most important property of an electrolyte model, often even more impor-
tant than its accuracy. During a project, process simulation might start in a part of
the plant where electrolytes are not involved. Later, when it is extended, electrolyte
components occur, and defining these components simply as heavy ends is always
tried first⁹. Although of course not state of the art, this is often a feasible approach
as long as the electrolyte concentration is low and not a decisive issue in the process.
9 This can simply be accomplished by assigning the same properties to the component as water, and
then overwriting the molecular weight with the correct value for maintaining the mass balance and,
also, overwriting the vapor pressure with an equation giving negligible values over the whole concen-
tration range.
When it finally turns out that electrolytes occur in a way that their character must be
correctly described, it is useful if at least the BIP matrix does not need to be revised.
The NRTL electrolyte model consists of the theoretically derived Debye–Hückel
term for the long ranging interactions due to the charges of the ions and an NRTL
based term for the short ranging interactions. Still, the local composition concept is
applied. The model is not restricted to systems of electrolytes just with water, other
solvents and side components are possible. However, the database is mainly based on
aqueous systems. Different kinds of parameters occur:
– the normal binary parameters between molecules;
– pair parameters between ion pairs and molecules;
– pair parameters between different ion pairs.
Nowadays, the possible electrolyte reactions and the species produced are generated
by the process simulation program. It is up to the user to neglect them or take them
into account. For instance, the dissociation reaction of ammonia
NH3 + H2 O → NH+4 + OH−
can easily been neglected if just the system NH3 /H2 O is regarded, as only a small
fraction of the ammonia dissociates, which is negligible. In a caustic environment,
even this hardly ever happens. However, in an acid environment, the ammonia will
dissociate, and in fact completely at low pH values. A famous example is the system
NH3 /CO2 /H2 O, where both NH3 and CO2 are weak electrolytes, but they keep the other
component in the solution as NH3 is a caustic and CO2 is an acid component.
In the process simulation program, the equilibrium constants for the generated
dissociation reactions are usually provided automatically, as well as the pair parame-
ters. Note that the pair parameters really refer to a pair of ions and not to single ones.
An example for the application of the NRTL electrolyte model is given in [11].
There are two options for the notation of the electrolytes. In the true component
approach, the ions are listed as ions. The advantage is that it is listed what really hap-
pens, the disadvantage might be that it is more difficult to keep the overview. In the
apparent component approach, the ions are recombined to components for the result
list. This is easier for discussion but does not always work in a plausible way. It can
happen that according to the balance components like NaOH and HCl coexist in the
aqueous solution, which one can hardly imagine.
2.5 Liquid-liquid equilibria
With increasing activity coefficients, two liquid phases with different compositions are
formed (miscibility gap). The concentration differences of the compounds in the dif-
ferent phases can be used for example for the separation by extraction. In distillation
processes, the liquid-liquid equilibrium (LLE) is often used when a decanter separates
2.5 Liquid-liquid equilibria | 59
the condensate of the top product into two liquid phases. The knowledge of the VLLE
(vapor-liquid-liquid equilibrium) behavior is of special importance for the separation
of systems by heteroazeotropic distillation. Many engineers are of the opinion that
the formation of an LLE does not take place in distillation columns. In fact, it does,
however, in contrast to a phase equilibrium arrangement the two phases do not sep-
arate due to turbulences (tray columns) or form thin layers which trickle down the
internals of the column (packed columns). In both cases, it is useful to treat the two
liquid phases as one homogeneous liquid for the determination of the overall liquid
composition and the physical properties of the liquid phase. For the phase equilib-
rium calculations, there is no other way than to consider the liquid phase split to get
the correct vapor composition. Liquid-liquid equilibria can be evaluated by the iso-
activity criterion:
(x i γ i )I = (x i γ i )II , (2.78)
as the product x i γ i is also called activity ai .
At moderate pressures, the liquid-liquid equilibrium behavior as a function of
temperature only depends on the temperature dependence of the activity coefficients.
For the calculation of LLE again g E -models like NRTL, UNIQUAC, or equations of state
with a g E mixing rule (Chapter 2.7) can be used, whereas Wilson is not appropriate [11].
The formation of two liquid phases can result in the formation of binary and higher
heteroazeotropes, which can for instance be used for the separation of systems like
ethyl acetate–water (Figure 2.18).
Ternary LLEs can be illustrated in a triangle diagram (Figure 2.16). It is more dif-
ficult to calculate reliable liquid-liquid equilibria (LLE) of a system containing three
or more components using binary parameters than to describe vapor-liquid or solid-
liquid equilibria (Chapter 2.6). The reason is that in the case of LLE the activity coeffi-
cients have to describe the whole concentration dependence correctly, whereas in the
case of the other phase equilibria (VLE, SLE) the activity coefficients primarily have to
account for the deviation from ideal behavior (Raoult’s law resp. ideal solid solubility).
This is the main reason why up to now no reliable prediction of a multicomponent
LLE behavior (tie lines) using binary parameters is possible. Fortunately, it is quite
easy to measure LLE data of ternary and higher systems at least up to atmospheric
pressure. Binary parameters can be fitted to ternary data as well, and in this way LLEs
can at least be correlated.
Also, even the fit to a binary mixture is often a bit more exhausting, as priorities
must be set. It hardly ever happens that a set of binary parameters can represent both
the vapor-liquid equilibrium of systems with an LLE (the so-called vapor-liquid-liquid
equilibrium VLLE) and the miscibility gap itself. It must be decided which capabili-
ties the parameters should have. Often, two different parameter sets are used, one for
LLE and one for VLLE. In most cases, the LLE data are more significant, especially for
systems with wide miscibility gaps.
Ternary diag for WATER/ETHANOL/ETAC
0.05 0.95
0.10 0.90
0.15 0.85
0.20 0.80
0.25 0.75
0.30 0.70
0.35 0.65
ate
0.40 0.60
Mo
cet
0.45 0.55
le f
yl a
rac
0.50
eth
0.50
eth
rac
0.55 0.45
an
le f
ol
0.60 0.40
Mo
0.65 0.35
0.70 0.30
0.75 0.25
0.80 0.20
0.85 0.15
0.90 0.10
0.95 0.05
0.05 0.15 0.25 0.35 0.45 0.55 0.65 0.75 0.85 0.95
Mole frac water
Fig. 2.16: Liquid-liquid equilibrium of the ternary system ethyl acetate–water–ethanol at t = 50 °C.
Moreover, the binary parameters for LLEs cannot be simply transferred from binary to
ternary and multicomponent mixtures. Doing so, one obtains more or less an estima-
tion. One should be aware that for a good description of multicomponent LLE at least
data from ternary mixtures are necessary.
Often, when a colleague tells me that his simulation results look strange, the first thing I do is
check whether the VLLE option is actually used for systems with a miscibility gap. Forgetting it is
a very common error in the simulation of vapor-liquid separations. Simulation results may look
strange in many ways. For this case, the trigger is that the temperature is way too low, not just by
a few degrees but by 100 K or so.
(Jo Sijben)
Calculating a phase equilibrium, there are two options in a process simulator. One is
the common VLE calculation. If it is known that miscibility gaps occur in the system,
the option VLLE (3-phase equilibrium) must be chosen so that the simulator checks
2.6 Solid-liquid equilibria | 61
whether there is an LLE before the equilibrium with the vapor phase is evaluated. If
this is forgotten, one gets strange phase equilibrium diagrams (Figure 2.17). The boil-
ing and the dew point curve then appear to be complete nonsense. The correct and
reasonable result is obtained with the 3-phase flash (Figure 2.18). In a flash in a process
simulation, the error is not so easy to detect, but whenever the result is not plausible,
it should be checked whether the VLLE option is necessary and chosen. One could
easily say that VLLE should always be chosen, but if it turns out to be not necessary, a
lot of calculation time has been wasted, as the VLLE option is quite time-consuming
in comparison with a simple VLE.
Generally, it can be said that systems with water and nonpolar organic substances
have strong intermolecular interactions and often form miscibility gaps. Therefore, at
least all the BIPs with water should be assigned in any project (Chapter 2.9).
3
2.5
2
p / bar
1.5
1
t = 80°C
0.5
0 Fig. 2.17: Calculation of the system
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
ethyl acetate–water at t = 80 °C with
x, γETAC / mol/mol the 2-phase flash.
1.6
1.4
1.2
1
p/bar
0.8
0.6
0.4 t = 80°C
0.2
0 Fig. 2.18: Calculation of the system
0 0.2 0.4 0.6 0.8 1
ethyl acetate–water at t = 80 °C with
x, γETAC/mol/mol the 3-phase flash.
2.6 Solid-liquid equilibria
Solid-liquid equilibria (SLE) are used for the synthesis and design of crystallization
processes, but taking them into account is also important to avoid undesired solids for-
mation. Information about solid-liquid equilibria can also be used for the adjustment
of binary parameters. SLEs are more complicated than VLEs or LLEs. Different types of
SLEs have to be distinguished, depending on the mutual solubility of the components
in the solid and in the liquid phase. However, the most important one, the simple
eutectic system, is comparably easy, and it is the only one which does not require a
specialist.
The eutectic system is characterized by the total immiscibility of the components
in the solid phase. This is advantageous for crystallization, as the crystallized phase
has a high purity. One theoretical stage is sufficient to obtain the pure compounds [11].
Usually, there are liquid mechanical inclusions so that the crystallization must at least
be repeated once, but from thermodynamics alone a pure solid phase is generated.
Fortunately, about 80 % of the systems behave in this way. Figure 2.19 shows the solid-
liquid equilibrium of the eutectic system benzene–naphthalene. Both solid phases
crystallize in pure form. Generally, solids are formed at a low temperature. Consider a
mixture with x1 = 0.5. When it is cooled, it reaches the liquidus line at T ≈ 320 K. The
first naphthalene crystals are formed. Cooling down further, the amount of solids in-
creases according to the lever rule, while the concentration of the liquid phase moves
towards the eutectic one. At 300 K, it is approx. x1 = 0.7. When the eutectic tempera-
ture is reached, both components crystallize forming pure solid phases. The eutectic
temperature is lower than the melting points of the participating pure components.
Thermodynamics does not give information about the shape of the crystals and the
amount of liquid included. It is the art of crystallization to form the crystals in the
desired way (Chapter 7.3).
360
340
320
T/K
300
280
Fig. 2.19: Solid-liquid equilibrium of the
260 eutectic system benzene–naphthalene [11].
0 0.5 1
x1 / mol/mol Reproduced with permission.
For eutectic systems, the equilibrium condition can be written as [11]

∆h m,i T cLp,i − c Sp,i T m,i − T T m,i
ln (x i γ i ) = − (1 − )+ ( − ln ). (2.79)
RT T m,i R T T
The heat capacities of the solids are not often known, and, fortunately, the difference
in the corresponding term has a tendency to cancel out. Therefore, Equation (2.79) is
usually simplified to
∆h m,i T
ln (x i γ i ) = − (1 − ). (2.80)
RT T m,i
2.7 φ-φ-approach with g E mixing rules | 63
For an evaluation of Equation (2.80), only the activity coefficient and melting tempera-
ture and enthalpy of fusion as pure component properties must be known. Solid-liquid
equilibria are not very sensitive to pressure but much more to temperature. As the ac-
tivity coefficient γ i of the component i is strongly concentration-dependent, the mole
fraction x i in the liquid phase must be evaluated iteratively.
In process simulation programs, special solid components can be introduced
which do not take part in other phase equilibria. For crystallization, reaction blocks
can be introduced where the liquid component is transformed into the solid or vice
versa. The “reaction equilibrium” is calculated according to Equation (2.79) or (2.80),
respectively.
2.7 φ-φ-approach with g E mixing rules
Nowadays, in process simulation in the chemical industry the requirement for accu-
racy and reliability on the one hand and the occurrence of polar components on the
other hand make clear that advanced cubic equations of state with g E mixing rule
and individual α-functions should be preferred to the generalized equations of state
like Peng–Robinson or Soave–Redlich–Kwong with the quadratic mixing rule (Equa-
tion (2.37)). The generalized ones might still be tolerable in hydrocarbon processes;
however, the advanced cubic equations of state have no drawback there. The g E mix-
ing rule adopts the concept of the activity coefficients for use in the mixing rule for
the parameter a. Examples are the Predictive Soave–Redlich–Kwong equation (PSRK)
and the Volume-Translated Peng–Robinson equation (VTPR). For PSRK, the g E mixing
rule is
a a ii 1 gE b
= ∑ xi − ( 0 + ∑ x i ln ) , (2.81)
bRT i
b i RT 0.64663 RT i
bi
where g0E denotes the Gibbs excess energy at p = 0, i.e. at low pressure:
g E = RT ∑ x i ln γ i . (2.82)
i
To calculate the γ i , any appropriate equation like Wilson, NRTL, or UNIQUAC or a pre-
dictive approach (Chapter 2.10) can be used.
With the g E mixing rules, the general inability of the φ-φ-approach to describe
mixtures with polar components can be overcome.
There has been a great deal of discussion on whether the φ-φ-approach Equa-
tion (2.15) or the γ-φ-approach Equation (2.42) is the more favorable option for phase
equilibrium calculation, often in a more or less ideological way. The following items
point out the particular pros and cons:
– If supercritical components dissolve in the liquid phase to a considerable amount,
it is nowadays obligatory to use the φ-φ-approach, preferably with a g E -mixing
rule.
– Theoretically, the γ-φ-approach is valid in the whole subcritical region. However,

in the region close to the critical temperature of a component (approx. T > Tc −
15 K) the φ-φ-approach is more reliable.
– As long as supercritical components dissolve in the liquid phase to a minor ex-
tent, (e.g. nitrogen in water), there is no real disadvantage if Henry’s law with
the corresponding mixing rule is used instead of the φ-φ-approach. Although the
mixing rule Equation (2.62) is more or less arbitrary, it leads to the correct order
of magnitude of the gas solubility. As in processes these equilibria are usually not
reached, this information from process simulation is fully sufficient.
– If associating components are involved, the γ-φ-approach is obligatory. So far,
there is no established equation of state valid for both vapor and liquid phase for
associating components.
– One must be aware that with the φ-φ-approach all thermodynamic properties
(ρ L , p s , ∆h v , cLp ) are calculated just by the equation of state. This means that any
extension of the database for one property affects all other properties, giving a lot
of work. Furthermore, due to the limited number of parameters used, the accuracy
of the particular quantities is often not sufficient. An exception are, of course, the
high-precision equations of state for pure components. There is no option for set-
ting priorities in the enthalpy calculations (Chapter 2.8). With the γ-φ-approach,
all properties can be correlated separately and with the required accuracy, as long
as enough data are available.
– An often cited prejudice is that with the φ-φ-approach only data for a correlation
for cidp is required. In fact, this statement is not well founded. From the formal
point of view, all the thermodynamic properties as listed above can be calculated
without further data. However, in this case one has no information about the ac-
curacy of the equation of state. For process calculations, it is necessary to compare
the values obtained with experimental data and then probably adjust the param-
eters involved; a procedure with the disadvantage mentioned above that the pa-
rameters influence all quantities. Finally, a responsible use of the φ-φ-approach
requires presumably more preparation work than the γ-φ-approach.
– An argument against the γ-φ-approach is that it does not include the pressure
dependence of the activity coefficient. At high pressures, even small values of the
excess volume could have an influence on the results [49]. There are several ap-
proaches to encounter this argument. First, the fitting of data at the corresponding
temperatures and pressures should achieve some kind of error compensation so
that the data are still represented [49]. Second, a further correction taking the
excess volume into account could be included in the phase equilibrium condi-
tion (2.42) [50], which is, however, not considered in the process simulators. And
finally, it is clear that the φ-φ-approach takes the excess volume into account in
a formally correct way; but there is hardly any evidence that the excess volume is
represented correctly.
– For the adjustment of the parameters, there is often a disagreement between the
vapor pressure equation used, which is usually well-founded, and the pure com-
ponent vapor pressure given as a data point in a binary vapor-liquid equilibrium
data set. It is a common and successful practice to replace the vapor pressure from
the correlation by the pure component vapor pressure given in the particular data
set just for the parameter fitting procedure to avoid inconsistencies with the rest
of the data and to get more reliable values for the γ∞
i -values (vapor pressure shift-
ing). In the process simulation, the vapor pressure correlation is then used again
together with the adjusted binary parameters [11]. In the γ-φ-formalism, this is an
easy change, whereas in the φ-φ-approach the individual α-function would have
to be manipulated, which has also an influence on the other quantities.
– It is a considerable disadvantage for the project administration that the g E mixing
rules use the equations for the determination of the activity coefficients in a differ-
ent context. The same parameters do not yield the same activity coefficients. This
leads to confusion when both approaches are used in a project simultaneously.
2.8 Enthalpy calculations
For the evaluation of heating and cooling duties in a process, a correct description of
the specific enthalpies is decisive. As all components are more or less present in both
the liquid and the vapor phase, the difficulty is that a continuous enthalpy description
in both phases is necessary. The following quantities can contribute to the enthalpy:
f
– standard enthalpy of formation ∆h0 at t = 25 °C in the ideal gas state, used as
reference point for the specific enthalpy;
– ideal gas heat capacity cid p;
– enthalpy of vaporization ∆h v ;
– liquid heat capacity cLp ;
– enthalpy pressure correction of the vapor phase (h − hid );
– excess enthalpy h E .
First, it should be remembered that the absolute value of the enthalpy is normally
meaningless, only differences between specific enthalpies can be interpreted. A sin-
gle value for the enthalpy is only useful if a reference point is given. With an arbitrary
choice of the reference point (e.g. h = 0 for t = 0 °C), the calculation of chemical reac-
tions is awkward; it makes only sense if only pure components are involved. In process
simulation programs, the use of the standard enthalpy of formation as reference point
for the enthalpy makes sure that the enthalpies of reaction can be correctly calculated.
This is further explained in Chapter 10.1.
Only the standard enthalpies of formation and the ideal gas heat capacities are
explicitly necessary if equations of state are used, whereas the deviations from the
ideal gas can be calculated directly. Equations of state with generalized α-functions
are usually not accurate enough, therefore, individual α-functions of advanced cubic
equations of state with component-specific parameters have to be fitted to ∆h v , cLp , and
(h − hid ) to ensure that the equation of state works. It turned out that the adjustment
of cLp is sufficient to obtain good results [22].
For the γ-φ-approach, there are more options available, as the particular quan-
tities are not independent of each other. For a pure substance, knowing three of the
four quantities cid p , ∆h v , c p , (h − h ), the fourth one can be evaluated. Unfortunately,
L id
it turns out that this does not work for cid L

p and c p . The reason is that the slope of ∆h v
is not accurate enough if only values of this quantity are adjusted [51]. There are two
ways for the description of the enthalpy in process simulators, illustrated using hT di-
agrams (Figures 2.20, 2.21). For a pure substance, they show the bubble point line, the
dew point line and the ideal gas enthalpy at p = 0 for guidance.
(A) Vapor as the starting phase:
With this route, the cLp is the quantity which is determined indirectly (Figure 2.20).
Liquid enthalpies of the pure components are determined with considerable de-
viations via
T
h L,i (T) = ∆h0f,i + ∫ cid

p,i dT + (h − h )i (T, p si ) − ∆h vi (T) .
id
(2.83)
T0
–8600
T
–8700 ∫c id
p,i dT
Ideal gas enthalpy Δh0f
T0 (h − hid)(T, psi(T ))
–8800
Enthalpy of Δhvi Critical point
h/J/g
–8900 saturated vapor

Enthalpy of
–9000 saturated liquid
–9100
T0
–9200
–100 –75 –50 –25 0 25 50 75 100 125 150 175
t/°C
Fig. 2.20: Enthalpy description of a pure liquid using the vapor as the starting phase [11].
A way to overcome this difficulty is to fit the correlations for the enthalpy of va-
porization simultaneously to both enthalpy of vaporization and cLp data [51]. The
effect of pressure on the liquid enthalpy is considered to be small and therefore
neglected.
–2200
T
∫c id
p,i dT
T0
(h − hid)(T, ps (T ))
–2400
(h − hid)(Ttrans, ps(Ttrans))
h/J/g
∫ c dT
L
p Δhvi (Ttrans)
–2600 T0
href (Tref) h, liquid

h, saturated
vapor
–2800 h, ideal gas
269 299 328 358
T/K
Fig. 2.21: Calculation of the enthalpy of saturated vapor of a pure substance using the liquid as the
starting phase [11].
(B) Liquid as the starting phase:

The reference enthalpy is set for a liquid state and adjusted in a way that the stan-
dard enthalpy of formation in the ideal gas state at t = 25 °C is met. The transition
to the vapor phase is performed at a certain arbitrary temperature, usually the
normal boiling point. Figure 2.21 illustrates the calculation route for an enthalpy
of a saturated vapor using route (B). The enthalpy of vaporization is calculated
indirectly with this route, the results are usually sufficiently accurate at low tem-
peratures. They become even qualitatively wrong in the vicinity of the critical
point [11]. Despite the fact that it is calculated directly, it can turn out that cLp
is again the problem quantity. It is often only measured for temperatures below
the normal boiling point, and its extrapolation to high temperatures is often bad.
For both routes, the change from pure components to the mixture is performed at the
system temperature. For gases, the mixing takes place in the ideal gas state, where the
excess enthalpy is zero. Liquid enthalpies are linearly mixed, the excess enthalpy is
often neglected.
Correlations for cid L
p , ∆h v , and c p supported by the various process simulation pro-
grams are discussed in [11]. The most important ones are as follows:
– for cidp:
– Aly-Lee equation:
2 2
C/T E/T
p = A+B(
cid ) +D( ) . (2.84)
sinh (C/T) cosh (E/T)
– PPDS equation:
p = RB + R(C − B)y [1 + (y − 1)(D + Ey + Fy + Gy + Hy )] ,

cid 2 2 3 4
(2.85)
with y = T/(A + T).

– polynomial:
p = A + BT + CT + DT .
cid 2 3
(2.86)
– for cLp :
– polynomial
cLp = A + BT + CT 2 + DT 3 + ET 4 + FT 5 (2.87)
– PPDS equation:
A
cLp = R ( + B + C(1 − Tr ) + D(1 − Tr )2 + E(1 − Tr )3 + F(1 − Tr )4 )
1 − Tr
(2.88)
– for ∆h v :
– DIPPR¹⁰ equation
∆h v = A(1 − Tr )B+CTr +DTr +ETr
2 3
(2.89)
– PPDS equation:
∆h v = RTc (A(1 − Tr )1/3 + B(1 − Tr )2/3 + C(1 − Tr ) + D(1 − Tr )2 + E(1 − Tr )6 )

(2.90)
In the particular equations, A, B, . . . , F are the adjustable parameters. R is the general

gas constant.
It should be mentioned that most data for ∆h v are based on the Clausius–Clapey-
ron equation (2.7). Using a good vapor pressure equation and an appropriate equation
of state for v󸀠󸀠 , an error of approximately 2 % can be expected. The application of the
Clausius–Clapeyron equation should be avoided at vapor pressures p s < 1 mbar and
in the vicinity of the critical point, as the description of v󸀠󸀠 becomes more and more
weak in this area [11].
2.9 Model choice and data management
Any problem, no matter how complicated it may be, has a simple, obvious and generally com-
prehensible wrong solution.
(Harald Lesch)
10 Design Institute for Physical Property Data

Process simulators offer a vast amount of models, which often confuses the user, gen-
erating more disorientation than opportunities. The following paragraph should pro-
vide some clarification. Despite the fact that new ideas and solutions are frequently
presented in the literature, the number of models in use can in principle be restricted
to six. These models are the following.
– Standard activity coefficient model combined with an equation of state, e.g. NRTL-
PR:
A standard activity coefficient model should cover approx. 80–90 % of the cases.
An appropriate equation of state should be used for the vapor phase to make full
use of the capabilities of Equation (2.42). This equation can be used in principle
for temperatures up to the lowest critical point. The Peng–Robinson equation is
the favorite one of the author, definitely, there are other options. If single critical
points are by far exceeded (e.g. presence of an inert gas), the Henry concept can
be applied (Equation (2.59)). When they are exceeded only by a small extent, it
is theoretically wrong but pragmatic just to extrapolate the vapor pressure curve,
however, this should not be done for main components. It does not make sense
in process engineering to use the ideal gas equation for the vapor phase, even if
pressures are low. With the ideal gas equation, the enthalpy description is inaccu-
rate (Chapter 2.8). Moreover, in a later state of a project the pressure relief devices
must be designed (Chapter 14.2). In this case, the thermodynamic model is applied
at pressures far above the normal operation pressure, which then requires a rea-
sonably accurate equation of state for the vapor phase anyway. For the density, it
should be taken care that the mixing rule (Equation (2.66)) is applied. As well, the
correct representation of the enthalpies should be considered.
– Standard activity coefficient model combined with an equation of state for vapor
phase association, e.g. NRTL-ASS:
If carboxylic acids (formic, acetic, propionic, acrylic, butanoic, etc.) or HF play a
decisive role in the process, the use of an association model for the vapor phase
is important (Chapter 2.3.3). One should be aware that there are no models which
consider both association and the nonidealities caused by elevated pressures. The
problems concerning cLp via enthalpy route (A) should be considered [51].
– Electrolyte model, e.g. ELECNRTL:
Small amounts of electrolytes can be treated as heavy ends as a workaround.
If they play a major role, an electrolyte model considering the dissociation re-
actions and the interactions of the ions should be used [47, 48]. The electrolyte
model can as well be combined with equations of state for the vapor phase, e.g.
Peng–Robinson or Redlich–Kwong for high pressures or the association model in
case HF occurs.
– Equation of state model (PR, PSRK, VTPR):
If the process is mainly operated at high pressures, if supercritical components
play a major role with high concentrations in the liquid phase so that Henry’s
law cannot be successfully applied or if supercritical components change to the
subcritical state in one apparatus, the application of the φ-φ-approach with an

equation of state valid for both the vapor and the liquid phase is obligatory (Chap-
ter 2.2). Well-known examples are the polypropylene or natural gas conversion
processes. The standard generalized equations of state like Peng–Robinson [21],
Soave–Redlich–Kwong [52] or Lee–Kesler–Plöcker [53] have the disadvantage
that they perform well for nonpolar molecules (ethane, propane, propylene,
etc.), but the more the molecules have a polar character the more inaccurate they
become. A solution for this problem are the advanced cubic equations of state
like PSRK or VTPR [11]. They involve pure component parameters in the so-called
α-function (Equations (2.33), (2.34)). With this option, the vapor pressures, heat
capacities and enthalpies of vaporization of any components can be successfully
represented. There are still weaknesses in the liquid density description, even the
volume translation in the VTPR equation does not give satisfactory results.
The mixing rules of the advanced cubic equations of state are based on the ac-
tivity coefficient approaches (g E mixing rules). Combination of these advanced
equations of state with electrolyte models are available as well, however, as men-
tioned above, they can still not be combined with vapor phase association models.
– High-precision equation of state:
Unexpectedly, there are a lot of problems in process simulation which are related
to pure components, e.g. the utilities steam and cooling water, inertizations with
nitrogen or CO2 or major parts of the LDPE¹¹ process, where ethylene is the only
component. These problems can be covered very effectively by the use of the so-
called high-precision equations of state, which are equations adjusted individ-
ually for the particular components [26–28]. They represent all thermodynamic
properties within their experimental uncertainty over a wide range of pressures
and temperatures and should be used whenever possible. However, except for
natural gas no high-precision equations of state for mixtures are available. In
these cases, compromises have to be made¹². In the hydrocarbon processes, the
Lee–Kesler–Plöcker equation is a good approach, using accurate pure component
representations for hydrocarbons and applying corresponding mixing rules. For
polar components, this is not an option.
– Polymer model:
Polymer applications are often covered by representing the polymer as a high-
boiling component. A reasonable representation of the combinatorial part of
the activity coefficient, which often shows negative deviations from Raoult’s
law for molecules differing large in size, should be considered (Flory–Huggins,
UNIQUAC). There are models for polymers which are by far more complicated
11 low density polyethylene.

12 Copolymerization in LDPE is a good example. For instance, it has to be decided whether for a mix-
ture of 95 % ethylene and 5 % propylene at p = 100 bar it is more important to describe the pressure
effect sufficiently or to describe the deviation caused by the 5 % impurity.
but represent the character and the effects in polymer mixtures accurately, e.g.
PC-SAFT¹³ [54].
The choice of a good model does not imply that simulation will be correct. It is even
more important to compile the binary interaction parameters (BIPs) for the various
possible combinations. For n components in a process, n(n − 1)/2 binary combina-
tions are possible. For a typical project in a chemical plant with 40 components this
makes 780 binary parameter sets. In hydrocarbon business, up to 200 components are
possible, giving 19 900 binary parameters sets. In both cases it is not possible to pro-
vide a perfect matrix in a reasonable time scale. For a chemical process, the following
procedure is recommended:
An EXCEL table is set up containing a matrix with all the components, showing a
color code for the possible options. Figure 2.22 gives an example.
Component H2 H2O BDO THF PAC PROH BUT-AC BUOH

H2
H2O
BDO
THF
PROP-AC
PROPANOL
BUT-AC
BUTANOL
not necessary just set experimental data

Simulator Database to be revised estimated
Fig. 2.22: Illustration of the binary interaction parameter matrix in a project.
To fill this matrix, first the BIPs given by the databanks of the process simulator should
be listed. Next, it should be checked which parameters are decisive for the process.
This is, of course, not an exact assessment, nevertheless, for components which oc-
cur together in any block with considerable concentrations, the corresponding BIP
should be important. For these cases, it is the responsibility of the user either to adopt
the parameters from the simulator or to adjust own ones from experimental data from
the established databanks [55, 56]. The latter is highly recommended, as the quality
of the parameters can be assured in this way. If the data situation turns out to be not
13 perturbated-chain statistical associating fluid theory.

sufficient, own phase equilibrium measurements can be initiated to overcome this sit-
uation. The adjustment of BIPs to experimental data is thoroughly described in [11].
Less important parameters can be estimated (Chapter 2.10). Parameters where the
current situation is not acceptable should be marked as “to be revised” or similar. It
must be clear that the BIPs for component pairs where both components occur only at
very low concentrations, e.g. in the ppm region, are not important, it has no influence
on the results if they are omitted. All BIPs with water should be assigned, either by
the simulation program, by adjustment to experimental data or by estimation, as the
nonidealities of water with organic components are usually large. Other parameters
can just be set; e.g., the BIPs for n-hexane–n-heptane can be assumed to be zero (ideal
mixture), as long as nothing better is available. In this way, a detailed overview about
the BIP situation can be obtained, and the quality of the simulation results is easier to
assess.
For the hydrocarbon business, the situation is different. The number of param-
eters involved is so large that a thorough check is not possible within a reasonable
time. Often, it cannot be distinguished between important and less important compo-
nents, as all components occur only in relatively low concentrations. As hydrocarbon
mixtures usually do not show major nonidealities and the interactions with water can
easily be covered, it might be an option to use Modified UNIFAC with an appropriate
equation of state or PSRK (Chapter 2.10).
In this case, process simulation can be skipped!

(Hans Haverkamp, after his boss suggested that he should vary the binary parameters until the
column behavior is met)
Fitting physical properties to reproduce operation results is something one should not
do. Process simulation which meets the data obtained from operation are a strong in-
dication that the process is understood, or it can help to detect errors. Fitting physical
properties to operation data will certainly reproduce the data, but the simulation is of
no use. Extrapolation to other operation states will simply not work.
2.10 Binary parameter estimation
BIPs can be estimated using group contribution methods, as illustrated in Figure 2.23
for the system ethanol–n-hexane. In group contribution methods it is assumed that
the mixture does not consist of molecules but of functional groups. Ethanol can be
divided in a CH3 -, a CH2 -, and an OH-group, whereas n-hexane consists of two CH3 -
and four CH2 -groups. It can be shown that the required activity coefficients can be
calculated as long as the interaction parameters between the functional groups are
known. Furthermore, if the group interaction parameters between the alkane and the
alcohol group are known, not only the system ethanol–n-hexane, but also all other
2.10 Binary parameter estimation | 73
–OH –CH3
–CH2 –CH2 –CH2
–CH2 –OH
Fig. 2.23: Illustration of the group contribution

Ethanol: –CH3 –CH2 –OH
concept [11].
n-Hexan: –CH3 –CH2 –CH2 –CH2 –CH2 –CH3
Reproduced with permission.
alkane-alcohol or alcohol-alcohol systems can be predicted. The great advantage of

group contribution methods is that the number of functional groups is much smaller
than the number of possible molecules [11].
The UNIFAC (universal quasi-chemical functional group activity coefficients)
group contribution method has first been published in 1975 [57]. Like UNIQUAC,
it consists of two parts. The combinatorial part is temperature-independent and
takes into account the size and form of the molecules, whereas the residual part
is temperature-dependent and considers attractive and repulsive forces between the
groups.
The interacting groups are called main groups. They often consist of more than
one sub-group. For example, in the case of alkanes the subgroups are CH3 -, CH2 -,
CH-, and C-groups. The different sub-groups have different size parameters, the so-
called van-der-Waals properties, which represent the volume and the surface of the
groups. By definition, the group interaction parameters between groups belonging to
the same main group are equal to zero.
Meanwhile, the UNIFAC method has almost been replaced by the Modified
UNIFAC (Dortmund) method [58, 59]. Its main improvements are [11]:
– an empirically modified combinatorial part to improve the results for asymmetric
systems;
– temperature-dependent group interaction parameters;
– adjustment of the van-der-Waals properties;
– additional main groups, e.g. for cyclic alkanes or formic acid;
– an extension of the database, besides VLE data;
– activity coefficients at infinite dilution;
– excess enthalpy data;
– excess heat capacity data;
– liquid-liquid equilibrium data;
– solid-liquid equilibrium data of simple eutectic systems;
– azeotropic data.
Most important for the application of group contribution methods for the synthesis
and design of separation processes is a comprehensive and reliable parameter matrix.
Because of the importance of Modified UNIFAC for process development, the range of
applicability is continuously extended by filling the gaps in the parameter table and
revising the existing parameters with the help of new data. Since 1996, the further re-
vision and extension of the parameter matrix is carried out by the UNIFAC consortium,
and the revised parameter matrix is only available for its sponsors [60].
For the estimation of binary parameters, Mod. UNIFAC is used in a way that artifi-
cial data are generated, and the parameters of the current model are adjusted to them.
The advantage to the direct use as a model is that the well-known systems do not need
to be estimated with Mod. UNIFAC, and their accuracy is not lost.
Still, Mod. UNIFAC has some weaknesses:
– Isomer effects can not be predicted.
– Unreliable results are obtained for group contribution methods if a large number
of functional groups occurs in the molecule, as it is the case for pharmaceuticals.
Functional groups which are located closely together are often not represented
sufficiently, e.g. the configuration –C(Cl)(F)(Br) in refrigerants (proximity effect).
– Poor results are obtained for the solubilities and activity coefficients at infinite
dilution of alkanes or naphthenes in water [11].
– Systems with small deviations from Raoult’s law are difficult to predict, which is a
problem if also the differences in vapor pressures are small. In these cases, often
qualitatively wrong characteristics for the mixture are obtained.
The advanced cubic equations of state PSRK and VTPR use a g E mixing rule, which can
be used as a predictive tool if UNIFAC or Modified UNIFAC are used for the calculation
of g E . For PSRK, the original UNIFAC method can be used, whereas Modified UNIFAC
yields bad results in this combination. For VTPR, an own matrix has been built up [60,
61]. For both equations, additional groups have been introduced so that light gases can
be described as well. Certainly, as equations of state both can of course be used in the
supercritical region.
In recent years, the COSMO-RS model¹⁴ has been developed [62], which works
without adjusted parameters. Its accuracy compared to UNIFAC is a bit worse, but it
is applicable in any case. However, it should be used only by a professional user, as
well as Molecular Modeling. Molecular Modeling can generate data from quantum-
mechanical calculations of the interactions of the molecules. This is a fascinating
demonstration of how far these interactions are understood. However, in engineer-
ing applications more detailed justifications of the model are required, therefore, the
author is pretty convinced that the general structure of fitting models to experimental
data will remain.
14 COnductor like Screening MOdel for Real Solvents.

2.11 Model changes
For the handling of enthalpies in a process simulation program, the change of a model
between two blocks is often critical. This problem has much to do with the enthalpy
description. Between the two blocks, the simulation program hands over the values for
p and h to describe the state of the stream. According to the particular models used in
the two blocks, the stream is assigned with two different temperatures that may differ
significantly [11].
Example
A vapor stream consisting of pure n-hexane (t1 = 100 °C, p1 = 2 bar) is coming out of a block which
uses the Peng–Robinson equation with the φ-φ-approach. It is transferred to a “nothing-happens-
block” (adiabatic, same pressure) working with the ideal gas law. Which error will be produced due to
the model change?
Solution
According to the Peng–Robinson equation of state, the specific enthalpy of the stream leaving the first
block is determined to be h1 = −1807.3 J/g at p1 = 2 bar. Using an activity coefficient model with an
ideal gas phase, the coordinates for p2 = p1 and h2 = h1 refer to the vapor state at t2 = 96.6 °C. Using
cLp ≈ 2 J/(g K), the corresponding enthalpy difference
J
∆h ≈ cLp ∆T = 2 ⋅ 3.4 K = 6.8 J/g
gK
will be missing in the energy balance.
Therefore, care must be taken when the thermodynamic model is changed in a flow-
sheet. It is recommended that a dummy heat exchanger is introduced between two
blocks operating with different models, which is defined in a way that inlet and outlet
states are the same in spite of the different models. Another option is to carry out a
model change where both models yield at least similar results, maybe in a block oper-
ating at low pressure. In general, one should stick to the most comprehensive model
in a flowsheet; nevertheless there are cases (e.g. association, electrolytes occurring
only in parts of the flowsheet) where a model change cannot be avoided.
2.12 Transport properties
The transport properties dynamic viscosity, thermal conductivity and surface tension
must be known for the design of many pieces of equipment, e.g. columns or heat
exchangers. Their correlations and mixing rules and further details are thoroughly
explained in [11] or [63]. Therefore, only general remarks are given here and the pit-
falls are explained.
– Dynamic viscosity of liquids:
The dynamic viscosity of liquids is probably the most important transport quan-
tity. Among others, it has an influence on heat exchangers, distillation and ex-
traction columns, and the pressure drop of pipes. Similar to the vapor pressure,
it can cover several orders of magnitude and it is similarly difficult to correlate
and extrapolate, although nowadays there are excellent correlation tools which
do this easily. The dynamic viscosity of liquids starts at high values at the melt-
ing point and then decreases logarithmically with increasing temperature. The
pressure influence is comparably small, but it should be considered at large pres-
sures p > 50–100 bar. Correlations refer to the dynamic viscosity at saturation
pressure. Correction terms for the pressure influence are available [11]. Also, just
recently the first reasonable estimation method has been developed [64]. How-
ever, the most crucial thing and probably the weakest part of process simulation
in general is the mixing rule. Several ones are available, but in most cases the
logarithmic relationship
ηM ηi
ln = ∑ x i ln (2.91)
η0 i
η 0
is used, where η0 is the unit used to make the argument of the logarithm dimen-
sionless. Equation (2.91) can only reproduce the order of magnitude of the result,
and in extreme cases not even that. It does not claim to give a good result, it is
essentially not more than a way to come to a number. Even simple systems like
methanol-water can exhibit large maxima, where the deviation of Equation (2.91)
can be up to 100 % [11]. The lucky circumstance is that most applications are not
strongly sensitive to errors in the calculation of small viscosities around or even
below 1 mPa s. Things become worse when larger viscosities become involved and
differences between the two components occur.
Example
Calculate the dynamic viscosity of a brine consisting of 40 mol% ethylene glycol (1,2-ethanediol, EG,
M = 62.068 g/mol) and 60 mol% water (W, M = 18.015 g/mol) at t = 20 °C.
Solution
The pure component viscosities are according to [42]
ηW = 1.01 mPa s ,
ηEG = 21.23 mPa s ,
giving a calculated viscosity for the mixture according to Equation (2.91):

ηM
ln = 0.6 ⋅ ln 1.01 + 0.4 ⋅ ln 21.23 = 1.228 ⇒ η M = 3.415 mPa s .
mPa s
A more probable value can be taken from [65] after recalculation of the EG concentration from 40 mol%
to 69.7 mass%. The result is 7.08 mPa s, with a deviation of more than 100 % from the calculated value.
In fact, this is a deviation which might have a significant influence on equipment design.
– Thermal conductivity of liquids:

The thermal conductivity of liquids decreases with increasing temperature almost
linearly, except the region in the vicinity of the critical point. The order of mag-
nitude is approx. 0.1–0.2 W/(K m) for almost all liquids. An exception is water;
here the thermal conductivity is in the range 0.6–0.7 W/(K m), with a maximum
at approx. 150 °C. There is a mixing rule which does not produce major errors. The
influence of the pressure is similar to the one for the liquid viscosity; the values
usually refer to the saturation line, and it is a good approach to take it for pressure-
independent, but at p > 50–100 bar a correction term should be applied [11, 63].
– Dynamic viscosity of gases:
In contrast to the viscosity of liquids, the dynamic viscosity of vapors or gases
increases with increasing temperature almost linearly. The order of magnitude
is approx. 5–20 µPa s. The thermal conductivity of gases is hardly ever measured,
and most of the values are calculated ones from a well-defined estimation method
[11]. As for liquids, a pressure-correction should be applied when the pressure
exceeds p > 50–100 bar [11]. Given values usually refer to the ideal gas state at
low pressures. Mixing rules are available.
– Thermal conductivity of gases:
The qualitative behavior of the thermal conductivity of gases is similar to the
dynamic viscosity of gases. The typical order of magnitude is 0.01–0.03 W/(K m).
As well, for the pressure dependence the same statements hold. However, some
interesting remarks must be given: The thermal conductivity of hydrogen and
helium, the so-called quantum gases, is significantly higher; it is in the range
0.16–0.25 W/(K m) in the usual temperature range 0–250 °C for both substances.
This is the reason why helium or hydrogen are used as carrier gases in gas chro-
matography with thermal conductivity detectors. The high thermal conductivity
corresponds to the base line; when the sample passes the detector, the thermal
conductivity and therefore the heat removal from the detector decreases. The
detector temperature rises and gives a corresponding signal.
A second interesting item is that for very low pressures (p < 10−6 bar), the thermal
conductivity of gases is no more a physical property of the substance but depends
more on the dimensions of the vessel where the gas is located [11].
– Surface tension:
The surface tension occurs in various formulas used for the design of equipment
in process engineering, but the author is not aware that any of these equations are
sensitive to it. The surface tension refers to the phase equilibrium. A typical order
of magnitude is 5–20 mN/m. For water, the surface tension is significantly higher
(approx. 75 mN/m at room temperature). The surface tension decreases almost lin-
early with increasing temperature and becomes zero at the critical point. Mixing
rules are available, but for systems with water special ones must be applied.
3 Working on a process
The difference between fiction and simulation is that both simulation and fiction deceive and
betray, but at least simulation creates an image that is congruent with reality.
([66])
A process simulation is the attempt to evaluate the characteristic quantities of a pro-

cess with well-defined calculations of the particular process steps. It is also a target to
identify the sensitivities of a process and to find out how it reacts to disturbances. Pro-
cess simulation is a tool for the development, design and optimization of processes in
the chemical, petrochemical, pharmaceutical, energy producing, gas processing, en-
vironmental, and food industry [11]. It provides a representation of the particular basic
operations of the process using mathematical models for the different unit operations,
ensuring that the mass and energy balances are maintained. Today simulation models
are of extraordinary importance for scientific and technical developments. The devel-
opment of process simulation started in the 1960s, when appropriate hardware and
software became available and could connect the remarkable knowledge about ther-
mophysical properties, phase equilibria, reaction equilibria, reaction kinetics, and the
particular unit operations. A number of comprehensive simulation programs have
been developed, commercial ones (ASPEN Plus, ChemCAD, HySys, Pro/II, ProSim)
as well as in-house simulators in large companies, for example, VTPlan (Bayer AG)
or ChemaSim (BASF), not to mention the large number of in-house tools that cover
the particular calculation tasks of small companies working in process engineering.
Nevertheless, all the simulators have in common that they are only as good as the
models and the corresponding model parameters available.
Various degrees of effort can be applied in process simulation. A simple split bal-
ance can give a first overview of the process without introducing any physical rela-
tionships into the calculation. The user just defines split factors to decide which way
the particular components take. In a medium level of complexity, shortcut methods
are used to characterize the various process operations. The rigorous simulation with
its full complexity can be considered as the most common case. The particular unit
operations (reactors, columns, heat exchangers, flash vessels, compressors, valves,
pumps, etc.) are represented with their correct physical background and with a model
for the thermophysical properties. Different physical modes are sometimes available
for the same unit operation. A distillation column can, for example, be modeled on the
basis of theoretical stages or using a rate-based model, taking into account the mass
transfer on the column internals. A simulation of this kind can be used to extract the
data for the design of the process equipment or to optimize the process itself. During
recent years, dynamic simulation has become more and more important. In this con-
text, “dynamic” means that the particular input data can be varied with time so that
the time-dependent behavior of the plant can be modeled and the efficiency of the
process control can be evaluated. For both steady-state and dynamic simulation, the
80 | 3 Working on a process
correct representation of the thermodynamics, i.e. thermophysical properties, phase

equilibria, mass transfer, and chemical reactions mainly determines the quality of the
simulation. However, one must be aware that there are a lot of pitfalls beyond thermo-
dynamics. Unknown components, foam formation, slow mass transfer, fouling layers,
decomposition, or side reactions might lead to unrealistic results. The occurrence of
solids in general is always a challenge, where only small scale-ups are possible (ap-
prox. 1 : 10) in contrast to fluid processes, where a scale-up of 1 : 1000 is nothing un-
usual. For crystallization, the kinetics of crystal growth is often more important than
the phase equilibrium itself. Nevertheless, even under these conditions simulation
can yield a valuable contribution for understanding the principles of a process.
Today process simulations are the basis for the design of plants and the evalu-
ation of investment and operation costs, as well as for follow-up tasks like process
safety analysis, emission lists, or performance evaluation. For process development
and optimization purposes, they can effectively be used to compare various options
and select the most promising one, which, however, should in general be verified ex-
perimentally. Therefore, a state-of-the-art process simulation can make a considerable
contribution for both plant contractors and operating companies in reducing costs.
3.1 Flowsheet setup
Sometimes, the reality is different from the truth …

(Comparison between plant and simulation data)
Figure 3.1 shows the symbols for some of the most important blocks used in the process
simulator ASPEN Plus. Their main functions can be briefly explained as follows.
1. Heater block:
In a heater block, a process stream is heated up or cooled down. It is not regarded
how this is achieved. Usually, it is done with utility (e.g. steam, cooling water,
brine), but it can also happen that the heat is exchanged between two process
streams. When using the heater block, one should be aware that the simulator
does not check whether the driving temperature differences are sufficient to trans-
fer the heat. It should also be clear that the use of the heater block is not connected
to a special kind of equipment.
2. Heat exchanger block:
In contrast to the heater block, the heat exchanger calculates the heat transfer
between two media, and it is checked whether this heat transfer can take place
or not, if a minimum driving temperature difference is not available. Clearly, this
is the better way to model a heat transfer, but one should be aware that only the
target conditions of one stream can be specified. The second stream is a result of
the specification, and this can lead to an increased effort for convergence.
3. Valve:
Performs an adiabatic throttling with hin = hout .
4. Pipe:
Evaluates the pressure drop in a specified pipe, considering friction losses, geode-
tic height influence and pressure changes due to positive or negative acceleration.
Also, the two-phase pressure drop (Figure 12.7) can be considered, and the pipe
can be divided into segments to update the vapor fraction continuously. Often,
the most significant influence is the one of the geodetic height. The friction pres-
sure drop can often be neglected, as later the pipe diameter will be dimensioned
in a reasonable way during the basic engineering anyway. The height influence
is independent of the pipe diameter. If one wants to consider the height without
already fixing a friction pressure drop, the pipe diameter can be arbitrarily set to
d = 2 m. Such a large diameter ensures that the velocities are so low that the fric-
tion pressure drop will be negligible, but it will develop the full height influence.
5. Pump:
Elevates the pressure of liquids. In process simulation, the pump symbol and their
function are more or less place holders for the pumps in the process. Only for big
pumps the power consumption is relevant, e.g. cooling water pumps. The author’s
opinion is that the main function of the pumps in a simulation flowsheet is to
make the engineer remember that a device for the pressure elevation is necessary.
Formally, the pressure elevation and the efficiency of the pump can be specified.
However, during process simulation the layout is not available so that the exact
tasks of the pumps cannot be defined, and their efficiency is strongly dependent
on the type and on the conditions chosen where the pumps must have its optimum
efficiency (Chapter 8.1). This is not really a problem for process simulation, as the
change of temperature and enthalpy due to the pressure elevation of a liquid is
usually small. In a later stage of the project, usually only the big pumps with large
volume flows or pressure elevations are updated.
6. Compressor:
Elevates the pressure of gases. In contrast to pumps the power required for com-
pressors is significant, and the temperature elevation is important (Chapter 8.2).
7. Vessel:
A vessel just for the intermediate storage of liquids is often omitted in process
simulation, as it does not change the thermodynamic state. If the vessel has a
process function (e.g. separation of vapor and liquid, heating the liquid by jacket
heating, mixing), it is considered.
8. Splitter:
Divides a stream and leads the particular parts to their destinations.
9. Mixer:
Unites various streams and evaluates the state of the resulting stream. The out-
let pressure after the mixing must usually be defined. Normally, it is the lowest
pressure of the participating streams.
10. Separator:
Divides the inlet stream and leads the particular parts to their destinations. In
contrast to splitters, it can be defined for each component which fractions are led
to the particular streams.
11. Multiplier:
Multiplies the mass-flow of an inlet stream with a factor. Of course, this is a block
with a function that will not be invented very soon, but it is useful for the transfer
from batch to continuous operating mode or for dividing the process into lanes
and their reunification.
Their are a number of other blocks (reactors, distillation columns, decanters, absorp-
tion columns, extraction columns), which are discussed in detail in the corresponding
chapters.
There is often a misunderstanding about the meaning of a process simulation. One
must be aware that all blocks described above refer to an equilibrium state, where both
inlet and outlet streams are constant. However, equilibria are only reached if the resi-
dence time is infinite, and this is of course never the case. The design of the equipment
should be performed in a way that the blocks approach this equilibrium in a reason-
able way. For example, for a decanter it is necessary that the residence time is long
enough for the two liquid phases to form and separate. If this time is not provided,
the decanter will perform worse than calculated, which often leads to the conclusion
that the equilibrium is not described correctly. The same holds for many other blocks.
Sometimes, an efficiency can be introduced, as it is the case in columns, or kinetic ap-
proaches can be applied, as in reactors. The knowledge about the process steps and, of
course, the choice of an appropriate thermodynamic model are necessary presump-
tions for a useful application of process simulation. Otherwise, the GIGO principle
holds, i.e.:
Garbage in, garbage out!

(The GIGO principle)
Figure 3.2 shows the typical scheme of a chemical plant. One or more feed streams
containing the raw materials are passing a preparation step, e.g. a purification, com-
pression or a heating step. Then, the main reaction step takes place, giving the main
products and a number of by- or coproducts¹. In a separation section, the valuable
products are isolated and purified to meet the specification. Nonvaluable by-products
are separated and sent to disposal. In most cases the raw materials will not be con-
verted to a full extent. Due to economic reasons, it is of course desirable to collect and
1 A coproduct is a product which necessarily occurs when the main reaction takes place. A by-product
is a product which occurs due to undesired side reactions. In some cases, co- or even by-products are
valuable and can be sold, but usually, they are subject to disposal.
1 Heater 7 Vessel
2 Heat exchanger 8 Splitter
3 Valve 9 Mixer
4 Pipe
5 Pump 10 Separator
6 Compressor 11 Multiplier
Fig. 3.1: Some blocks in ASPEN Plus. Screen images of Aspen Plus® are reprinted with permission by
Aspen Technology, Inc. AspenTech® , aspenONE® , Aspen Plus® , and the Aspen leaf logo are trade-
marks of Aspen Technology, Inc. All rights reserved.
recycle them back so that they are not lost. When it is tried to calculate the process
steps in Figure 3.2 sequentially, two major difficulties come to the fore:
1. The recycle stream cannot be known in advance. It is itself a result of the process
calculation. An iterative procedure is necessary, where the recycle stream is first
estimated. Then, the process calculation is carried out, giving the recycle stream
as a result. If estimated and calculated recycle stream are identical within a cer-
tain tolerance, the result can be accepted. Otherwise, it must be estimated again
and the procedure must be repeated. In this case, the recycle stream is called a
tear stream. This kind of task is a special ability of the above-mentioned process
simulators. How they work is explained below.
2. The other remarkable point is that the recycle stream might contain other com-
ponents than the ones listed in Figure 3.2. Each component occurring in the pro-
cess must have an outlet; otherwise it will accumulate in the process. If a compo-
nent behaves in a way that it is neither concentrated in the product or by-product
stream nor removed in a side reaction in the reactor, the only way to get rid of it is a
purge stream, where a defined (and hopefully small) amount of substance is split
from the recycle stream (or another appropriate one) and led out of the process.
In this way, the concentration of the accumulating components will rise up to a
certain level, where the removal of the components on one side and the formation
and feed on the other side are in equilibrium.
Recycle of A and B Purge
Feed: Products:
A+B C+D
Preparation Reaction Separation
By-products:
A+B C+D E+F
Fig. 3.2: Typical scheme of a chemical plant [11].

There are several strategies to achieve conversion of a flowsheet, i.e. to get a result
where the estimated compositions and states of all tear streams agree with the calcu-
lated ones within a certain tolerance. For this purpose, the simulation programs use
specialized methods. The simplest one is the direct substitution method, where the
new estimation for the tear stream X is the calculated stream G(X) from the previous
flowsheet calculation path:
XDirect,k+1 = G(X k ) . (3.1)
The direct substitution method is slow but sure; it gives results even in cases where
other methods are unstable. With few exceptions, the convergence course indicates
that the error steadily decreases from step to step. If this is not the case, it makes sense
to interrupt the flowsheet convergence and have a look at the tear stream history in
order to see whether a component accumulates. In case the iteration switches between
two solutions, it makes sense to give another starting point or change the convergence
method.
The standard method for flowsheet convergence is the Wegstein method. In prin-
ciple, it is an extrapolation of the direct substitution, taking into account the last two
iteration steps for the next estimation of the tear stream. It is usually faster than Di-
rect Substitution, but it can happen that the calculated error can also increase from
one step to the next, and only every few steps an improvement is achieved. In case
convergence takes many steps, it is more difficult to judge whether the convergence
target will be met. The calculation is as follows. First, the last two evaluation steps are
characterized by
G(X k+1 ) − G(X k )
s= . (3.2)
X k+1 − X k
Extrapolation gives an estimate for the next iteration:
G(X k+2 ) = G(X k+1 ) + s(X k+2 − X k+1 ) . (3.3)
At convergence, X k+2 = G(X k+2 ) is expected; therefore, we get
X k+2 = G(X k+1 ) + s(X k+2 − X k+1 ) . (3.4)
Substituting q = s/(s − 1), the final Wegstein iteration formula is obtained:
X k+2 = G(X k+1 ) ⋅ (1 − q) + X k+1 ⋅ q . (3.5)
The Newton method is the fastest approach for achieving convergence in a flowsheet.
It is the transfer of the well-known solving method for nonlinear equations, where the
derivative is used to get the next iteration step:
f(x)
x k+1 = x k − . (3.6)
df(x)/dx
For multivariable functions, instead of the derivative the Jacobian matrix is used:
X k+1 = X k − J(X k )−1 G(X k ) , (3.7)
with
∂G(X i )
J ij (X k ) = . (3.8)
∂X j
For the Newton method, a good starting point is extremely important in order to
achieve convergence. Far away from the solution, the Newton method is not really
a good choice. The evaluation of the Jacobian matrix is a huge effort. Therefore, as
long as progress in the iteration is made, the calculation of derivatives is avoided. The
number of components should not be too large.
The Broyden method is a useful modification. The Jacobian matrix is only calcu-
lated at the first iteration. Therefore, it is faster but maybe less reliable than the Newton
method.
Although convergence is often the most difficult part of working on a flowsheet,
it is strongly recommended to work it out. It is the only way known by the author to
find out whether one of the components accumulates in the process. The following
example explains that it is necessary even in cases assumed to be self-evident.
Case study
For the separation of an ammonia/water mixture (1000 kg/h ammonia, 1000 kg/h water), a two-
column system is provided (Figure 3.3). The target is to get both ammonia and water with a high purity
so that the ammonia can be reused and the water can be disposed as waste water. In column C1,
ammonia is taken overhead in a way that practically all the water goes to the bottom. This means that
some of the ammonia is lost and remains in the water at the bottom. In column C2, the remaining
ammonia is taken overhead with a certain amount of water. The water withdrawn at the bottom is
practically ammonia-free. The overhead stream of column 2 is recycled to column 1. As a result, both
the ammonia and the water outlet can be purified to any necessary extent. The question is now: Can
one imagine that this arrangement works terribly in practice?
P1
REC-1
RECYCLE
C2
NH3
C1
FEED
WASTE H2O
Fig. 3.3: Ammonia–water separation with two distillation columns.

Screen images of Aspen Plus® are reprinted with permission by Aspen Technology, Inc.
AspenTech® , aspenONE® , Aspen Plus® , and the Aspen leaf logo are trademarks of Aspen Tech-
nology, Inc. All rights reserved.
A practical system never consists of only two components. In this case, which actu-
ally happened, an additional organic component occurred. Consider benzene to be
this component, and its amount in the feed shall be 1 kg/h. The plant manager was
desperate, as it was simply impossible to run this arrangement continuously. A sim-
ilar panic came up when it was first tried to simulate the process and to converge
the flowsheet calculation, there was no way to get a result. Both participants have
clear indications of what went wrong. The plant manager had to empty his plant reg-
ularly, and each time huge amounts of organic substance were found. Analogously,
in simulation the iteration history indicated that benzene was the component which
accumulated, which was the reason for the convergence difficulties.
What happened both in the plant and in simulation was the following: In the
first column C1, the cut was done in a way that relatively large amounts of ammo-
nia remained in the bottom stream. As benzene is by far less volatile than ammonia,
it could not get to the top of the column. This is at first favorable, as the ammonia
product is not contaminated with the organics. Together with the whole water and
the rest of ammonia the organics were transported to column C2. In C2, it was made
sure that no ammonia remains in the waste water by taking part of the water overhead.
Benzene and water form a heterogeneous azeotrop. Regarding the separation cut, this
azeotrope is a light end, and the benzene is completely in the overhead stream. Again,
this is considered to be fine at first, as the organic compound should not occur in the
waste water. The strange result is that for the benzene no way is left to leave the system,
and it accumulates in the cycle until the arrangement ceases to do its job. An exit for
the organic component must be provided; in this case, a liquid-liquid separator was
chosen (Figure 3.4, B1). Still, the benzene accumulates, but at a certain level a phase
split occurs. The organic phase can be removed continuously, while again both the
ammonia and the waste water have no impurities.
P1
REC-1 RECYCLE
C2
NH3
C1
FEED
9
B1
L1 WASTE H2O
PURGE
Fig. 3.4: Ammonia–water separation with two distillation columns and a decanter.
The conclusion from this incident is:
Never forget the toilet!

(Process development wisdom)
The most important thing for the feasibility of a process is the value creation of the
products in comparison with the raw materials. Before a process is going to be devel-
oped or the erection of a plant is decided, it must be clear that the added value of the
products is attractive. This check is not really complicated, the only things one must
know are the stoichiometry and the prices of the substances involved.
Example
EDC (1,2-dichloroethane) is produced via the reaction
C2 H4 + Cl2 󳨀→ C2 H4 Cl2 .
The prices² and the molecular weights for the particular substances are listed in Table 3.1.
Table 3.1: Fictitious prices and molecular weights of ethylene, chlorine, and EDC.
Ethylene: 500 €/t 28.053 g/mol

Chlorine: 50 €/t 70.905 g/mol
EDC: 200 €/t 98.959 g/mol
Can this process be feasible?
Solution
Let us consider that 1 kmol ethylene reacts with 1 kmol chlorine. This means that
28.053 kg C2 H4 + 70.905 kg Cl2 󳨀→ 98.959 kg C2 H4 Cl2 .
The corresponding values are
28.053 kg ⋅ 0.5 €/kg + 70.905 kg ⋅ 0.05 €/kg 󳨀→ 98.959 kg ⋅ 0.2 €/kg
This means that the value of the raw materials is 17.57 €, compared to 19.79 €for the product side. In
fact, there is a value generation in this case with these assumed prices, but it is comparably small.
Taking into account other operation and capital costs, the process might not be feasible.
Although the product has a lower price per kg than one of the raw materials, it is possible that
there is a value generation due to the increase of the molecular weight. Only approx. 30 % by mass of
the EDC molecule come from the ethylene; the rest comes from the inexpensive chlorine.
2 in this example completely fictitious,

The cost structures of the various kinds of chemical production, i.e.

– basic chemicals,
– fine chemicals,
– specialty chemicals,
– pharmaceuticals, and
– polymers.
are rather different. Basic chemicals and polymers are manufactured in large special-
ized plants with extremely large capacities (typically: 300 000 tons per year or even
more). For the economy, the differences between the market prices of products and
raw materials are decisive, as in the example above. The costs for energy and for the
disposal of the waste products and waste water grow steadily and have already a sig-
nificant influence. The large capacities are advantageous, as the invest costs of a plant
do not rise proportionally to the capacity. Also, appropriate logistics for the handling
with the huge amounts of substances is necessary. It is favorable to work at large chem-
ical sites, where all utilities are available and where the products can be further used
directly without a major transport infrastructure. Because of the strong competition,
the margins are low. The profit in basic chemicals production is based on the safe
sales of large amounts which are ensured by long-term contracts. The same holds for
polymers.
In contrast, the manufacturing capacity of pharmaceutical products is by far
lower, often only several hundred kg per year. The costs for raw materials, energy,
and the investment costs for the plant are not really important. However, before a
pharmaceutical product gets permission, a huge research effort is necessary before
the first revenues are obtained, often more than a billion €. This means that the risk
for such a development is high. Furthermore, once the product has permission, the
reliability of the production has to be ensured.
An intermediate situation is encountered for fine, specialty, and agro chemicals;
the capacities are by far lower, but the margins are much higher than for basic chem-
icals. For fine chemicals, the focus is on the molecule. Usually, it has a complex syn-
thesis with many steps. Because of the low capacities, fine chemicals are in most cases
produced in batch processes, often in campaigns. There might even be no “dedicated
plant”, i.e. a plant which is used only for the manufacturing of one product. Therefore,
the optimization of the process is usually not the key issue; the reliable synthesis and
purification is more important. For process engineers, there is also a wide of variety
of jobs in exhaust air purification, waste water treatment, and waste product manage-
ment.
Specialty chemicals are similar to fine chemicals, but the main focus is on the final
application. Examples are paints, detergents, de-icing agents, or adhesives. The target
is not necessarily the sale of the product but the technical solution for the customer.
3.2 Heat integration options
As mentioned, the feasibility of a process is mainly determined by the value creation

due to the price difference between raw materials and the product. However, the value
creation itself can hardly be influenced by the engineering, it is more a question of
“yes” or “no”. Less important, but susceptible to good process engineering are the
utility costs, especially for steam and electricity. Therefore, heat integration measures
are a field where process engineers can develop their strength, i.e. suggesting reason-
able ways to save energy without significantly increasing the complexity of the plant.
Before starting with heat integration, one should be aware that the utility costs
extremely depend on the location. Currently (2016), energy of each kind is extremely
inexpensive in the United States, Saudi Arabia, or Qatar, whereas in Europe or in Asia
energy costs are significant. The steam price varies from 4 $/t to 30 $/t; it is clear that
the effectiveness of steam saving measures must be considered when deciding where
the plant will be built.
First, some standard options for heat integration are presented which occur fre-
quently, based on the case that an aqueous solution has to be concentrated. For a
better understanding, it should be pointed out that for dilute solutions large amounts
of water must be removed to increase the concentration. For example, to increase the
concentration slightly from 1 % to 2 %, about half of the water has to be removed. To
concentrate the solution from 10 % to 20 %, only 5 % of the water at the beginning has
to be evaporated.
– Reverse osmosis³:
For very dilute solutions, it is possible to use reverse osmosis as a first step to get
rid of a large part of the water without thermal energy. Details and an example can
be found in Chapter 7.1. There are membranes available which let only water⁴ pass,
whereas all other components are retained. In this way, water can be removed
until the osmotic pressure is reached. As long as the durability of the membrane is
sufficient, reverse osmosis should be considered. In the low concentration region,
large amounts of water can be removed according to the effect described above
while the energy consumption is just given by the pump energy for the pressure
elevation (usually approximately 60 bar).
– Multieffect evaporation:
Figure 3.5 shows one of the possible arrangements for multieffect evaporation
with four effects. The evaporators I–IV are arranged in series, where the pres-
sure is decreasing from effect to effect. The product to be concentrated is fed to
the system in the first effect, which is heated with fresh steam. The vapor gener-
3 Strictly speaking, reverse osmosis is not a heat integration but a heat saving measure. However, it
can support the other methods to a great extent.
4 and some other molecules similar to water in size.
ated is the heating agent for the second effect, where the product outlet of the first
effect is further concentrated. The process proceeds in this way: the vapor gener-
ated is the heating agent for the next effect; feed and vapors are in cocurrent flow
(forward feed arrangement). The advantage is that no pumps are used, the flow
of the product stream is achieved by the pressure difference between the effects.
It is the most useful arrangement if the feed is already hot (i.e. not significantly
subcooled) or if the concentrated product must not be exposed to high tempera-
tures [67]. Also, a backward feed arrangement is possible, where countercurrent
flow is realized. It is the appropriate arrangement if the flow is cold, i.e. strongly
subcooled, as the fresh cold feed is evaporated at the lowest temperature and does
not need to be heated up first. It is also the best option if the concentrated product
becomes highly viscous. In this arrangement, the concentrated product with the
highest viscosity is processed at the highest temperature so that the heat transfer
remains acceptable. Of course, backward feed arrangement needs pumps for the
transfer of the product solution.
Of course, in multieffect evaporation one must also take care that for each effect
a sufficient driving temperature difference is available.
It is quite easy to estimate how much fresh steam is saved with multieffect evap-
oration. With two effects, the first effect takes the steam and generates about the
same amount of vapor for heating the second effect. Therefore, about 12 or 50 %
Feed
Generated
vapor
Steam
Condensate
I II III IV
Product
Condensate
Cooling water
Fig. 3.5: Forward feed arrangement for multieffect evaporation.

of the fresh steam with one effect is necessary. The same considerations hold for
other numbers of effects: 33 % for 3 effects, 25 % for 4, 20 % for 5, 17 % for 6, and
14 % for 7 effects. These values indicate that more than 3 or 4 effects do not save
much more energy, not to mention that it will become difficult to provide suffi-
cient driving temperature differences and that all effects have almost the same
price, independent of their effectiveness. Only for very large plants, more than
4 effects really make sense.
– Mechanical vapor recompression (MVR):
For dilute solutions, the boiling point elevation is usually not significant. In this
way, a moderate compression of the generated vapor can elevate the dew point
of the vapor by approx. 8–10 K. The corresponding arrangement is explained in
Chapter 8.2 (Figure 8.13). Also, an example is calculated. The energy added to the
vapor by the compressor is used to elevate the dew point temperature of the vapor.
Even the compressor losses (Chapter 8.2) are not wasted but remain in the system,
causing superheating of the vapor. The necessary pressure increase can often be
achieved with blowers (Figure 8.14), the simplest form of a compressor which has
even relatively low investment costs. Mechanical vapor recompression gains the
highest possible energy savings. There are cases where up to 100 t/h of fresh steam
consumption can be replaced by a few MW el., which are by far more inexpensive
[67]. The drawback is that often more heat transfer area is necessary. The 8–10 K
mentioned above are sufficient for heat transfer to take place but is not extraor-
dinarily high. With fresh steam, the driving temperature difference can be more
or less chosen; of course, it is limited to be within certain standard ranges (Chap-
ter 4.5). However, in most cases it is several times larger so that the heat transfer
areas can be smaller. Mechanical vapor recompression has disadvantages in vac-
uum applications [67]. The main ones are:
– in a vacuum, the vapor volumes are larger so that larger compressors and
pipes are required;
– air leaks can badly affect the efficiency of vapor recompression.
However, these are only drawbacks which are not generally prohibitive to vapor
recompression in vacuum applications.
There seems to be a certain trend to mechanical vapor recompression, as it is prob-
ably the most effective steam saving measure. Concerning the blowers with their
limited pressure ratio, a multieffect compressor arrangement can be used. Blow-
ers require hardly any maintenance. At first glance, it looks as if someone wants to
pull himself up by his own bootstraps, but correctly designed mechanical vapor
recompression definitely works and has a lot of successful references.
– Thermal vapor recompression (TVR):
In thermal vapor recompression [67], instead of a compressor a jet pump (Chap-
ter 8.3) is used to increase the pressure of the generated vapor for reuse. Figure 3.6
shows a typical arrangement. The technical principle of a jet pump is explained
in Chapter 8.3. There are several differences to the conventional compression.
Feed
Steam
Product Condensate
Cooling water
Fig. 3.6: Thermal vapor recompression arrangement.
First, the compression is not driven by electrical energy but by fresh steam (“mo-
tive steam”). Therefore, TVR should only be considered if high-pressure steam is
available and low-pressure steam is sufficient for the process. Although it depends
of course on the special case, the amount of motive steam is usually in the same
order of magnitude as the suction steam. Therefore, a rule of thumb says that TVR
can replace one effect. The motive steam does not produce condensate which can
be reused but waste water, as it is contaminated by the vapor coming from the
product. The unbeatable advantage of TVRs are the low investment costs, their
reliability and the low space demand. However, there are difficulties in operation.
TVRs are designed for a single operation point; changes in operation might result
in serious performance breakdowns. In these cases, their behavior is difficult to
predict.
For heat integration, besides the effective reuse of vapor latent heat, the optimization
of the heat exchanger network is an important issue. It can be performed in a system-
atic way, where the network is analyzed as a whole. The Pinch method [68] determines
the most efficient heat transfer network just from the point of view that the use of ex-
ternal utility is minimized.
T T
H H
Fig. 3.7: TH-diagram foundations.
The main tool of the Pinch method is the representation of the heat streams in a TH
diagram (Figure 3.7). In a process, hot streams require cooling, where their enthalpy
is reduced (arrows from right to left). Cold streams require heating (arrows from left to
right), where the enthalpy is increased. There are two sorts of streams: The first ones
change the temperature during heating and cooling according to
Q̇ = ∆ Ḣ = mc
̇ p ∆T (3.9)
giving inclined arrows in the TH diagram⁵. The second sort of streams are those which
are evaporated or condensed (latent heat), where the temperature remains constant,
according to
Q̇ = ∆ Ḣ = m∆h
̇ v. (3.10)
In this case, the arrows are horizontal⁶. The heating and cooling demand of the process
can be visualized by drawing the composite curve (Figure 3.8). For each temperature,
the involved heat capacities of the streams are added up. At each point, the slope of
the curve is represented by 1/(mc ̇ p ). Finally, the latent heats are added for each tem-
perature. In this way, the composite curves for the hot streams and the cold streams
are built. The curves can be shifted parallel to the H-axis so that the cold composite
curve is below the hot composite curve.
The so-called pinch must be defined, i.e. the minimum driving temperature dif-
ference ∆Tmin (Figure 3.9) where heat transfer is considered to make sense, e.g. 10 K.
Again, the cold composite curve is shifted until the minimum temperature difference
between cold and hot composite curve is equal to ∆Tmin . In the region where the hot
composite curve is above the cold composite curve, it is possible that the hot streams
can be cooled by the cold streams and vice versa. However, at the ends the curves usu-
ally overlap. In these regions, the process does not cover the demand for heating and
cooling, respectively. Therefore cold and hot utility (steam, cooling water, etc.) is nec-
essary. The enthalpy differences represented by the overlapping spaces indicate the
5 In the following section explaining the Pinch method, it is assumed that the heat capacities of the
streams remain constant so that straight arrows are produced. In reality, this is not the case, giving
slight curvatures.
6 This is an approximation for illustration as well. In fact, for mixtures changing the phase the tem-
perature will change.
160°C 160°C
0
=2
20
Cp
120°C 120°C
Cp =
0
t => t =6
40 Cp
80°C 80°C
=
40
p
C
=
p
C
40°C 40°C
H H
Fig. 3.8: Construction of a composite curve.
Hot utility demand
160°C
120°C
Hot Composite Curve
t
Pinch point
80°C ΔTmin = 10 K
Cold Composite Curve
40°C Cold utility demand
Fig. 3.9: Hot and cold composite curves.
minimum hot and cold utility demand, respectively. If the curves are shifted more to-
wards each other so that the overlapping is further reduced, ∆Tmin would be underrun
(pinch point, Figure 3.9). Hot and cold composite curves can then be separated into a
region below and a region above the pinch (Figure 3.10). If heat integration should be
applied in the optimum way, heat must not be transferred across the pinch, or, more
detailed:
– Don’t use steam below the pinch!
– Don’t exchange heat between streams on different sides of the pinch!
– Don’t use cooling water above the pinch!
– The temperature levels of a thermal engine shouldn’t be on different sides of the
pinch.
– A heat pump should be operated across the pinch.
Furthermore, the grand composite curve can be introduced (Figure 3.11). After shift-
ing the composite curves to ∆Tmin = 0, for each temperature the enthalpy differences
between hot and cold composite curve are represented. With this construction, the
Minimum hot
utility demand
Below the pinch
Above the pinch
Minimum cold utility demand
Fig. 3.10: Regions below and above the pinch.
Hot utility Hot utility
T T
Grand composite curve
Shifted composite curves

Cold utility Cold utility
H H
Fig. 3.11: Construction of the grand composite curve.
necessary temperature of the various utilities can be determined. The “pockets” play
the key role in this context. Pockets are formed when the process can deliver part
of the required utility on its own. As a result, part of the utility can be supplied at
a more convenient temperature level, i.e. lower for heating agents and higher for cool-
ing utility (Figure 3.12)⁷. Having set these constraints, one can set up an appropriate
7 Of course, the shifting of the composite curves must be reversed when the utilities are finally chosen.
Hot utility
Necessary temperature
of the hot utility
T
“Pockets”
Necessary temperature
of the cold utility Cold utility
Fig. 3.12: Meaning of the pockets.
heat exchanger network. Also, distillation columns [69] and other equipment can be
described.
One should be aware that the pinch method just makes suggestions. Often, heat
integration measures increase the complexity of the flowsheet, or material reasons
prevent the results from being implemented. Also, the spatial distance between hot
and cold streams can be so large that heat integration is awkward. Nevertheless, pinch
analysis has proved to be a useful tool to get an overview on the heat integration op-
tions. In the author’s view it is even superior to the exergy analysis, although the
latter is more elaborated according to thermodynamics. The pinch analysis reflects
to heat integration, which can usually be implemented without changing a success-
fully working process. The exergy analysis covers other issues as well, e.g. dilution or
neutralization as contributions to chemical exergy. Exergy analysis does not give an
answer whether avoiding the losses is in line with the targets of the process or not.
The opportunities for improvement often require major changes of the process, which
might be linked with additional test phases. Also, the mechanical exergy often yields
only minor contributions, and they can easily be detected by having a simple look at
the pressure levels in the process.
3.3 Batch processes
Most processes in fine chemicals, specialty chemicals, and pharmaceuticals are not
operated as continuous but as batch processes, meaning that a specified amount is
produced within a certain time. Often, batch products are not manufactured in dedi-
cated plants but in multipurpose units, where several products can be produced in the
same plant, just according to the demand. The engineering of such a plant comprises
1. Reaction
1.1. Purge unit V201. Purge 3 times with NITROGEN. Then setup blanket to maintain 1 atm pressure in
the unit.
1.2. Charge V201 with 120 kg of Int-I.
1.3. Charge V201 with 140 lb of DCM-Liquid.
1.4. Heat unit V201 to 30 C. The heating time is 15 min. Dissolve 100 % of all solids.
1.5. Charge V202 with 50 kg of R-Liquid.
1.6. Transfer contents of unit V202 to V201. The transfer time is 15 min.
1.7. React in unit V201 via Reaction-R1-R2. The reaction is Isothermal. The final temperature of the
batch is 30 °C.
1.8. QC-Test the material in unit V201. Test for LC. The expected test time is 10 min. The batch should
meet the following specification: Test for completion of reaction.
1.9. Charge V203 with 250 liter of B-MeOH.
1.10. Transfer contents of unit V203 to V201. The transfer time is 15 min.
1.11. React in unit V201 via Reaction-R3-R4. The reaction is Isothermal. Reaction occurs over 45 min.
The final temperature of the batch is 30 C.
1.12. Charge V201 with 20 liter of METHANOL. The charge time is 15 min.
1.13. Cool unit V201 to ‒15 °C. The cooling time is 15 min.
2. Separation
2.1. Start sweep in unit T901.
2.2. Filter the batch from unit V201 in filter F201. The mother liquor, named Mother Liquor 1, is sent to
T901. The cake contains 95.5 % of D-Solid in the Solid phase. 99 % of IMPS-1 in the Solid. 99 % of
IMPS-1 in the Solid phase, 99 % of IMPS-2 in the Solid phase and 99 % of Pp in the Solid phase.
The moisture content in the final cake is 30 %.
2.3. Charge V201 with 300.3 liter of WATER.
2.4. Filter the batch from unit V201 in filter F201. The mother liquor, named Mother Liquor 2, is sent
to T901.
2.5. Wash the cake in unit F201. For each wash, use 300.3 liter of WATER. Wash the cake 1 times. Use
a Slurry wash. The moisture content in the final cake is 60 %.
2.6. Transfer contents of unit F201 to DR-100.
2.7. Dry the batch in unit DR-100. The drying time is 1.5 h. The moisture content in the final cake is
40 %.
2.8. Age the contents of unit T901 for 30 min. Agitation is OFF.
2.9. Decant in unit T901 over 40 gal/h. Separation is: 33.33 % of D-Solid goes to Top. 41.17 % of IMPS-
1 goes to Top, 66 % of METHANOL goes to Top, 41.17 % of IMPS-2 goes to Top and 91 % of WATER
goes to Top. Unspecified materials go to Bottom. The upper layer (category: High BOD waste),
named High-BOD-Layer, is sent to Storage.
2.10.Transfer contents of unit T901 to Disposal. The transfer stream (category: Organic Waste) is nmed
Heavy-Organic-Layer.
2.11. Stop sweep in unit T901.
2.12.Transfer contents of unit DR-100 to DR-101.
2.13.Dry the batch in unit DR-101. The drying time is 10 h. The moisture content in the final cake is
0.5 %.
2.14.Transfer contents of unit DR-101 to Drums.
Fig. 3.13: Example for a batch process recipe.

not only making up the dimensions but also scheduling the charges of the equipment,
the duration of the various process steps and the optimization of the load of the plant.
Meanwhile, tools are available which enable a comprehensive documentation and vi-
sualization of the process. Flowsheets, equipment lists, mass balance, the contents
of the particular pieces of equipment, the emissions and the time schedule can be
generated with a batch simulation.
The basis of the batch simulation is the recipe, which is in principle a standardized
process description. A special language consisting of a limited number of expressions
has been developed, covering all the possible steps in a batch process. It contains the
whole information about the process. The recipe links the unit operations together. It
is similar to a laboratory instruction. As the language consists of standard phrases,
an automatic translation into other languages is easily possible. Figure 3.13 shows an
example of a batch recipe. As well as for the continuous process, a simple flowsheet
(“equipment diagram”) can be derived from the recipe (Figure 3.14). It visualizes the
flow between the pieces of equipment and relates them to the recipe steps. For each
V203
1.1. Purge-4
1.2. Purge-5
1.3. Purge-7
1.12. Purge-25
2.3. Purge-34
1.10. Transfer-20 2.2. Filter-32

2.4. Filter-39
1.5 Charge 11 1.1. Purge-1
1.1. Purge-1
1.6. Transfer-13
1.1. Purge-1
V202 V201
1.4. Heat-10
1.7. React-17
1.11. React-24
1.13. Cool-28
Fig. 3.14: Example of an equipment diagram.

Vessel volume
1.200
Liquid + Solid volume (liter)
1.000
800
600
400
200
0
0 100 200 300 400 500 600
Time (min)
Fig. 3.15: Equipment content visualization.

piece of equipment, the content as a function of time can be visualized (Figure 3.15),
giving valuable advice for its design. Moreover, it can be evaluated when emissions
take place and what the amount of emissions is. Especially, this is most relevant for
the exhaust air concept (Chapter 13.4). In contrast to a continuous process, where the
exhaust air amount is usually well defined, the exhaust air in batch processes is hard
to trace, for example, exhaust air is generated each time when a vessel is filled or
flushed. If the temperature in a vessel rises, vessel breathing takes place due to thermal
expansion.
However, the heart of a batch simulation program is the schedule view (“Gantt
Chart”). An example is shown in Figure 3.16. It is the most valuable information for
the staff to operate and organize the process and to evaluate whether it is feasible at
all. There are tools for optimization of batch schedules, the following example shall
illustrate the large potential of such a tool in a very simple case.
Fig. 3.16: Example for a Gantt Chart.

Approach 1
5 5 5 2 2 2
Vessel 1
A B
2 2 2 4 4 4
Vessel 2
Makespan = 29 hrs
Time
Approach 2
5 2 5 2 5 2
Vessel 1
A B A B A B
2 4 2 4 2 4
Vessel 2
Makespan = 25 hrs
Time
Fig. 3.17: Example for makespan optimization.
Example
A batch process for a specialty chemical consists of the following operation steps:
1. producing an intermediate of product A in vessel 1 in 5 h;
2. producing an intermediate of product B in vessel 1 in 2 h;
3. finalizing product A in vessel 2 in 2 h;
4. finalizing product B in vessel 2 in 4 h.
For producing an appropriate amount for selling, three cycles for each product are necessary. Optimize
the makespan, i.e. the time needed for production in the plant. The cleaning of the vessels must take
place after each step and is already included in the given durations of the process steps.
Solution
Figure 3.17 shows two options.

In Approach 1, vessel 1 first produces the whole amount of the intermediate of product A. After a
batch of A has been finished, it is finalized in vessel 2. Then, vessel 1 is used to produce the interme-
diate of product B. When the first batch of B is ready, in vessel 2 the last batch of product A has just
been finalized, so one can directly continue with product B. The makespan for Approach 1 is 29 h. It
has certain advantages, as both vessels can be operated partly in parallel. Moreover, in both vessels
only one product change takes place so that the cleaning effort is comparably small.
However, as mentioned above the cleaning must be performed anyway and is already consid-
ered in the time schedule. In this case, Approach 2 has considerable advantages, as an alternating
production of the products A and B takes place. The overlapping times, where both vessels are used
in parallel, are much larger. Therefore, the makespan is reduced to 25 h, giving a capacity increase of
almost 14 %. This example has been executed more or less by manual rearranging of the time frames,
as the process is simple enough. For complex cases, a capable software for optimization is required.
3.4 Equipment design
Besides process simulation, equipment design is the second large task for engineer-
ing calculations. There is a clear trend in the process simulator programs to integrate
process simulation and equipment design. For instance, a heat exchanger could be
represented in different ways with increasing complexity:
– a simple heater block, which just regards the energy balance of one of the two
streams, yielding the necessary duty;
– a heat exchanger block, taking into account the energy balance of the second side
of the heat exchanger as well. Especially if both streams are process streams, it
often happens that an additional tear stream is introduced, when the conditions
of the second stream are a result of the first stream. An advantage is that the tem-
peratures of the two streams are compared in each segment so that it is supervised
that the cold stream has a lower temperature than the hot stream everywhere.
– the heat transfer coefficient can be taken into account, either by setting it to a
fixed value or even by a calculation, taking into account the specification of the
heat exchanger.
In the author’s opinion there are advantages to clearly separate process simulation
and equipment design. Sticking to the example of the heat exchanger, reducing the
complexity is normally very useful for process simulation, and estimated heat trans-
fer coefficients will only give a brief impression of the necessary equipment. It is useful
that the process simulation just defines the requirements of the process which should
be achieved by an adequate equipment design. In the design procedure itself, the engi-
neer gets a certain feeling and experience for the sensitivities of the various adjusting
screws. It is, however, necessary to adjust some parameters in the simulation after the
design has been set up, e.g. the pressure drop. Nevertheless, it can still be decided
whether the value obtained in the process design or a maximum allowable value is
used in the simulation.
Another item is the validity of the documents during an engineering project. If
the design parameters of a heat exchanger or a column are directly involved in the
process simulation, it might happen that minor changes due to new information about
the process have to be considered, e.g. a fouling factor in a heat exchanger or a choice
for a more inexpensive packing type. In this case, the outlet streams of this piece of
equipment will slightly change, which in turn has an influence of all the connected
downstream equipment and recycles. Their specifications are probably only slightly
affected, but no more consistent with the mass balance – a nightmare for quality man-
agement. To take the mass balance only as constraints for the equipment design is
more effective and less sensitive to changes.
Equipment design comprises the choice of the kind of equipment, the fixing of
design options (e.g. for a shell-and-tube heat exchanger: hot fluid on shell side or
tube side), the determination of the dimensions including both the process and the
mechanical aspects and the limits of operation conditions. In process engineering,
there is a peculiarity concerning the overdesign: a piece of equipment in process en-
gineering can be too small or too large; both incidents can lead to a lack of function.
In other engineering disciplines, a piece of equipment which is too large is usually
more expensive and therefore uneconomical, but the function is achieved. Therefore,
special care must be taken to make sure that the particular pieces of equipment can be
used in all required application cases, and, furthermore, the applications theirselves
should be questioned and discussed. Due to the limited accuracy in process engineer-
ing calculations, safety factors are certainly necessary, but nonreasonable ones can
be detrimental to the function of the equipment [70].
In the following chapters, the main design aspects of various kinds of equipment
are outlined.
3.5 Troubleshooting
Round up the usual suspects!

(Claude Rains as Captain Renault in “Casablanca”)
Process simulation is a complex exercise and therefore prone to errors and mistakes.
In most cases⁸ only one single error is responsible for failure in process simulation. It is
usually no use to detect even more failures, as they all might refer to the same reason.
The most important thing is to keep calm and analyze what happened. In fact, as in
the citation above, there are a number of usual suspects, and, in contrast to the movie,
there is a high probability that they are responsible for the failure. And even if not: the
defined work on the process helps to obtain better understanding of the calculation
and the behavior of the simulation, and often it is the key for fixing the failure at the
end of the day.
8 but, to my regret, not always.

If process calculations give strange results or do not match the process, the fol-
lowing things should be done first.
– Check if systems with an LLE are calculated with the three-phase-flash (VLLE):
If systems with an LLE are calculated without taking the LLE into account (Chap-
ter 2.5), strange results are obtained, and they are not always so obviously wrong
as in Figure 2.17. Undoubtedly, this is the error which occurs most often. Unex-
pected mixing temperatures or crazy phase equilibria are clear hints that the setup
in the phase equilibrium calculation might be wrong.
– Model change:
Model changes lead to inconsistencies in the enthalpy description (Chapter 2.11).
Any region where a model change takes place must be carefully checked, even if
nothing suspicious happens.
– Binary parameters:
Check whether the decisive binary pairs are reliable.
– Check whether the numbers of theoretical stages in columns are adequate. If small
concentrations occur, it might be useful to apply a mass transfer based calculation
(p. 184).
– Components are missing:
Often the wrong components are being looked at. One must double-check with
the plant manager that the component list is correct.
– Enthalpy description:
The liquid heat capacity is always a quantity which causes errors (Chapter 2.8).
If energy balances of liquid phases look strange, one should carefully check
whether the errors in cLp are acceptable.
– Component removal:
When strange physical properties or phase equilibria occur and the reason is not
identified, it might happen that one of the components is the reason because of
a wrong parametrization. First, a plot of the quantity to be checked should be
generated, showing all pure components. If it is still not clear which one it is, the
strategy to remove the components one by one can be applied. If the error vanishes
after having removed one component, the erroneous component is identified and
should be further examined.
– Component accumulation:
As in the case study (p. 86) with the ammonia-water system, it often happens that
no outlet for one or more components has been provided. A simulation cannot
converge if one component has no option to leave the system. Process simulation
programs usually save the tear history, where it can be checked for which compo-
nents difficulties occur to fulfill its mass balance.
– Convergence of column calculation:
Distillation is the most crucial unit operation block in a process simulation. For
flowsheet convergence, it is not crucial if during flowsheet evaluation a column
once does not fully meet the convergence criterion, as long as it is not the case
in the final step. If a column calculation has no tendency to converge at all, it is

strongly recommended to stop the process simulation. Otherwise, the simulation
will be continued with a meaningless result, which is then distributed into the
rest of the flowsheet. All the other blocks downstream get wrong input values and
yield bad results on their part. Finally, bad starting and input values are spread
all over the flowsheet, which makes convergence even worse.
– Wrong plant information:
Data coming from the plant staff might not be in the desired form. Mass balances
to be reproduced might be inconsistent, especially compositions might be biased.
Even technical terms can be used in different ways; a classic error source is a
wrong application of the reflux ratio (Equation (5.1)) in distillation columns.
4 Heat exchangers
Most of the heat exchangers do not work because of, but in spite of our design …
(Hans Haverkamp)
Heating and cooling of streams are essential unit operations in any process. They are
usually carried out in so-called heat exchangers, and their design is one of the main
requirements of a useful process engineering. There are a lot of different aspects that
have to be taken into account, and both process and construction engineers must give
their input to achieve a good solution.
In contrast to academia, the fundamentals of heat transfer play a minor role in
industrial applications. The necessary relationships are already integrated in com-
mercial heat exchanger programs like HTRI, HTFS, or ASPEN Heat Exchanger Design &
Rating. Instead, the focus is the reasonable use of the particular options for the design
of the apparatus. It is attempted to give a good explanation of these options in the fol-
lowing chapter; for the fundamentals, there are a lot of other textbooks available, e.g.
Baehr/Stephan [71] or the VDI Heat Atlas [72].
There are two main types of apparatuses for heat exchange, the shell-and-tube
type and the plate heat exchangers. Other types, e.g. spiral heat exchanger, double
pipes etc., are not or only briefly considered in this book. Much information about
them is given in [72]. The shell-and-tube type is still the more widely applied because
of its robustness and flexibility. It will be discussed in detail, as their specification is a
standard task of a process engineer. Plate heat exchangers are outlined more briefly;
usually, a vendor specialist is necessary to obtain an optimum design.
4.1 Something general
Heat exchangers transfer heat from one fluid to another one, the two of which do not
come into contact because of a separating wall. The design of a heat exchanger is a
classical heat transition problem, meaning that the heat has to be transferred from
one of the fluids to the separating wall, then go through the wall and finally be trans-
ferred from the wall to the second fluid. Fouling layers can make this heat transfer
more difficult; they can be considered in the calculation analogously to the separat-
ing wall (Figure 4.1).
There is a strong analogy to electrical engineering, where electric current and volt-
age are linked by the resistance in Ohm’s law:
U
R= . (4.1)
I
A series of resistances can be described by adding them up:
Rtotal = R1 + R2 + R3 + ⋅ ⋅ ⋅ . (4.2)
108 | 4 Heat exchangers
Hot temperature Cold fluid
Boundary layer
Surface temperature
Boundary layer
Surface temperature
Wall
Hot fluid Cold temperature
Fig. 4.1: Steps in the heat transfer process.
For heat transfer from a fluid to a wall, the standard approach is
Q̇ = αA∆T . (4.3)
Comparing Equations (4.1) and (4.3) and considering that Q̇ and I have the same mean-
ing as ∆T and U, a thermal resistance can be defined as
∆T 1
Rα = = . (4.4)
Q̇ αA
The heat transition can then be characterized by a series of thermal resistances:

1 1 1
Rk = = + Rfouling,1 + Rwall + Rfouling,2 + , (4.5)
kA α1 A1 α2 A2
with the overall resistance R k , defined as 1/(kA) to give it the same structural meaning
as α¹. The heat transfer areas A1 and A2 can differ, e.g. the inner and the outer surface
of a tube. The A in the term “kA” is arbitrary, it does not matter to which area it refers
unless k itself is evaluated. The thermal resistance of the separating wall Rwall can be
calculated according to the rules of heat conduction [71]. The fouling resistances are
usually set to empirical values (Table 4.2).
It should be clear that the magnitude of these resistances gives a clear guideline
how to improve the performance of a heat exchanger. Good heat exchanger calcula-
tion programs indicate the percentages of the particular thermal resistances. The high
thermal resistances should be counteracted, as the following simple example shows.
1 Note that in the English and American literature the heat transfer coefficient α is normally denoted
as h, and the heat transition coefficient k is normally denoted as U.
4.2 Shell-and-tube heat exchangers | 109
Example
A heat exchanger is calculated to have an α1 = 150 W/(m2 K) and an α2 = 2000 W/(m2 K). What im-
provement is required? The resistances of a fouling and separating wall should be neglected. The heat
exchange area is 100 m2 on both sides.
Solution
The overall resistance can be calculated to be

1 1 1 K
= + = 7.17 ⋅ 10−5
kA 150 W/(m2 K) ⋅ 100 m2 2000 W/(m2 K) ⋅ 100 m2 W
⇒ k = 139.5 W/(m2 K) .
Increasing the lower heat transfer coefficient by 10 % gives
1 1 1 K
= + = 6.56 ⋅ 10−5
kA 165 W/(m2 K) ⋅ 100 m2 2000 W/(m2 K) ⋅ 100 m2 W
⇒ k = 152.4 W/(m2 K) ,
while doubling the higher one gives
1 1 1 K
= + = 6.92 ⋅ 10−5
kA 150 W/(m2 K) ⋅ 100 m2 4000 W/(m2 K) ⋅ 100 m2 W
⇒ k = 144.6 W/(m2 K) .
A small increase of the lower heat transfer coefficient is much more efficient than a large change of
the higher one.
4.2 Shell-and-tube heat exchangers
Shell-and-tube heat exchangers are composed of a shell, which is in principle a pres-

sure vessel, and a tube bundle inside (Figure 4.2). The two fluids which are supposed
to exchange heat are on different sides of the tube; one inside the tubes and one out-
side the tubes within the shell-side. Often, only one of the streams involved is a process
stream, whereas the other one (e.g. steam for heating, cooling water) is a utility (Chap-
ter 13). On the other hand, it is desirable to reduce the consumption of utilities and
to cover the necessary heating or cooling duty from the process itself so that both
streams are process streams. This heat integration can save operation costs signifi-
cantly; the Pinch method for the optimization has been described in Chapter 3.2. The
heat exchangers can be classified with respect of the phase behavior of the streams;
they can maintain their phases (gas-gas or liquid-liquid exchangers), or the product
stream can condense (condenser) or evaporate (evaporator). In commercial heat ex-
changer design programs (e.g. HTRI, HTFS), the principle of the thermal calculation
Fig. 4.2: Transparent shell-and-tube heat exchanger. Test fluid with red dye.
Courtesy of Heat Transfer Research, Inc.
ṁ1 , t1,in
ṁ2 , t2,out
ṁ2 , t2,in
Fig. 4.3: Cell method for the thermal calculation
of a shell-and-tube heat exchanger.
ṁ1 , t1,out © Springer-Verlag GmbH.
of a shell-and-tube heat exchanger is the so-called cell method [72], which is illus-
trated in Figure 4.3.
Using a simple liquid-liquid heat exchanger, the procedure is illustrated in the
next chapter.
4.3 Heat exchangers without phase change
Each cell must be assigned with the actual state of the stream and its associated phys-
ical properties, with the geometry of the cell and with the appropriate relationships
which describe the heat transfer. In contrast to process simulation programs, com-
mercial heat exchanger design programs do not support physical property models
or their parameters except a number of common heat transfer fluids like steam or
water, cooling brines, thermal oils, and some common pure components and ideal
mixtures of them. Instead, the physical properties needed are generated before the
actual heat transfer calculation takes place. The communication between the physi-
cal property model and the heat exchanger design program is achieved with the help
of a so-called heat curve. An example for the product side heat curve of a liquid-
liquid heat exchanger without phase change is given in Figure 4.4. As can be seen, the
particular physical properties with respect to temperature, i.e. specific enthalpy, den-
sity, dynamic viscosity, thermal conductivity, and molecular weight, are tabulated.
Between these points, the program interpolates. Extrapolations are usually indicated
with warning messages. Each heat curve refers to a certain pressure. Normally, several
heat curves for different pressures are generated to account for pressure drop effects.
For a liquid-liquid exchanger, this is of minor importance. As well, the vapor fraction
of the stream is always zero in this case. The step size for the temperature can be varied
to obtain an appropriate number of points. Alternatively, the enthalpy can be used as
an independent variable. The pseudo-critical temperatures and pressures are of minor
physical significance but required by some correlations. To make a good interpolation
behavior possible, the user should take care that the distances between the points are
appropriate and that the whole temperature and pressure range occurring in the heat
exchanger is covered. The thermodynamics is completed by the specification of the
process in the heat exchanger.
p = 6 bar Liquid Properties

t h Vapor ρ η λ cp tc pc M
fraction (pseudo) (pseudo)
(°C) (J/g) weight (kg/m3 ) (mPa s) (W/(K m)) (J/(g K)) (°C) (bar) (g/mol)
100.000 −10 767 0 882.724 0.2953 0.2018 2.985 337.49 166.96 27.39
94.137 −10 784 0 888.618 0.3138 0.2034 2.967 337.49 166.96 27.39
88.240 −10 802 0 894.406 0.3345 0.2049 2.950 337.49 166.96 27.39
82.310 −10 819 0 900.089 0.3578 0.2054 2.934 337.49 166.96 27.39
76.348 −10 837 0 905.669 0.3842 0.2079 2.919 337.49 166.96 27.39
70.356 −10 854 0 911.147 0.4141 0.2084 2.904 337.49 166.96 27.39
64.335 −10 871 0 916.524 0.4482 0.2109 2.891 337.49 166.96 27.39
58.287 −10 889 0 921.802 0.4873 0.2124 2.878 337.49 166.96 27.39
52.214 −10 906 0 926.980 0.5324 0.2139 2.867 337.49 166.96 27.39
46.117 −10 924 0 932.060 0.5847 0.2154 2.856 337.49 166.96 27.39
40.000 −10 941 0 937.041 0.6457 0.2168 2.847 337.49 166.96 27.39
Fig. 4.4: Example for a heat curve without phase change.

Commercial heat exchanger design programs distinguish between three calculation

modes, i.e. the rating mode, the simulation mode, and the design mode. They can be
distinguished as follows.
– Rating mode:
The specified heat exchanger is calculated according to the process. As the main
result, it is indicated how much heat exchange area is excess or, respectively, miss-
ing (overdesign). This is the standard mode for the design of heat exchangers.
– Simulation mode:
It is evaluated how the specified heat exchanger would perform with the given
input streams, i.e. the actual outlet conditions are calculated.
– Design mode:
A heat exchanger design is evaluated which fulfills the requirement of the pro-
cess (outlet conditions, pressure drop). This is a very tempting approach. the heat
exchanger design is achieved by the famous mouse click. However, this mode is
time-consuming, and the result is not necessarily satisfactory; it should not be
taken as final but as a starting point for further rating mode calculations. Fur-
thermore, the user does not get a feeling for the sensitivities and the potential for
further improvement. The design mode is only recommended when the user has
really no idea about the design.
For the constructive details of a heat exchanger, the TEMA type (TEMA: Tubular
Exchanger Manufacturers Association Inc.) has to be fixed first. The TEMA type deter-
mines the general arrangement of the heat exchanger. Figures 4.5, 4.6, and 4.7 explain
the TEMA type code. Some popular choices are
– BEM: standard arrangement;
– BEU: U-type heat exchanger;
– BKU: kettle type reboiler (Chapter 4.5);
– AES: floating head, removable tube bundle;
– BJ21T: arrangement for vacuum condensers (Chapter 4.4).
A B C N D
Fig. 4.5: TEMA front end stationary head types.

A Channel and removable cover;
B Bonnet (integral cover);
C Channel integral with tubesheet, removable cover, and removable tube bundle;
N Channel integral with tubesheet and removable cover;
D Special high pressure closure.
Courtesy of Tubular Exchanger Manufacturers Association, Inc.
E F G H
J K X
Fig. 4.6: TEMA shell types.

E One pass shell;
F Two pass shell with longitudinal baffle;
G Split flow;
H Double split flow;
J Divided flow;
K Kettle type reboiler;
X Cross flow.
L M N P
S T U W
Fig. 4.7: TEMA rear end head types.

L Fixed tubesheet like A, stationary head;
M Fixed tubesheet like B, stationary head;
N Fixed tubesheet like N, stationary head;
P Outside packed floating head;
S Floating head with backing device;
T Pull through floating head;
U U-type bundle;
W Externally sealed floating tubesheet.
Additional information can be obtained from [73].

Next, the shell orientation (horizontal, vertical, inclined) must be defined and it has
to be decided which stream is on the shell side and which is on the tube side. This is
a strategic decision with often contradictory arguments. Usually, the heat transfer on
the shell side is better than in the tubes. Therefore, it is desirable to place the stream
with the worse heat transfer on the shell side. On the other hand, the tube side is
easier to clean, so that the stream showing more fouling should be placed in the tubes.
The latter argument is usually stronger; for example, cooling water as a notoriously
dirty fluid is placed in the tubes in almost all cases. A compromise can be found if
a removable tube bundle or a U-type exchanger is chosen; in these cases, the shell
side can be cleaned as well. Countercurrent flow is the default; cocurrent flow can be
specified. Additionally, several identical heat exchangers can be arranged in parallel
or in series.
Shell diameter and the length of the tubes mainly determine the heat transfer area.
Several other specification data for the tubes have to be defined. The tubes theirselves
are specified by their outside diameter (OD) and the wall thickness. A standard value
for the tube OD is 1′′ (25.4 mm). The most often applied alternatives are ¾′′ (19.05 mm)
and 1 ½′′ (38.1 mm), where it is possible to increase the number of tubes in a given
shell or reduce the velocity inside the tubes, respectively. The wall thickness is de-
termined by the mechanical stability. 2 mm is a reasonable value, for high-pressure
applications, larger wall thicknesses are probable.
The tube pitch and the tube layout angle define the arrangement of the tubes (Fig-
ure 4.8). The pitch is the distance between the centers of the tubes. The smaller the
pitch, the more tubes can be put into the shell. Large pitches can be useful to lower
the shell velocities for avoiding vibrations. The pitch is often given as a pitch ratio, i.e.
the pitch divided by the tube OD. Common pitch ratios are 1.25, 1.33, and 1.5. The tube
layout angle defines the pattern of the tubes with respect to the flow direction. The 30°
arrangement is the standard one. The 60° pattern is useful to avoid vibrations caused
by vortex shedding (Chapter 4.10). Besides these triangular patterns the square pat-
terns are used. The 45° pattern should not be used for gas streams because of possible
vibrations, the 90° arrangement is useful for cleaning purposes.
45° 30° 60°
90°
t
t
t t
t t
90° t 90° 60° 60° t
a b c d
Fig. 4.8: Tube pitch patterns. t = pitch.

© Springer-Verlag GmbH.
Tube-side
Straight-tube heat exchanger Shell-side fluid in
(two pass tube-side) fluid in
Tube sheet Tube bundle with

straight tubes
Shell
Tube sheet
Baffles
Shell-side
fluid out Tube-side
fluid out
Fig. 4.9: Sketch of a two-pass shell-and-tube heat exchanger [74].

© H Padleckas/Wikimedia Commons/CC BY-SA 3.0.
https://creativecommons.org/licenses/by-sa/3.0/deed.de
Finally, the number of tube passes can be specified (1, 2, 4, 8 in standard designs). At
the front and rear head, the tube stream can be divided by partition plates so that it
passes several times through the exchanger (Figure 4.9). The simplest way is the use
of a U-type heat exchanger, which has automatically two passes. As the cross-flow
area for the tube stream decreases with the number of passes, the tube velocity and
therefore the heat transfer coefficient increases. On the other hand, one direction is
in cocurrent flow, and the profile of the temperature difference between hot and cold
fluid along the flow path is distorted, often even leading to temperature crosses. It
makes sense to define more tube passes if the main thermal resistance is on the tube
side and if the temperature ranges of hot and cold fluid do not overlap. Otherwise, the
use of more tube passes can even be a disadvantage.
The material of the tubes is characterized by standard values (density, thermal
conductivity etc.). They can be overwritten if further knowledge is available.
The purpose of baffles is to direct the shell side flow and to prevent vibrations by
supporting the tube structure [75]. Again, there are some options for different types
(Figure 4.12).
The most common one is the single segmental baffle, which is in principle a circu-
lar plate where a segment has been removed. This is defined by the cut (Figure 4.10).
A reasonable cut should be in the range 20–25 %. If low-pressure gas flow is involved,
the first approach should be 40–45 %.
It can be distinguished whether the cut is perpendicular or parallel to the flow in-
let (Figure 4.11). The parallel orientation is preferred for condensing fluids so that the
condensate can be collected at the bottom. For liquids, the perpendicular orientation
Cut = h/D
Fig. 4.10: Baffle cut definition.

D Courtesy of Heat Transfer Research, Inc.
should be preferred, as it mixes existing fluid layers and avoids possible precipitation
of solids at the bottom. Single segmental baffles are certainly the cheapest ones be-
cause of their easy manufacturing, but they cause a comparably large pressure drop.
They are not recommended for viscous fluids [75]. The crossflow heat transfer to the
tubes is better than the longitudinal heat transfer. The more baffles are set, the more
this crossflow is achieved and the more the pressure drop increases. Therefore, the
baffle spacing can be varied within certain limits to achieve a satisfactory solution.
According to TEMA, the minimum baffle spacing is 20 % of the inner shell diame-
ter. For small heat exchangers it should not go below 2′′. A baffle spacing between
30–60 % of the shell inside diameter is usually a good starting point [75]. Baffle spac-
ing can be varied if vibrations occur; vibrations are less probable with more baffles
set, as the tubes get more support.
Perpendicular Parallel
baffle cut baffle cut Fig. 4.11: Baffle orientations.
The pressure drop can be significantly reduced if double-segmental baffles are used.
As can be seen in Figure 4.12, two kinds of baffles are alternating, with one having
the cut area in the center (“wing baffle”) and one having two circle segments as cut
areas (“center baffle”). From the thermal point of view, they are less effective. Also,
the so-called disk-and-donut baffles are in use, with one baffle type in the form of a
circular ring and one as a circular disk in the center. They are often used in gas-gas
applications to avoid vibrations.
Support plate
Baffle
Single‐segmental Double‐segmental No‐tubes-in‐window
Fig. 4.12: Common baffle types.
NTIW (“no tubes in window”) baffles (Figure 4.12) are used for mechanical stability
reasons. This option ensures that each baffle supports every tube. Tubes with long
areas without support are avoided. NTIW is a useful option if vibration problems oc-
cur. Also, the pressure drop is reduced. The disadvantage is that the shell diameter
must be increased to obtain the same heat transfer area. A baffle cut of 15 % is most
common [75].
Grid baffles are metal lattices that fix the tubes (Figure 4.13). Mainly, longitudinal
flow is produced instead of cross-flow. Grid baffles protect against tube vibration and
produce low pressure drops on the shell side.
Fig. 4.13: Rod-type baffles as an example for grid baffles.

Courtesy of TEMA India Ltd.
To finish the paragraph on baffles, tie-rods and sealing strips should be explained. A
small number of tie-rods are placed in the tube bundle instead of normal tubes to gain
more mechanical stability. They are not tubes, but rather massive sticks and do not
take part in the heat exchange. Sealing strips are placed on the circumference of the
inner diameter. They prevent a leakage flow between bundle and shell (Figure 4.14).
3–5 mm gap
Edge of baffle (baffle cut)

Sealing strips
3–5 mm gap
Dummy tubes
Tie rods
Fig. 4.14: Sealing strips.

In the rating mode, the specified heat exchanger design produces an overdesign as a
result, i.e. a statement whether the heat exchange area is too small or too large and
how much. A reasonable overdesign is 10–20 %². The overdesign must cover the vari-
ous uncertainties in the calculation, e.g. physical properties, uncertainties of the heat
transfer relationships or, if necessary, fouling effects (Chapter 4.9). The design must
be varied until the overdesign is in the desired range. However, there are a lot of other
items to be checked.
– Duty comparison:
The calculated duty must be equal to the duty reported in the process simulation.
This is a quite safe indication of whether the physical properties and the process
definition have been defined correctly in the heat exchanger design program.
– Flow fractions:
The flow fractions indicate to which extent the fluid on the shell side takes the
designated way. Heat exchanger design programs give an estimation about the
flow distribution. The percentages of the following fractions are regarded [76]:
2 It should be noted that the overdesign in the rating mode refers to the area of the heat exchanger. Of-
ten, it is checked whether the overdesign remains slightly positive if the load is increased by 10–20 %.
In fact, this is not equivalent, as increased load causes higher velocities and therefore better heat
transfer coefficients. This approach is less conservative.
– A fraction: The A fraction refers to the tube-to-baffle hole leakage stream. It

becomes large when the clearances between tubes and baffle holes are large
and when the baffle spacing is narrow, especially for single-segmented baf-
fles. At least, the A fraction is thermally effective and not lost.
– B fraction: The B fraction refers to the main crossflow stream through the bun-
dle, i.e. the desired way. Normally, it is approx. 60 % of the total flow. If the
B fraction is lower, too large clearances and a narrow baffle spacing are prob-
ably the reasons.
– C fraction: The C fraction is the bundle-to-shell crossflow bypass stream. It
should be less than 10 %. Additional sealing strips can decrease this fraction.
The C fraction is thermally not effective.
– E fraction: The E fraction is the baffle-to-shell leakage stream. It is thermally
not effective. There are hardly options to manipulate it. Double-segmental
baffles are advantageous in comparison with single-segmental baffles.
– F fraction: The F fraction is the tubepass partition bypass stream; it should be
lower than 10 %. The F fraction can be lowered by additional sealing strips.
– Thermal resistance distribution:
The percentage ratios of heat transfer on the shell side, heat transfer on the tube
side, thermal conductivity resistance of the tube and thermal conductivity resis-
tance of the fouling layer to the overall thermal resistance of the arrangement is
a useful guideline for improving the design. It can be found out which side de-
termines the heat transfer, so that the following design variations should focus
on this. Furthermore, the percentage of the fouling resistance should be watched
carefully; it can be assessed which impact the fouling factors have on the design.
An example is given in Chapter 4.9.
– Tube side velocities:
One should take care that the tube side velocities are sufficiently high, especially
if fouling is probable (Chapter 4.9). A range of 0.4–1.2 m/s is recommended for liq-
uids, the VDI Heat Atlas [72] even suggests 1.8 m/s. For vapors, the kinetic energy
is more relevant than the velocity, as the density of a gas can cover a wide range.
The recommended range is ρw2 = 30–270 kg/ms2 , corresponding to the velocity
of 5–15 m/s for air at p = 1 bar. The tubeside velocity is often a crucial point for the
heat exchanger design. In fact, there are examples where the choice of a smaller
shell diameter improves the heat exchanger performance, when the velocity in the
tubes and, subsequently, the heat transfer coefficient increase significantly. Also,
there is the option of providing multiple tube passes to increase the tubeside ve-
locity (see above).
– Shell side velocities:
The velocities on the shell side should not be too large to avoid vibrations. The
guidance values are w = 0.3–0.9 m/s for liquids and ρw2 = 30–130 kg/ms2 , cor-
responding to the velocity of 5–10 m/s for air at p = 1 bar.
– Vibrations:
At least, the most important vibration checks should indicate that vibrations do
not occur (Chapter 4.10).
– Physical properties:
As mentioned above, it should be checked whether the calculated duty meets the
expectation from the process simulation. The heat curves should cover the pres-
sure and temperature range of the process with a sufficient number of points for
interpolation.
For shell-and-tube heat exchangers, there is an interesting but not widely applied
option to increase the heat transfer coefficient in the tubes, where often not enough
turbulence is generated because of low velocities. Turbulence can be increased with
wire elements (Figure 4.15), which can be inserted into the tubes [77].
Fig. 4.15: hiTRAN element for increase

of the heat transfer in the tubes [77].
4.4 Condensers
For condensers, the heat curve on the product side looks slightly different. In contrast
to the liquid-liquid or gas-gas heat exchangers, it is important that dew and bubble
point of the stream are reproduced well by the heat curves. For the properties, both
phases are relevant. Due to the phase change, the influence of the pressure is larger.
Also, the calculated heat duty should agree with the one obtained in the process
simulation. However, there is often some work to do to prove it. In most cases, the
pressure drop in the process simulation is just a guess, and often a very conservative
one. In contrast to this, the heat exchanger design program calculates this pressure
drop properly. In most cases it will be lower than the set one assumed in the process
simulation. Therefore, due to higher outlet pressure the condensed fraction will be
larger than in the process simulation, and so is the duty. The simple solution is to
repeat the process simulation for the condenser setting the pressure drop according
to the one obtained with the heat exchanger design program.
4.4 Condensers | 121
From the constructive point of view, the condensate removal from the heat ex-
changer should be well defined. As mentioned above, the parallel baffle cut has to be
preferred. The condensate removal can be supported by an inclination (usually 1–2°),
where the nozzle for the condensate removal is the lowest point. In certain cases, the
condensate level in the heat exchanger can even be used for controlling the heat duty.
The higher the condensate level, the more tubes are flooded, and the less heat transfer
area can be used. At the vapor inlet, the tubes must often be protected against erosion
by droplets. For this purpose, impingement plates (Figure 4.31) are often used.
A special design is very useful if vacuum vapors have to be condensed. In normal
arrangements like BEM the calculated pressure drop is often larger than the pressure
itself, which is impossible. Increasing the shell diameter leads to huge apparatuses.
The BJ21T arrangement (Figure 4.16) enables condensation with a very low pressure
drop. The two large inlet nozzles guarantee a predistribution of the vapor at low ve-
locities, and vapor can reach the tubes easily.
0.5 m
Fig. 4.16: The BJ21T arrangement.

The dimensioning of a condenser has a special pitfall if inert gases are involved, es-
pecially when small amounts of condensables have to be removed from an inert gas
stream, e.g. waste air. There is a large difference to the normal case without inert gases,
where there is hardly any transport resistance for the vapor to get in contact with the
cold surface or, respectively, the boundary layer. If there are large amounts of inert
gases, the condensables must get to the boundary layer by means of diffusion, which
is usually slow and which must be regarded as the step determining the condensation
rate. Mass and heat transfer have a mutual influence on each other [78]. The state-of-
the-art calculation of this combined heat and mass transfer is thoroughly described in
[79]. However, this procedure is too complex for an application in a multicomponent
mixture. Simplified approaches [80] are often used in commercial heat exchanger de-
sign programs, but in general it must be recommended to take care when the heat
exchanger is designed.
Especially for condensers, the deaeration problem is a decisive issue. There must
be a clear route for defined and undefined (leakage) inert gas flows to leave the appara-
tus. For condensers, inert gases would significantly lower the heat transfer coefficient,
as described in the last paragraph. For each heat exchanger, the design must be indi-
vidually checked whether noncondensables are adequately removed. It should also be
verified that the air which is in the exchanger at the beginning can be removed during
operation.
4.5 Evaporators
The design of normal heat exchangers like condensers or liquid-liquid heat exchang-
ers can be regarded as a standard task in process engineering, whereas the design
of evaporators is usually not. Before one can encounter the particular difficulties in
calculation, one must choose the type of evaporator with respect to the properties
of the stream. Most evaporators are used as reboilers for column service; therefore,
this special arrangement shall be carefully considered. The most popular reboiler for
distillation columns is the thermosiphon reboiler with natural circulation.
Figure 4.17 shows the bottom of a column with a thermosiphon reboiler. On the
left hand side there is the removal of the bottom product, which has the same con-
centration as the circulation flow. The circulation flow enters the heat exchanger at
the bottom. Due to the height difference between the surface of the bottom liquid and
the inlet of the tube bundle (“static liquid head”) the pressure of the liquid is higher
than the saturation pressure of the liquid at the surface, i.e. the liquid is subcooled.
Inside the tubes, the liquid rises again. The pressure of the liquid decreases and it is
heated by the heating agent on the shell side, usually steam. Both effects compensate
the subcooling after a certain height (preheating zone) has been passed, and boiling
of the fluid begins. First, bubbles are formed, which become more and more numer-
ous. A distinctive two-phase flow leaves the heat exchanger through the outlet nozzle
and enters the column again, where it is split into a vapor flow going up the column
and a liquid flow going back to the bottom, where the cycle begins once again. The
driving force of the circulation flow is the density difference between the left leg of the
reboiler cycle, where there is only liquid, and the right one, where a two-phase flow
of the bottom product occurs. In the latter, the overall density is smaller due to the
bubbles, and therefore, the static pressure ρgH is larger in the left leg than in the right
one, causing the natural circulation without external stimulation.
The temperature difference between the heating agent and the product is the driv-
ing force for the heat exchange. For each evaporator type, there is a reasonable range.
For thermosiphon reboilers, this range is between 15 and 40 K. At lower temperature
Two-phase
Vapor flow
Liquid
Bottoms Boiling
Static zone
liquid
head
Start of
Calandria boiling
Preheating
Bottoms zone
product
Subcooled
liquid
Clear liquid
Fig. 4.17: Thermosiphon reboiler with natural circulation.
differences, especially below 10 K, circulation instabilities are likely to occur [81–84].

In these cases, only a few bubbles are formed, and a significant carry-over of liq-
uid does not take place. The circulation is low, and so is the heat transfer. There is
a large preheating zone. Periodically, these bubbles coalesce, and the plugs formed
push greater amounts of liquid upward, causing an increased circulation for a short
time (geysering). Small temperature differences are likely to occur if the reboiler has
an overdesign which is too large. According to
Q̇ = kA∆T (4.6)
and assuming that k remains in the same order of magnitude, a large heat transfer
area A means that a small driving temperature difference ∆T is formed, as usually
some process control valve maintains the duty Q̇ by adjusting the steam amount. For
a large area A, the valve will reduce the steam pressure on the shell side of the re-
boiler, and the condensation temperature will become lower and closer to the product
temperature. Another case is the startup phase. At the beginning, there is no fouling;
therefore, the heat transfer coefficient k is larger than expected (Chapter 4.9), and the
∆T gets smaller.
High driving temperature differences are just as undesirable as low ones either.
The natural circulation is then reduced by the rising pressure drop. At very high driv-
ing temperature differences there is the risk of burnout, i.e. the occurrence of film
boiling, where the heat transfer to the vapor is much lower than to the bubble flow. A
temperature difference rising beyond a critical value will lead to a lower vapor gener-
ation.
The residence time of the product in the bottom and in the reboiler is quite long,
and therefore the thermosiphon reboiler is not really gentle towards the product.
There is another reason why thermosiphon reboilers are not the first choice when
temperature-sensitive substances are involved. Usually the boiling points of these
components are relatively high, and the distillation is carried out in a vacuum to
avoid high temperatures at the bottom. In fact, the thermosiphon reboiler has prob-
lems in vacuum operation, especially at pressures below p = 150–200 mbar [85]. This
is illustrated in Figure 4.17. The static liquid head is the driving force for the ther-
mosiphon circulation. However, as mentioned above, due to hydrostatics the liquid is
subcooled at the entrance of the tubes. In a vacuum, the preheating zone in the tubes
with low heat transfer will be much larger, and at a certain point the thermosiphon
circulation vanishes due to the low driving force. The following example illustrates
the impact of the vacuum.
Example
Compare the subcooling of a thermosiphon reboiler with water as the bottom product at different
bottom pressures p1 = 1 bar and p2 = 100 mbar. The static liquid head is assumed to be H = 5 m. For
the density of water, ρ = 1000 kg/m3 is used in both cases. For simplification, the gravity acceleration
is 10 m/s2 .
Solution
At p1 = 1 bar, the bottom temperature can be calculated to be t s1 = 99.6 °C [29]. Following hydro-
statics, the pressure at the tube inlet of the reboiler will be p1,tube = p1 + ρgH = 1.5 bar. This in turn
corresponds to a boiling temperature of t s1,tube = 111.3 °C. The subcooling is approx. 12 K.
At p2 = 100 mbar, the bottom temperature can be calculated to be t s2 = 45.8 °C. Considering
hydrostatics, the pressure at the tube inlet of the reboiler is p2,tube = 0.6 bar, corresponding to a
boiling temperature of t s2,tube = 85.9 °C. The subcooling is considerably higher in this case, approx.
40 K. It can easily be guessed that the preheating zone will be much larger in the vacuum case. This will
result in a worse heat transfer, giving less circulation and, in a vicious circle, a larger preheating zone.
For this reason, it is recommended that thermosiphon reboilers should not be used below p < 0.2 bar
[85]. The tube length of thermosiphon reboilers in vacuum should be approximately 3 m.
The static liquid head mentioned in the example is one of the key quantities for the
function of the reboiler. It has to be defined for the heat transfer calculation, and it
should be indicated on the arrangement sketch for column and reboiler, although its
realization is not a matter of construction but of the level control. It has two effects
which partly compensate each other. The higher the static liquid head, the larger is
the preheating zone, which reduces the heat transfer, and the larger is the circulation,
which increases the heat transfer. Often, these two effects compensate [82]. A reason-
able value for the static liquid head is 90 % of the tube length (usually in the range
3–6 m).
Furthermore, the connections between the distillation column and the reboiler
have to be defined, i.e. length and diameter of the reboiler feed and outlet lines. For
the diameters, useful rules of thumb exist. The cross-flow area of the outlet line should
be approx. 80–100 % of the total cross-flow area of the tubes of the reboiler, whereas
25–50 % are sufficient for the inlet line. To avoid extremely large throughputs and low
evaporation rates, a throttle valve can be placed in the inlet line. The height of the rear
end head should be approx. 25 % of the shell diameter if it has an axial nozzle; for a
radial nozzle, it is recommended to add the nozzle diameter.
Finally, for the specification of a thermosiphon reboiler the definition of the pro-
cess is different from the one for a single-phase heat exchanger or for a condenser.
One should be aware that the flow rate of the product is not known, as its depends on
arrangement and construction of the equipment. Moreover, its determination by any
calculation program will be more a number for the order of magnitude than an exact
value. Instead, a reboiler is specified by the heat duty to be transferred. Usually, an
estimation for the outlet vapor fraction is required. A standard value is 0.2, an accept-
able range for the result is 0.15–0.25, for water, evaporation rates down to 0.05 can be
accepted. It is worth mentioning that the inlet pressure of a thermosiphon reboiler is
not a useful quantity, as it can hardly be determined. Instead, it makes sense to start
the calculation with the pressure at the surface of the liquid in the bottom of the dis-
tillation column. The local pressures obtained due to the interaction of static height
and pressure drops are then evaluated by the program. Table 4.1 gives a summary of
the mentioned design recommendations for thermosiphon reboilers.
Concerning the accuracy of the calculation of a thermosiphon reboiler, one must
be aware that the mutual interaction between heat transfer and the two-phase flow in
the tubes is really complex. The accuracy of the pressure drop calculation of the two-
Table 4.1: Recommended values for the design of thermosiphon reboilers.
Item Recommendation
Tube length 3–6 m (vacuum: 3 m)
Static liquid head 90 % of tube length
Cross-flow area, outlet line 80–100 % of tube cross-flow area
Cross-flow area, inlet line 25–50 % of tube cross-flow area
Height of rear head 25 % of the shell diameter
(for radial nozzle: + nozzle diameter)
Outlet vapor fraction 0.15–0.25, water: > 0.05
Driving temperature difference 15–40 K
Pressure range p > 200 mbar
phase flow (Chapter 12.1) in changing flow regimes determines the circulation rate,
which is in turn decisive for the heat transfer and the vapor generation in the tubes. It
is not probable that these dependencies can be accurately determined for both organic
and aqueous fluids in various arrangements. Nevertheless, the considerable success
of heat exchanger design programs indicate that at least the overall performance of
thermosiphon reboilers, i.e. the duty transferred, is predicted in a way that a reliable
design of commercial reboilers is possible.
There are a number of other types of evaporators. For vacuum applications, the
falling film evaporator is the usual alternative. Falling film evaporators (Figure 4.18)
consist of a vertical tube bundle. Usually, falling film evaporators are effective if the
tubes are long, often 8–9 m. The liquid to be evaporated is fed at the top and flows
down the tubes as a thin film due to gravity. Special distributors on the top of the
tubes ensure that there is an even distribution of the liquid into the tubes. On the
shell side, the heating agent, usually steam, is condensed. The vapor is generated in
the tubes and goes down in cocurrent flow with the liquid, supporting the downflow
of the liquid due to shear forces. Vapor and liquid are separated at the bottom in a
separator vessel.
Product
Distributor
Steam
Calandria
Separator
Condensate
Separator duct
Calandria base
Fig. 4.18: Falling film evaporator.

Concentrate Courtesy of GEA Group AG.
In a falling film evaporator, a gentle evaporation takes place. The residence time in the
heated zone is short, the temperatures can be kept low, as it can be operated in the vac-
uum, and the temperature differences between heating agent and product can be kept
low as well (usually 8–20 K). There is a small liquid holdup, giving quick reaction times
on changes of the operation conditions. Furthermore, in contrast to thermosiphon and
forced circulation reboilers (see below) the falling film evaporator is pretty insensitive
against foaming.
The k-values of falling film evaporators is in the range k = 700–1200 W/(m2 K),
where the main heat transfer resistance is usually caused by the heat conduction
through the film. Care must be taken that the heat flux does not exceed a critical value.
Otherwise there is the danger that the film will dry out locally and a hot spot be formed.
Another countermeasure is to operate with a liquid recycle so that the film thickness
increases due to the increased mass flow. For the so-called coverage, defined as the
ratio between liquid volume flow rate and total wetted circumference of all tubes, a
range of 1.2–1.5 m3 /(m h) is a good approach.
Falling film evaporators can be used up to viscosities of 500 cP, although one can-
not expect Newtonian behavior in this range, and the performance at high viscosities
will certainly be lower. The limitation for the application is fouling. Due to the slow
gravity flow of the liquid, there is no abrasive effect.
The specification of a falling film evaporator in a heat exchanger design program
has some peculiarities. In contrast to the thermosiphon reboiler, the inlet flow into the
heat exchanger at the top of the calandria is quite well defined, either by the process
for the single-pass option or by the recirculation pump capability. During the speci-
fication, the coverage should be checked. Defining the inlet pressure on the product
side of the falling film evaporator is unpleasant. In fact, only the outlet pressure is
well defined. The inlet pressure is a result of the pressure generated by the pump, the
static height of the tube inlet and the pressure drop of the distributor. Especially in the
vacuum operation, the performance of the falling film evaporator is quite sensitive to
the inlet pressure, and therefore its calculation does not make much sense, especially
because the exact performance data of the pump is barely known. It is reasonable to
determine it iteratively. With a reasonable guess of the inlet pressure, the design pro-
gram evaluates an outlet pressure which should match the given value. If this is not
the case, the inlet pressure is varied according to the difference of the calculated and
the known outlet pressure until the outlet pressure fits.
Another alternative to the thermosiphon reboiler is the forced circulation reboiler
(Figure 4.19). In fact, this is a liquid heater. The heated liquid is then expanded into
the recipient, usually a column, through a valve, causing the evaporation. The heat
transferred in the heater becomes sensible heat according to
Q̇ = mc
̇ p ∆T . (4.7)
The larger m,̇ the lower is ∆T, and the lower are the thermal stability problems. On
the other hand, a strong pump is necessary for large volume flows and relatively low
pressure differences. The forced circulation reboiler is relatively insensitive against
fouling, as the evaporation takes place outside the apparatus. Due to the relatively
high velocities in the tubes (1.5–2 m/s), there is an abrasive effect that prevents the start
of fouling. The high velocities on the product side are also related to a high through-
put, which in turn causes a low temperature change and therefore less thermal stress.
However, the forced circulation reboiler is very sensitive against foaming.
The kettle reboiler is a simple and robust alternative to the thermosiphon reboiler,
both for vacuum and pressure applications (Figure 4.20). It causes no vibration prob-
lems, and evaporation rates up to 80 % are possible. Its drawback are possible en-
Vapor
Gravity separator
Calandria
Steam
Concentrate
Fig. 4.19: Horizontal forced circulation

Condensate evaporator.
Product Courtesy of GEA Group AG.
Fig. 4.20: Sketch of a kettle reboiler.

trainment and its affinity to fouling, as there is no defined flow which can remove the
dirt, and heavy boiling substances have a long residence time in the heat exchanger.
It is a relatively expensive piece of equipment.
For the design, it has to be taken into account that the kettle reboiler is effective
as an evaporator, not as a liquid heater. The heat transfer takes place due to bub-
ble formation. Feeding subcooled liquids must be avoided [86], as in this case the
dominating heat transfer mechanism is natural convection with low velocities and
without the support of baffles, giving a low k-value. To enable the formation of bub-
bles, a minimum temperature difference between heating agent and product should be
maintained, at least 12–15 K. For high pressure applications, the driving temperature
difference can be lower.
Wiper
Feed basket
Condenser
Evaporator
Heating
jacket
distillate
Residue
Fig. 4.21: Sketch of a thin film evaporator.
Although they do not belong to the shell-and-tube heat exchangers, two other evapo-
rators should be mentioned which have the main purpose of gentle evaporation and
product conservation. In thin film evaporators (Figure 4.21), the product is distributed
on the inner side of a tube with a heating jacket outside. It forms a film. Inside the tube,
a drive shaft with an attached wiper rotates and keeps the film thickness constant. The
residence time in thin film evaporators is normally less than 1 min. They are appropri-
ate for low pressures down to 1 mbar. Pressures below 1 mbar are not possible because
of the pressure drop caused by the transport from the evaporator to the condenser.
Extremely high heat transfer coefficients are possible. However, the prediction capa-
bilities are low. The design should be performed by a vendor who has carried out a
pilot trial with a reasonable scale-up. A classical failure is the use of laboratory data
for the determination of the k-value; these laboratory data have usually been obtained
in an equipment made of glass, where the low thermal conductivity of the glass deter-
mines the heat transfer. For further information, [85] is a good starting point.
For applications in rough vacuum (p = 1–10−3 mbar) and high vacuum (p =
10 –10−7 mbar) short path evaporators (Figure 4.22) are used for extremely high
−3
boiling substances. The principle is to keep the distance between evaporator and
condenser as short as possible, usually only a few cm. Because of the low pressures,
the temperatures can be kept low as well, and the evaporation is extremely gentle
towards the product. On the other hand, the smaller the distance between evaporator
and condenser is, the greater is the danger of entrainment. The design should again
be performed by experienced vendors. It is essential that low boiling substances are
Feed
Distribution
plate
Heating
Wiper agent, out
system
Inner
condenser
Heating
agent, in
Residue Vacuum
Cooling
medium
Distillate Fig. 4.22: Sketch of a short path evaporator.
completely removed before. From gas theory, the Langmuir–Knudsen equation gives
an upper limit for the evaporation capacity per heating area [87]:
̇
m/A p √ M T −1
( 2
) = 1575 ( ) . (4.8)
kg/(m h) max mbar g/mol K
For p = 10−3 mbar, the order of magnitude is
̇
(m/A)max = 1.5 kg/(m h) .
2
(4.9)
4.6 Plate heat exchangers
Plate heat exchangers are an option for combining very high heat transfer coefficients
with a large heat transfer area per volume. They consist of profiled heat exchanger
plates which separate the two media. When they are assembled, the profiles form
many parallel connected channels which form the heat transfer area. These chan-
nels change their direction regularly, and the flows are directed in a way that the gaps
filled with cold and warm medium alternate (Figure 4.23). The corrugation of the plate
profiles increases the turbulence of the flows and therefore improves the overall heat
transfer coefficient k.
There are further advantages of plate heat exchangers. As can be seen in Fig-
ure 4.23, the plates are mounted in a skid between two massive plates. This arrange-
ment is easy to dissemble, giving the opportunity of an easy cleaning procedure. De-
sign corrections or capacity increases can easily be achieved by adding more plates to
the skid.
4.6 Plate heat exchangers | 131
Fig. 4.23: Principle sketch of a plate heat exchanger.

Courtesy of Alfa Laval Mid Europe GmbH (Germany).
A serious problem are the gaskets between the plates. For aggressive media and many
organic liquids the plates must often be soldered or welded, where the strong advan-
tage of the easy cleaning opportunity gets lost again.
The k-values of plate heat exchangers can be extremely high. They can also be
used for highly viscous media, and they show less fouling due to high velocities. Foul-
ing factors used for shell-and-tube heat exchangers are generally too high for plate
heat exchangers. The disadvantages are the high maintenance costs for the gaskets
and the permanent risk of leakage, or, alternatively, the impossibility of cleaning.
Plate heat exchangers are not only used as liquid-liquid heat exchangers. Mean-
while, they are increasingly used as evaporators and condensers as well. Plate
condensers have the advantage that asymmetric channels can be formed; wide
ones for the vapor side and narrow ones for the cooling water to maintain an ap-
propriate velocity to achieve enough turbulence (Figure 4.24). The same advantage
can be be claimed for evaporators, where even high-viscosity media can be han-
dled. Driving temperature differences of only 3–4 K can be taken into consideration,
which is especially important if mechanical (Chapter 8.2) or thermal (Chapter 8.3)
vapor recompression or special materials are involved. Another advantage is the
small holdup, giving short startup and shutdown phases and performing a gentle
evaporation of temperature-sensitive substances. Several evaporation options can be
realized, e.g. the thermosiphon reboiler, the single pass evaporator, the falling film
evaporator (Figure 4.25) or the forced circulation evaporator.
The design calculations of plate heat exchangers is usually supported by the var-
ious commercial design programs. However, the particularities and the design rules
are not as common as they are for shell-and-tube heat exchangers so that large differ-
ences between a quick program call and a professional design performed by a vendor
Fig. 4.24: Asymmetric channels for condensation in plate heat exchangers.

Courtesy of Alfa Laval Mid Europe GmbH (Germany)
Vapor outlet
Feed inlet Demister
Steam inlet
Plates
Condensate
outlet
Fig. 4.25: Plate heat exchanger

used as falling film evaporator.
might occur. The commercial programs enable even a nonprofessional to point out the
advantages of plate heat exchangers if their application is possible. The principles of
the heat transfer calculation are well described in [88].
4.7 Double pipes
The double pipe is the simplest construction for a heat exchanger. This is just two
concentrical pipes, where one stream is in the inner tube and the other one in the
ring.
4.8 Air coolers | 133
Fig. 4.26: Double-pipe heat exchanger [89].

© Wiley-VCH Verlag GmbH & Co. KGaA. Repro-
duced with permission.
Generally, double pipes suffer from the fact that the heat transfer area remains low, so
that large duties can hardly be realized. Instead, double pipes are often used for heat
tracing to compensate for unintended heat losses.
4.8 Air coolers
If the heat transfer on one side is significantly worse than on the other side, it is deter-
mined solely by the bad side. In this case, it makes sense to increase the heat transfer
area selectively on this side. An example is the air cooler. The heat transfer on the prod-
uct side is much better than the one to the environmental air. Therefore, the tubes are
equipped with fins (Figure 4.27), and the heat transfer is improved by blowers which
create forced flow with comparably high velocities (Figure 4.28). The heat transfer to
the fins takes place by heat conduction. Calculation procedures for this problem are
described in [71] and [90].
The investment costs for an air cooler are higher than for a conventional shell-and-
tube water cooler because of the large heat transfer area and the costs for the blowers
including their drives. On the other hand, operation costs are lower, as there is no
cooling water consumption. Also, no piping is necessary on the service side, and no
fouling takes place. There are certain other aspects for the assessment of air coolers,
e.g. their high noise level or their large space demand. Process control is difficult if
the air cooler is exposed to rain, snow or sun radiation. But the main criterion is the
fouling aspect: Cooling water contains hardness components (Chapter 13.3), which
might precipitate at wall temperatures higher than 65–75 °C, and as long as no other
measures are taken, the air cooler should be taken into account in these cases.
The final design of an air cooler should be performed by a vendor.
Fig. 4.27: Finned tubes.

Courtesy of Kelvion (www.kelvion.com).
Fan
Finned tube bundle Fan ring
Heating coils Air plenum
Louvers
Pulley
Drive belt
Motor
Supporting structure
Fig. 4.28: Air cooler.

Courtesy of Kelvion (www.kelvion.com).
4.9 Fouling
If you consider fouling, you will get fouling.

(Andreas Doll)
In general, heat exchangers suffer from a gradual deterioration of the heat transfer
caused by so-called fouling. The streams often contain dissolved or suspended ma-
terials which may deposit on the surface of the heat transfer areas and form a layer.
These layers usually have a low thermal conductivity; therefore, they form an extra
thermal heat transfer resistance which lowers the overall heat transition coefficient.
Especially, streams containing hardness components are prone to the formation of
4.9 Fouling | 135
fouling layers. The solubility of hardness components becomes lower with increasing
temperature; their deposition is a typical case for fouling.
Formally, this heat transfer resistance can be calculated by taking into account
the thickness of the layer and its thermal conductivity. However, neither of them can
be determined in advance. Therefore, the additional fouling resistances used (untruly
called “fouling factors”) are more or less just set by the user, in most cases according to
the in-house design rules of each particular company. Fouling factors are in the range
0–600 ⋅ 10−6 m2 K/W. They can be interpreted according to Table 4.2.
Table 4.2: Typical fouling factors.
Fouling Factors Interpretation Example

10−6 m2 K/W
0 no fouling caustic soda

100 formal consideration of fouling steam
200 low fouling overhead products
300–400 moderate to strong fouling cooling water
500 strong fouling dirty products
600 very strong fouling very dirty products
A collection of fouling factors can be found in [91]. Shell-and-tube fouling factors can-
not be directly applied in the plate heat exchanger design. In plate heat exchangers,
the velocities are considerably higher so that less fouling occurs.
Nevertheless, the fouling consideration is often subject to discussion. In [91] it is
argued that the consideration of fouling factors often leads to heat exchangers which
are too large, and therefore the velocities are lower. High velocities counteract foul-
ing because of possible abrasion of the fouling layer. Therefore, it often happens that
fouling only occurs because fouling had been considered in the design. In [91] it is rec-
ommended that fouling should not be considered. A usual safety margin (15–20 %),
an appropriate design with a large B fraction (> 65 %), and a reasonable baffle cut
(20–25 %) should ensure that fouling is avoided. Auxiliary measures like a small par-
allel heat exchanger to compensate fouling if it occurs nonetheless or recycling of part
of the cooling water return to keep a sufficient velocity at plant startup when there is
no fouling could also help. A similar recommendation is given in [92].
In practice, the situation of an engineer who follows this argumentation is not
easy. There is no obvious success story when fouling is avoided by setting the fouling
factor to zero, as there is no proof. On the other hand, it is a serious design mistake if
the heat transfer area is too small due to fouling when no fouling factor has been con-
sidered. One should at least be skeptical if fouling factors are arbitrarily increased or
considered, although the medium is known to be clean. When in doubt, there should
be a tendency towards the lower fouling factor. Traditions in plant engineering are
difficult to overcome.
As mentioned, the heat exchanger design programs outline how the overall heat
transfer resistance is composed. When the heat transfer conditions are very good on
both the product and the service side, the fouling factors can account for a large per-
centage of the heat transfer resistance. Often, this is the case for thermosiphon reboil-
ers with steam as heating agent and water on the product side. In these cases, the
size of the heat exchanger is determined by the more or less arbitrary fouling factors.
Special care must be taken. If a thermosiphon reboiler is too large, it might not work
properly (see above). Without experience, heat exchangers like this are very difficult
to design.
4.10 Vibrations
Dark Arts is a mandatory subject.

(Harry Potter and the Goblet of Fire)
Tube vibrations can cause severe mechanical damage to a heat exchanger. They are
induced by the shellside flow across the tube bundle.
There are several mechanisms which can create vibrations; the most important
ones are fluidelastic instability and vortex shedding.
Due to its pressure, the fluid exerts a certain force on the tubes, pushing them
apart. On the other hand, the tubes act like a spring, and a reset force is built up.
Fluid force, reset force, and damping of the fluid form an oscillating system. If the am-
plitudes exceed a certain level, severe mechanical damage can be the consequence
(Figure 4.29). This phenomenon is called fluidelastic instability. Vortex shedding is
caused by periodic formation of vortices in the tube bundle (Figure 4.30). Short-term
failures can take place if the frequency of the vortex formation approaches the reso-
nance frequency of the tubes. Regions which are prone to vibration damage are [93]:
– tubes with large unsupported spans between two baffles;
– tubes located in baffle window region at the tube bundle periphery;
– U-bend regions;
– tubes beneath the inlet nozzle;
– tubes in tube bundle bypass area.
4.10 Vibrations | 137
Fig. 4.29: Tube failure caused by fluidelastic

instability [93].
© Hydrocarbon Processing.
Fig. 4.30: Vortex shedding in tube array [93].

The calculation of tube vibration phenomena is not very well founded. There are some
rules and plausibility considerations that should be taken into account. Many param-
eters can affect tube vibrations, and most of them also affect the thermal performance.
The proper support of the tubes is essential in avoiding vibrations [93]. If the unsup-
ported span of the tubes is long, its natural frequency is low and resonance phenom-
ena are more probable. Tubes supported by baffles have an unsupported span equal to
the baffle spacing. Tubes in the baffle window have much larger unsupported spans
and are therefore more susceptible to vibrations. Reducing the spacing L of single-
segmental baffles increases the cross-flow velocity proportional to L−1 , while the nat-
ural frequency increases proportional to L−2 so that resonance becomes less probable.
Also, other baffle types can be tried (Figure 4.12). In heat exchangers with double-
segmental baffles, the cross-flow velocities are much lower. “No-tubes-in-window”
baffles (NTIW) are usually vibration-free, as the unsupported span of the tubes is re-
duced by 50 % for the same number of cross-passes. However, as the tubes in the
window region are omitted, they require a larger shell diameter to maintain the heat
transfer area. In a rod-baffled exchanger (Figure 4.13), the shell-side flow is parallel to
the tubes, and vibrations are hardly possible.
Tubes at entrance and exit areas are exposed to the highest velocities and there-
fore susceptible to vibrations. To avoid damage, clearances or impingement plates can
be provided (Figure 4.31). The tube pitch and the layout pattern can be varied to avoid
vibration. The larger the tube pitch, the lower are the cross-flow velocities, and flow-
induced vibrations become less probable. However, the heat transfer deteriorates, and
a larger diameter has to be chosen. For avoiding vortex shedding, it often proves to be
Height under inlet nozzle

Impingement plate 154,053 mm
Fig. 4.31: Clearance under inlet nozzle

and impingement plate.
205,004 mm Courtesy of Heat Transfer Research, Inc.
successful to change the angle of the tube arrangement (60° pattern in Figure 4.8). An
angle of 45° can be tried to avoid fluidelastic instability.
Finally, an increase of the stiffness of the tubes can be tested, e.g. by choosing a
material with a higher elastic modulus or by choosing a larger tube diameter or wall
thickness. However, usually these items are in most cases determined by corrosivity
issues or by the heat transfer itself, respectively [93].
Using commercial programs for heat exchanger design, empirical criteria must be
applied to estimate whether vibrations are probable. For fluidelastic instability, the
vibrations increase with increasing velocity on the shell side. One must take care to
stay below 80 % of the critical velocity in the whole shell side. The most critical loca-
tions are the ones where the velocity is high, e.g. at the inlet nozzles or at the bundle
entrance. To avoid fluidelastic instability, any measure of reducing the shellside ve-
locity might be useful, e.g. increasing the shell diameter, choosing larger nozzles or
increasing the pitch. It can also be recommended to increase the clearance between
nozzle and tube bundle. Also, increasing the natural frequency by reduction of the
baffle spacing or the change of the baffle type is another promising option. To avoid
vortex shedding, one must stay away from the resonance frequency. Besides reducing
the velocities, the change of the tube arrangement has often proved to be successful.
A detailed description of the tube vibration phenomena is given in [93] and [94].
5 Distillation and absorption
The most important thermal separation process in technical applications is distilla-
tion. The name often causes some confusion. In basic lectures, “distillation” means
a single stage comprising evaporation and condensation. The multiple arrangement
of separation stages in a column is called “rectification”. Although this is sometimes
used, the colloquial term used in industry for such a “rectification” is in fact “distilla-
tion”, which is used in this book as well. The reason for the wide spread of distillation
is the use of simple heat as a utility which can easily be added and removed. The phase
equilibrium between vapor and liquid is the foundation of distillation, where the den-
sity difference between the phases is so large that their separation is relatively easy.
Distillation can be explained according to the following scheme.
A simple separation stage consisting of evaporation and condensation achieves
an enrichment of the low-boiling substances in the condensate, but according to the
vapor-liquid equilibrium all components occur in both phases. Generally, pure com-
ponents are not obtained. Several stages are necessary for further purification. Fig-
ure 5.1 shows an example for such an arrangement. Considering a binary mixture, in
the upper part the low-boiler is purified by repeatedly sending the overhead stream
to a further separation stage. In principle, one can in fact obtain the low-boiling com-
ponent in an arbitrarily specified concentration, and, accordingly, the heavy end as
well, as shown in the lower part of Figure 5.1. However, the drawback of this process is
obvious: Only small amounts of both light and heavy ends are obtained, as no use is
Cooling .
water VN
Cooling .
water VN–1
Cooling . . N
water VJ+1 VN–2
. N–1 .
VJ LN
J+1 . .
LJ+1 LN–1
Feed .
J VJ–1
.
Steam V2
. J–1
LJ .
V1
Steam . . 2
LJ–1 L3
. 1
Steam L2
Steam .
L1
Fig. 5.1: Series of separation stages.

140 | 5 Distillation and absorption
Cooling .
VN
water
.
VN–1
. N
VN–2
N–1 .
. LN
VJ+1
N–2 .
. LN–1
VJ
J+1 .
. LN–2
LJ+1
J
.
VJ–1 .
. V3
LJ J–1
.
. V2
LJ–1 3
.
. V1
L3 2
.
L2 1
.
Steam L1
Fig. 5.2: The prestage of a distillation column.
made of the intermediate fractions where a number of separation stages have already
been applied. This situation can be improved if these particular fractions are led to the
next stage above or, respectively, below. Both condensers and evaporators can be left
out if the vapors are directly forwarded to the next stage without condensing. Only at
the ends of the sequence an evaporator and a condenser are necessary (Figure 5.2). Va-
por and liquid are moving in countercurrent flow. A vertical arrangement of the stages
then leads to the well-known distillation columns. The principle of countercurrent
flow can be applied to more or less all thermal separation processes like extraction or
even adsorption (“simulated moving bed”, Chapter 7.2).
Figure 5.3 shows the typical terms describing a distillation column. The mixture
to be separated is continuously led into the column. It is called “feed”. Several feeds
are possible. The lower end of the column is called the “bottom”. At the bottom, the
reboiler generates vapor to provide the necessary heat input into the column. At the
upper end of the column (“top”), the overhead stream is led into a condenser, which
removes heat from the column at a lower temperature level. Part of the condensate
is led back into the column (“reflux”), which forms the essential liquid flow from the
5 Distillation and absorption | 141
Condenser
(cooling water)
Distillate
Reflux
Feed
Side draw
Add. feed (optional)
(optional)
Reboiler
(steam)
Fig. 5.3: Distillation column
Bottoms and the most important terms.
top to the bottom of the column, giving countercurrent flow with the vapor. The rest
of the condensate is removed from the column as a product, the “distillate”. The ratio
between reflux flow Ṙ and distillate flow Ḋ is called the reflux ratio ν:
ν = R/̇ Ḋ . (5.1)
The column part above the feed is called rectifying section, the part below the feed is
the stripping section. In the column there are internals which enable a good contact
and mass transfer between vapor and liquid phase.
Generally, there are two options to perform distillation and absorption processes:
packed columns and tray columns. In packed columns, there is a continuous mass
transfer along the column. Their advantage is the significantly lower pressure drop,
which is the decisive criterion in vacuum distillations. In tray columns the mass trans-
fer is performed stage-wise; their advantage is that they have no wetting problems and
that they are less sensitive against fouling. Very good monographs for distillation are
the books of Baerns et al. [8], Kister [95, 96], Sattler [97], and Stichlmair and Fair [98].
The material costs of a distillation column depend on the amount of material for
the column cylinder, which is proportional to both the diameter D and the height¹ H,
and on the costs for the packing or trays, which are proportional to H and to D2 . H is
determined by thermodynamics (number of separation stages), whereas D is deter-
mined by hydrodynamics (characteristics of the trays or the packing) and thermody-
namics (determination of the internal flows inside the column)².
Before starting, the influence of the pressure on a distillation column should be
clarified: The higher the column pressure is, the lower are the volume flows. Therefore,
higher pressure results in a higher capacity of the column. On the other hand, higher
pressure usually (not always) results in a worse separation behavior due to the phase
equilibrium, and therefore to a worse separation performance.
1 neglecting top and bottom.

2 Furthermore, the pressure has a significant influence on the wall thickness and, subsequently, on
the material costs.
5.1 Thermodynamics of distillation and absorption columns
There are two ways for the calculation of distillation columns. The equilibrium calcu-
lation uses the presumption of a theoretical stage, which represents full development
of the phase equilibrium on this stage. In practice, this assumption is not valid. A tray
does not represent a theoretical stage. One can introduce efficiencies (Chapter 5.3),
which are usually in the range of 23 . Alternatively, the number of stages is reduced,
e.g. 60 trays represent 40 theoretical stages. For packed columns, a certain packed
height is taken for one theoretical stage (HETP value, see Glossary).
The concept of the theoretical stage is very widely applied, but there are certain
constellations where it leads to qualitatively and quantitatively bad results. In these
cases, the mass transfer and the phase equilibrium area must be taken into account
by the calculation. Still, the phase equilibrium is most important, as it determines the
driving forces for the mass transfer. The application of rate-based models is obligatory
for the calculation of absorber columns, if high purities are required.
Distillate
Reflux
Feed Li–1 Vi Stage i
Vi+1 Li
Bottoms Fig. 5.4: Mass balance on a tray.
For an equilibrium column with theoretical stages, the determination of the necessary
number of stages and the reflux ratio are essential for the development of a distillation
process. With the concept of the theoretical stage, this can be achieved by solving the
so-called MESH equations (material balance, phase equilibrium, summation condi-
tion, heat balance). For a column with N stages and n components, these equations
are
– material balance for each component on each stage: n ⋅ N equations;
– phase equilibrium conditions for each component on each stage: n ⋅ N equations;
– summation conditions (∑ x i = 1, ∑ y i = 1) on each stage: 2N equations;
– heat balance on each stage: N equations.
5.1 Thermodynamics of distillation and absorption columns | 143
All N ⋅ (2n + 3) equations have to be solved. The corresponding unknowns are the
compositions x i and y i of each component (2n variables), the flows of liquid and vapor
and the temperature of the two phases in equilibrium (3 variables), all of them on
each stage (times N). For example, for a column with 60 stages and 20 components,
there are altogether 60 ⋅ (2 ⋅ 20 + 3) = 2580 equations, most of them nonlinear. If
chemical reactions occur (e.g. reactive distillation) additional equations would have
to be considered.
The mathematics of solving this system of equations are described in [8] and [95].
Modern process simulators offer very stable and well-established algorithms for the
solution. In case the column does not converge, it is often a trial-and-error procedure
to test the various options. The convergence history and the error messages can give
valuable information. The calculation of the column can only converge if there is both
vapor and liquid on each stage. If the error messages or the profile indicate that vapor
or liquid are missing on certain stages, one should change the specification in a way
that the amount of the missing phase on these stages is increased. A typical error in the
setup of a column is that the specified distillate or bottom flow is larger than the feed,
in this case, the algorithm has of course no chance to converge. If column convergence
does not work, it is often useful to change and simplify the column specification until
its calculation converges and a profile is available, even if it is a completely wrong one.
Slight variations of the specification towards the correct one often give a feeling about
the sensitivities, and with a valid column profile as starting point it is easier to achieve
convergence.
Often, a specification “out of balance” occurs. Consider a mixture of 500 kg/h of
component A (light end) and 500 kg/h of component B (heavy end). If e.g. the bottom
stream is defined to be 510 kg/h, one cannot expect it to be pure B, as it contains at
least 10 kg/h A, corresponding to approx. 2 %. On the other hand, at least 10 kg/h of
component A are lost. One must be aware that a variation of the reflux ratio or the
number of stages does not help at all in this case.
Essentially there are two ways for the representation of a distillation column in
the process simulator: the compact approach (Figure 5.5) and the detailed approach
(Figure 5.6). During process development, it is strongly recommended to use the com-
pact approach to reduce the effort when separation sequences are changed. When the
process is fixed, it makes sense to move over to the detailed approach, where the re-
sults for the streams of the whole condenser system are accessible more easily so that
they can directly be used for the design. The definition of so-called “pseudo-streams”
to extract this information is not necessary. Multiple condensers on different temper-
ature levels can easily be specified. Convergence is certainly a larger effort but usually
not more difficult. However, the reflux stream must be estimated first, otherwise the
upper stages might dry out, causing column calculation to stop. Another disadvantage
is that the reflux ratio cannot be entered directly; it must be converted into a split ratio.
DIST
FEED
BOT
Fig. 5.5: Compact column approach.

INERTS
OVERHEAD COND
REFLUX DIST
FEED
BOT
Fig. 5.6: Detailed column approach.

5.2 Packed columns
In packed columns, a packing volume distributes the liquid entering the bed at the top
and creates a large mass transfer area between the liquid and the vapor, which are in
countercurrent flow. Large packing volumina are divided into several beds. Between
the beds, the liquid is collected and redistributed by a collector/distributor unit. This is
necessary because the liquid tends to run together within the packing and concentrate
near the column wall. Figure 5.7 shows the principal constitution of a packed column.
Reflux from
condenser
Packing
Collector
Feed
Distributor
Vapor from
reboiler
Fig. 5.7: Constitution of a packed column.

Courtesy of Prof. Dr. J. Stichlmair.
Packed columns are mostly used at low pressures, where many systems exhibit larger
separation factors³, making a design with a lower tower height possible. Because of
the low pressure drop, a lower pressure can be achieved in the stripping section of the
column, and it is possible to limit the bottom temperature. For temperature-sensitive
substances, this is often important to avoid decomposition. Packed columns are sen-
sitive against fouling but less sensitive to foaming systems than tray columns (at least
random packings). Because of the low holdup, they are more sensitive to operation
changes. The separation efficiency depends on the specific surface, the kind of pack-
ing, the packing material and the kind of system. Packings can be made of metal,
plastic, glass, graphite, or ceramic.
3 The reason is that the ratio between the vapor pressures of the components is usually larger at low
temperatures.
There are two types of packed columns.

– Random packings:
The development of random packing elements during recent decades is charac-
terized by the aspiration to create structures that are more and more open (Fig-
ure 5.8). Packing elements of the 1st generation were spheres, cylindrical rings
(“Raschig rings”) or saddles. They were easy to manufacture, but their pressure
drop was still high. Furthermore, the packing elements of the 1st generation often
suffered from maldistribution, and their use was limited to small columns with
diameters below 500 mm. In the 2nd generation, the shell areas have been pene-
trated, which had already a drastic effect on the pressure drop. The most popular
packing element of this generation is the Pall ring, which is still in use. In the 1970s
and the 1980s, the packing elements of the 3rd generation came up, which only
consist of the framework, and due to the large free cross-section area their pres-
sure drop is even lower. The progress of this development is that the vapor load
at constant column diameter can be significantly higher for the modern packing
elements. Therefore, the change of the packing has become a standard option for
a capacity increase. The progress in the separation efficiency was comparably low.
The trend was continued in the 1990s with the development of the 4th generation
packing elements (Figure 5.9), with completely new shapes, giving extremely low
pressure drops and high flooding points (p. 148). It must be emphasized that the
importance of the distributor performance has increased, as the 4th generation
packing elements have hardly any self-distribution.
The larger the nominal size of a packing element is, the lower is its pressure drop,
but also its specific surface and therefore its separation efficiency. The nominal
size of a packing element should be lower than 10 % of the column diameter. Oth-
erwise there are too many empty volumes at the area near the wall, where the
liquid has no sufficient contact to the vapor phase.
Fig. 5.8: Raschig-Ring, Pall-Ring, and ENVIPAC: packing elements of the 1st, 2nd, and 3rd generation.
© Raschig GmbH, © ENVIMAC Engineering GmbH.
Fig. 5.9: Raschig-Super-Ring as an example

for a 4th generation packing element.
© Raschig GmbH.
Fig. 5.10: Sulzer Mellapak as an example

for structured packings.
© Sulzer Chemtech Ltd.
– Structured packings:
Structured packings have a regular geometry. They need appropriate distributors;
if this is ensured there is hardly any stream formation or wall effect. At low liquid
loads (< 20 m3 /(m2 h)) structured packings are more effective than random pack-
ings. The separation efficiency does not depend on the column diameter. Struc-
tured packings have a larger maximum load, a better efficiency and lower pressure
drops than random packings. An example is the Sulzer Mellapak (Figure 5.10).
At liquid loads < 10 m3 /(m2 h) wired packings (e.g. Montz A3, Sulzer BX) are an-
other alternative, having an even lower HETP⁴ value (0.1–0.2 m, i.e. 5–8 theoretical
stages per m) and even less pressure drop. On the other side, wired packings must
4 height equivalent of one theoretical plate, see Glossary.

be well wetted, which is often not the case for aqueous systems. Furthermore, they
are significantly more expensive and extremely sensitive against fouling. At higher
liquid loads, their application does not make sense; their advantages do not be-
come effective. It should be noted that structured packings are not more effective
in general. Random packing is clearly the better choice at high liquid loads and if
a larger holdup is required to generate residence time on the packing.
Packed columns can fail if the vapor load is too high or the liquid load is too low.
The upper limit for the vapor load is the flood point, where the countercurrent flow
between vapor and liquid breaks down. In this case, a froth layer is formed, the liquid
is accumulated and finally carried over to the top. The flood point decreases with in-
creasing liquid load, as the free cross-section area for the vapor flow is more and more
filled with liquid. The minimum liquid load is a less strict criterion. If it is not reached,
the wetting of the packing is so bad that the efficiency decreases significantly. As a rule
of thumb, for random packings liquid loads⁵ of 10 m3 /(m2 h) for aqueous systems and
5 m3 /(m2 h) for organic systems can be considered as limiting values. For structured
packings, nowadays very low liquid loads down to 0.2 m3 /(m2 h) can be realized. The
limiting factor is the quality of the distributor. One must always be aware that dis-
tributor and packing form a package which should be in the hand of one vendor. A
badly chosen distributor type can have a significant influence on the performance of
the packing [99].
For the design of packed columns their separation efficiency has to be regarded
first. Manufacturers usually give HETP values as a function of the vapor and the liquid
load. Often HETP also depends on the pressure and the kind of the system. Kister [100]
gives the following rules of thumb for estimating HETP.
– For random packings:
HETP = 93/a p . (5.2)
– For structured packings:
100
HETP = 0.1 m + , (5.3)
ap
with a p as the specific surface area.
Equation (5.2) is valid for modern (i.e. at least Pall rings) random packings with nom-
inal diameters of 1′′⁶ or larger. Unexpectedly, it has been found out that random pack-
ing elements smaller than 1′′ do not necessarily have a lower HETP value [100], proba-
bly due to maldistribution effects. Equation (5.3) is valid for Y-structured packings (e.g.
Sulzer Mellapak 250 Y, 45° inclination to the horizontal), whereas for X-structured
packings (60° inclination, e.g. Sulzer Mellapak 250 X) the resulting HETP should be
5 Definition in Equation (5.5)

6 ′′ = inch, 1′′ = 25.4 mm. Another abbreviation is “in”.
multiplied by 1.45. Both equations (5.2) and (5.3) hold for low surface tensions σ <
25 mN/m, which is typical for an organic system. In aqueous systems (σ ≈ 70 mN/m)
the results should be multiplied by 2, for amine and glycol systems (σ ≈ 40 mN/m)
by 1.5. The effect of pressure on packing efficiency is widely discussed but poorly un-
derstood, as maldistribution effects can always be involved as well. The prevailing
opinion [95] is that the effect of pressure is low at least at pressures p > 100 mbar.
Below this value, it is suspected that the efficiency decreases for random packings,
whereas for structured packings a decrease for high pressures (p > 15–20 bar) has
been observed.
With the HETP value, one can assign theoretical stages to the particular sections of
the column and simulate it with an equilibrium model. One should have in mind that
there are influences on the packing performance which are not covered by any calcu-
lation. The HETP value can be larger due to layers on the packing or bad wettability,
which is often the case for aqueous systems. After making a first guess, the results
should be discussed which the packing vendor to make sure that the required number
of stages per m is met. The HETP values are determined by means of test measure-
ments with an almost ideal, narrow-boiling binary mixture, e.g. chlorobenzene/ethyl
benzene (Figure 5.11), cyclohexane/n-heptane, or isobutane/n-butane [99, 100]. The
separations of these mixtures are sensitive to the number of stages so that it is not
difficult to evaluate.
68
67
66 p = 100 mbar
t / °C
65
64
63
0 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1
x,y chlorobenzene
Fig. 5.11: pxy diagram for the system chlorobenzene/ethyl benzene at p = 100 mbar.
From process simulation, one gets the liquid and vapor flows which encounter
each other between the stages, including their properties like density and viscosity.
They are the basis for the hydrodynamic calculation, which determines the column
diameter.
Usually, several load cases have to be regarded. The hydrodynamic considerations

are working with simplified physical models (channels, particles circulated around,
particle clusters). These models comprise equations for the flood point, the pressure
drop, and the holdup, i.e. the liquid content of the packing during operation. They
depend on each other in a complex way. Figure 5.13 illustrates the courses of pressure
drop and holdup as functions of vapor and liquid load. The vapor load is represented
by the so-called F-factor:
F = w√ρ , (5.4)
where w is the vapor velocity referring to the free cross-flow area and ρ is the vapor
density. It represents the square root of the kinetic energy of the vapor. Its unit Pa0.5
is usually omitted. A reasonable order of magnitude of the F-factor is F = 2. F = 0.5
would represent a relatively low vapor load, F = 3 a relatively high one. The liquid
load is the parameter in the set of curves. It is defined by
liquid volume flow
B= . (5.5)
cross-section area
Liquid loads can strongly differ; their range covers 100 m3 /(m2 h) as a very high load,
10–40 m3 /(m2 h) as “normal” liquid loads and 0.5–5 m3 /(m2 h) as low ones.
Fig. 5.12: Montz Type S distributor for very low liquid loads.
Courtesy of Julius Montz GmbH.
The larger the liquid load, the larger is also the holdup of the packing. The holdup
is the relative liquid content of the packing; it is a key quantity for the calculation
models. It causes the pressure drop to rise, as for the flow of the vapor less cross-
flow area is available. One can see that at the load point the drastic rise of pressure
drop and holdup starts (Figure 5.13). These courses are difficult to reproduce with the
calculation models. Therefore, they have a number of parameters adjustable to exper-
imental data of the packing. The most popular models are the one by Stichlmair [101],
0.3
0.2
m
wL = 80
h
0.1
H·ϘL·g
0.06 60
Δр
0.04
Pressure drop
40
0.02
20
0.01
10
0.006
0
0.004
0.003
0.2 0.4 0.6 10 2 4 m/s
Gas velocity wG
3·10–1
80
1·10–1 60
Liquid hold up h
2 40
2
20
2
10
2
2 wL = 5 m/h
2.8
1·10–2
0.1 0.2 0.4 0.6 0.8 1.0 1.5 2 3 m/s
Superficial gas velocity wG
Fig. 5.13: Course of pressure drop and holdup as a function of vapor and liquid load.
its further development by Engel [102] and the model of Billet and Schultes [103]. It
is important to note that the correlations used have limited physical value and are
optimized for use within the particular model. It makes no sense to mix them, e.g.
calculate the holdup according to Billet/Schultes and the pressure drop according to
Engel.
The hydrodynamic calculation starts with the thermodynamic calculation of the
column, which evaluates the liquid and vapor flows going from stage to stage. Using
these loads it is then checked whether a specified packing type fulfills a number of
criteria. The design criteria for a packing are as follows.
– Distance to flood point:
The flood point denotes the vapor load where the liquid is accumulated in the
packing and finally carried over the top. It depends on the liquid load. The par-
ticular calculation models [102, 103] have a built-in flood-point correlation. Their
accuracy can be estimated to be ± 30 %. The Kister–Gill correlation [95] might be
slightly more accurate but uses a packing-specific parameter. Therefore, the strat-
egy is to set the vapor load in the case with the maximum load to 70 % to be on the
safe side. One should try to get close to these 70 % and not to stay below and set
additional safety margins. Otherwise, the vapor load in the minimum case might
be too low. Furthermore, one should have in mind that the uncertainty of ± 30 %
could also mean that 130 % of the calculated value could be the actual flood point.
Therefore, it might happen that the packing does not perform very well if the load
is not adequate. Packing vendors have often more experience with the use of their
products and can sometimes take the responsibility to make a design closer to the
flooding point than 70 %.
– System flooding:
System flooding occurs if even large droplets are carried over the top by the vapor
flow without the influence of column equipment, i.e. in the empty column. A cor-
relation is given in [100]. Normally, system flooding is the last flooding criterion
which indicates flooding. It is, however, relevant in packings with very open struc-
tures, e.g. Mellapak 125X. Another application is the check of flooding data from
vendors, which are definitely too optimistic if they exceed the system flooding
limit.
– Load point:
The load point refers to the vapor flow where the vapor starts to influence the
liquid film. At this point, the efficiency of the packing is at its maximum, and one
is far enough away from the flood point. One should try to operate the column at
this load point, but as a design criterion it is not useful.
– Sufficiently low pressure drop:
The pressure drop can be a direct criterion for the diameter of the column. In many
applications, there is a limitation for the bottom temperature to avoid decompo-
sition. In these cases, a certain pressure drop over the entire column must not be
exceeded, as the pressure in the condenser is usually fixed.
There are two pressure drops: the dry pressure drop and the pressure drop of the
irrigated packing, which is the relevant one for design and where the dry pressure
drop has a contribution. The pressure drop increases first linearly with increas-
ing vapor load, with the liquid load as an additional parameter. Beyond the load
point, the pressure drop rises more rapidly, and at the flood point theory says that
it is infinite. In practice, flooding or, respectively, inoperability of the column is
reached already at finite pressure drops. For the Sulzer packings, there is even a
special model where flooding is defined as the vapor load where the pressure drop
is 12 mbar/m [104].
– Minimum liquid load:
A minimum liquid load according to packing and distributor should be kept to
avoid maldistribution. A lot of packing operation problems are due to maldistri-
bution [105]. An adequate distributor must be chosen. Special distributors can
handle very low liquid loads down to 0.04 m3 /(m2 h) (Figure 5.12). A reference
value for the number of droplet holes is 60–150/m2 . As a rule of thumb, after a
packed bed length of 6 m redistribution should take place, with additional space
requirement for collector and distributor. The need for redistribution can vary.
For thin columns, the accumulation of liquid at the column wall is more pro-
nounced, so that redistribution must take place earlier. For high liquid loads,
the packed bed length might be extended. Anyway, it is strongly recommended
to ask the manufacturer about an appropriate maximum length of a packing bed.
For vacuum application, the pressure drop of the distributor might be relevant.
As correlations do not make too much sense due to the wide variety of distributor
types, a value of ∆p = 1 mbar is usually reasonable as a first guess.
One should always have in mind that hydrodynamic calculations do not claim to be
very accurate. An uncertainty of ± 30 % is a reasonable assumption.
Figure 5.14 shows the load diagram of a packing, where the flooding line is de-
picted as a function of the liquid load. It can be seen that a minimum liquid load is
necessary to reach the operation region. With increasing liquid load, the vapor load at
the flooding point decreases, at high liquid load even rapidly. It should be mentioned
that the design criterion of 70 % of the vapor load at the flood point (flooding factor
70 %) can be interpreted differently. It can either mean that flooding occurs when both
vapor and liquid load are increased simultaneously from 70 % to 100 %. This flooding
factor is called FFLG and is usually relevant for the design of distillation columns.
Another flooding factor can be defined if flooding occurs when only the vapor load is
increased from 70 % to 100 %, maintaining the liquid load constant (FFL). This would
be a reasonable quantity for the design of absorption columns. The engineer must
decide which one is more relevant for the particular application case.
101
m/s
Flooding
Gas load VG/AC
Minimum
100 liquid
load
Operating region
10–1
10–4 10–3 10–2 m/s 10–1
.
Liquid load VL/AC
3.6 m3/m2h 36 m3/m2h
Fig. 5.14: Load diagram for a packing.
5.3 Tray columns
The lack of a reliable calculation method does not cancel the necessity of providing a reasonable
design for a tray.
(Volker Engel)
In tray columns, a stage-wise mass transfer takes place by means of horizontal trays
(Figure 5.15). A weir causes the accumulation of liquid coming from the tray above.
Over this weir, the liquid leaving the tray goes into the downcomer to the tray below.
The vapor can rise from tray to tray through openings characterizing the type of tray.
When it passes the liquid, it is split into small bubbles with a large mass transfer area
and distributed in the liquid. A so-called froth area is formed, giving good presump-
tions for an intensive mass transfer. In fact, there are three ways in which a tray can
be operated:
– In the bubble regime the liquid is the continuous phase. The vapor as the disperse
phase is rising through the liquid as bubbles. The bubble regime often occurs at
high pressure applications.
– The froth regime is the preferred one. A froth layer with a large interphase be-
tween vapor and liquid is formed. Both phases are more or less continuous.
– At high vapor loads, low liquid loads or in vacuum, trays often operate in the spray
regime; in this case, the vapor phase is the continuous one. Note that the appli-
cation range of many correlations does not cover the spray regime. In the spray
regime, weeping (p. 169) and unsealed downcomers (p. 167) should be avoided⁷.
In [95], the limiting formulas for the occurrence of the particular regimes are given.
One should always try to get froth regime; spray regime should be avoided if possible.
Vapor to condenser
Man hole
Reflux from
condenser
Downcomer
Sieve tray
Tray fitting
Man hole
Weir
Liquid feed Inlet weir
Sieve tray
Downcomer
wall
Hydrostatic
Vapor from sealing
reboiler
Fig. 5.15: Constitution of a tray column.

Liquid to reboiler Courtesy of Prof. Dr. J. Stichlmair.
7 The guideline in the spray regime is: hold on to your liquid!

Essentially, there are three types of trays.

– Sieve trays:
Through a sieve tray, the vapor from the tray below gets into the liquid through
holes in the plate of the tray. It is dispersed in the liquid. Today relatively large
holes up to d = 20 mm are used to avoid blocking by fouling. The maximum frac-
tional hole area is 11–12 %. The vapor prevents the liquid to leave the tray through
the holes (weeping); nevertheless, for this purpose a sufficient vapor velocity is
necessary. After shutdown, sieve trays are emptied through the holes. Sieve trays
(Figure 5.16) are quite simple to manufacture and have a relatively low pressure
drop. They are also easy to clean. As the bubbles entering the tray from below are
not diverted, sieve trays are prone to entrainment, meaning that droplets from the
froth layer get to the tray above. As this liquid is transported in the wrong direc-
tion, the efficiency of the tray can be significantly lowered. Sieve trays are not well
suited for large load variations. The turndown, i.e. the ratio of max. and min. load,
is approx. 2 : 1 [106]. Due to their simplicity, sieve trays are easy to specify and to
calculate. The tray efficiency of sieve trays is as good as for other tray types.
Fig. 5.16: Sieve tray.

Courtesy of Ludwig Michl GmbH.
– Bubble cap trays:

The principle of a bubble cap tray is that the rising vapor entering the tray above
is diverted by the bubble cup located above the holes (Figure 5.17). It enters the
liquid parallel to the tray, in contrast to the sieve tray. Therefore, the entrainment
is comparably low. They are not emptied after shutdown. Bubble cap trays have
a good efficiency and a wide-spread load range. However, the pressure drop and
tendencies to fouling and corrosion are relatively high, the manufacturing is ex-
pensive, and the cleaning is difficult. Because of the large number of types, bubble
cap trays are also difficult to specify. The hydrodynamic calculations are compli-
cated [107–110] and require a large number of input parameters. Nevertheless,
some special constructions like the Bayer-Flachglocke (Figure 5.18) are still con-
sidered to be indispensable. Using extremely low weir heights, the pressure drop
and the holdup on the tray can be kept very low.
Fig. 5.17: Bubble cap tray.

Fig. 5.18: Bayer-Flachglocke.

– Valve trays:
On valve trays, the hole of the tray is covered by movable valves (Figure 5.19). Sim-
ilar to the bubble cap tray, the vapor enters the liquid parallel to the tray, giving
less entrainment. The lift of the valve determines the opening area for the vapor; it
is self-adjusting to the vapor load. Figure 5.20 shows a typical pressure drop char-
acteristics of a valve tray. At low vapor loads, the valves are fully closed, and the
vapor enters the tray through the open crevices of the valves. The pressure drop
rises with increasing vapor load. At point CBP (closed balance point), the valves
begin to open. Further increase of the vapor load leads to a wider opening of the
valves, and the pressure drop stays constant. After they are fully open (point OBP,
open balance point), the pressure drop at increasing vapor load rises again. Thus,
the valves are in some way self-adjusting to the vapor load. This is the main ad-
vantage of valve trays. Their turndown is much higher than the one of sieve trays
(approx. 4.5 : 1 [106]). In fact, on Figure 5.19 there are two kinds of valves. On sort
is equipped with an additional plate on the top, giving extra weight. At low loads
only the light valves will open, and at high loads the heavy ones are working as
well. This option gives even more flexibility at differing loads.
The pressure drop of valve trays is between the ones of bubble cap and sieve
trays. Weeping can usually be avoided. Valve trays are sensitive to fouling. The
maximum fractional hole area is 13–14 %. Valve trays are by about 20 % more
expensive than sieve trays [106]. There are many types of valve trays available.
The data of various manufacturers have been collected in professional software
programs for hydrodynamic calculations. Valve trays are the most common trays,
their market share is approximately 70 % [95].
– Fixed valve trays:
Fixed valve trays are, in principle, sieve trays with a roof over the sieve holes (Fig-
ure 5.21). They do not move during operation, as the name suggests. The roofs
might prevent some entrainment, as the vapor is at least diverted to the side open-
ings. Nevertheless, the performance of fixed valve trays is pretty similar to that of
sieve trays. The maximum fractional hole area is 12–13 %. The investment costs
are approx. 10 % higher, and the main operation advantage is a slightly better
turndown (approx. 2.5 : 1) [106].
Fig. 5.19: Valve tray.

Δр
. . .
VCBP VOBP Vgas
Fig. 5.20: Typical course of the pressure drop of a valve tray.

Courtesy of WelChem GmbH.
Fig. 5.21: Fixed valve tray.

– Dualflow trays:
The dualflow tray is more or less a sieve tray without downcomer and weir. Vapor
and liquid go through the holes in countercurrent flow. The liquid goes down to
the tray below by weeping. This is limiting for the vapor load. Dualflow trays are
often used when the capacity has to be increased for an existing column. This is
achieved by using the cross-flow area of the downcomer as active area as well.
Dualflow trays are well-suited for systems tending to polymerization. Due to the
missing downcomer, they are slightly less expensive than sieve trays, but their
turndown is supposed to be lower (approx. 1.5 : 1 [106]). The efficiency of a du-
alflow tray is approximately 80 % of a normal tray.
Because of the hydrostatic pressure of the liquid on a tray, which has to be passed
by the vapor, tray columns have a significantly higher pressure drop than packed
columns. This is a disadvantage especially for vacuum distillations. The elevated
pressure leads to higher temperatures at the bottom of the columns, and the resi-
dence times on the trays are larger because of the greater holdup. This gives higher
decomposition rates; usually, the substances exposed to vacuum distillation are sen-
sitive against high temperatures. Tray columns are relatively insensitive against liquid
load variations and fouling, but susceptible to foaming. For small column diameters
(< 800 mm), man holes do not make much sense. Tray columns are then designed as
so-called cartridge columns (Figure 5.22), which usually have sealing problems at the
column wall.
Fig. 5.22: Cartridge column.

WDC
HW,in
WRSP
TS
HW,out HCL
IDcol
AA
LW,in LCL LW,out

ADC
FPL
Fig. 5.23: Dimensions in tray geometry.

For the specification of trays, the main geometric data are defined in Figure 5.23. In
this context, the particular abbreviations mean:
IDcol inner column diameter
AA active area
TS tray spacing
LW,out outlet weir length
LW,in inlet weir length
ADC downcomer area, top
WDC downcomer width

HCL downcomer clearance
FPL flow path length
HW,in height of inlet weir
HW,out height of outlet weir
LCL apron length
WRSP width of inlet weir
Generally, the residence time on trays is not large enough to reach equilibrium be-
tween vapor and liquid phase. For the thermodynamic calculation of the column,
one of the standard procedures is to take three trays for two stages, e.g. simulate a
60-trays column with 40 theoretical stages. Although this is often successful, this ap-
proach has difficulties with widely boiling systems, where the column is far away from
equilibrium. For the equilibrium calculation, it is better to characterize the quality of
a tray by the so-called Murphree efficiency:
y n − y n−1
EM = , (5.6)
yeq (x n ) − y n−1
where yeq is the equilibrium concentration and n − 1 refers to the tray below tray n. The
Murphree efficiency always refers to a certain component. Normally it is in the range
EM = 0.6–0.7, corresponding to the approach with the reduction of stages described
above. There is a problem with the product streams of the column. If Murphree effi-
ciencies are used, these product streams are not in equilibrium as expected. To avoid
simulation errors with pieces of equipment that require a single phase at the inlet
(compressors, pumps), an efficiency of EM = 1 should be assigned to the trays where
product streams leave the column.
The approach of taking three trays for two theoretical stages is certainly too simple
to be fully correct. High viscosities of the liquid and high separation factors have a
negative effect on the Murphree efficiency. The O’Connell correlation (5.7) takes these
effects into account. The Murphree efficiency is estimated to be [111]
ηL −0.245
EOC = 0.492 ⋅ ( α) , (5.7)
mPa s
where the viscosity refers to the feed at arithmetic average temperature in the column.
The relative volatility α should be evaluated with the separation factors of the key
components of the separation:
y LK /xLK
α= , (5.8)
yHK /xHK
where the subscripts “LK” for “light key” and “HK” for “heavy key” are used. Equa-
tion (5.7) tends to be slightly conservative, indicating that the usually assumed
60–70 % are not unreasonable. There are other influences on the tray efficiency,
which are discussed in [111]. When the fractional hole area is increased from 8 % to
14 %, the efficiency decreases by 5–10 %. It also depends on the hole diameter, an

increase from 0.25′′ to 0.5′′ gives an increase in efficiency by 5–10 %. Further increase
shows almost no effect; the efficiency even decreases again if the hole diameter is
drastically increased to 2′′ and more. The flow path length has a significant influence
on the efficiency: it increases the contact time between liquid and vapor. Doubling
the flow path length gives an increase of the tray efficiency by 5–15 %. One should
consider that e.g. the step from a one-pass to a two-pass tray has a negative effect on
efficiency, as the flow path lengths are shortened. The influence of weir height and
tray spacing are marginal, as well as the one of the load, as long as the hydrodynamic
criteria are fulfilled (Figure 5.24). The effect of pressure is already considered in Equa-
tion (5.7), as higher pressures cause higher temperatures with lower viscosities. For
high pressures (p > 10 bar), there is a negative effect due to increased entrainment
[111]. Of course, the strongest negative effect on tray efficiency is caused by maldistri-
bution.
The use of the Murphree efficiency has another advantage: the numbering of trays
is different in simulation and in construction. In process simulation, the trays are
numbered from top to bottom. Reboiler and condenser are counted as stages. In con-
struction, the trays are numbered from bottom to top. Therefore, a confusing and error-
prone renumbering procedure has to be performed. It is much easier to assign the trays
in the construction when the Murphree efficiency is used. If the reduction of stages had
been applied, it is much more complicated, not to mention that the factor is often for-
gotten when the overall pressure drop of the column is evaluated.
The hydrodynamic calculation of tray columns has a different philosophy than
the one for packed columns. While for packed columns self-contained models have
been developed with certain physical foundations, only single and incoherent corre-
lations are available for the particular failure criteria of tray columns. These criteria
are in most cases empirical but quite well confirmed due to the great experience with
the use of distillation trays. The correlations have limited accuracy and applicability.
If different correlations are compared, contradictions often occur. Often, the correla-
tions are based on measurements with the water/air system. The particular correlation
equations can be found in [95, 97, 98].
Figure 5.24 shows the load diagram of a sieve tray, which illustrates the range
where the sieve tray can be operated with a good efficiency. It is typical for trays that
the range for the vapor load is comparatively small, while the liquid load can be widely
varied. The upper limits can be interpreted as absolute limitations; when they are ex-
ceeded, the tray cannot be operated. The lower limits are recommendations; falling
below them might cause a bad efficiency of the tray. The criteria for limiting the oper-
ation range of a tray are as follows.
– Froth height:
The height of the froth layer should be lower than the tray spacing, with a certain
margin.
2.0
Maximum vapor load
Froth layer
too high
Jet
Vapor load G/Aak [m/s]
1.0 Operating region

flood
Downcomer
0.8 choke flood
0.6
Minimum Minimum
weir load vapor load
0.4
0.2
10 10–3 10–2 10–1
Liquid load L/Aak [m/s]
Fig. 5.24: Load diagram of a sieve tray.

– Entrainment and jet flood:

Entrainment is the carry-over of liquid droplets to the tray above. This would lower
the tray efficiency significantly and has a drastic impact if high purities of the over-
head product are required. The entrainment usually determines the maximum
vapor load of the column. There are correlations available which can evaluate
the order of magnitude of the entrainment. Besides the column diameter and the
diameter of the tray holes, the tray spacing has a decisive influence on the entrain-
ment (Figure 5.25). Common tray spacings are 350–600 mm, in North America,
only 600 mm are generally used. The entrainment reacts sensitively on variations
of these parameters; for example, an increase of the vapor load by a factor of 10
can increase the entrainment rate by 4–5 orders of magnitude. For a new tray de-
sign, it is best to keep the entrainment as small as possible. For a check of an
existing column, 10 % entrainment might be tolerable as long as very high puri-
ties are not required. If the entrainment is so high that not only droplets but larger
parts of the liquid are carried over, the jet flood case is reached as the limit of oper-
ability. Massive liquid entrainment has the consequence that liquid is impounded
on the trays, and finally, the column floods. While normal entrainment just leads
to a bad tray efficiency, jet flood makes the operation of the tray impossible. The
measures to avoid jet flood are to increase the active area or to use smaller holes,
as long as there is no danger of plugging. As mentioned above, more tray spacing
helps.
10–1
dh < 10 mm
10–2
10–3
Relative entrainment E/G [m3/m3]
ΔH·Δp·g
2
=
Fmax
10–4 33
40 50
60 75
10–5
100 50
1
10–6 20 0
300
10–7
10–8
0.03 0.06 0.1 0.2 0.4 0.6 1.0 2.0
WG F
2 XXX
0.6 1 2 3 WBL 2.5 [φ·σ·Δр·g]
Vapor load
Fig. 5.25: Influence of tray spacing on liquid entrainment.

Courtesy of Prof. Dr. J. Stichlmair [112].
– Pressure drop:
The pressure drop is an indication of the vapor load. It is not a direct limitation; the
increase of the bottom temperature plays a large role in vacuum columns, where
trays are rarely used. If the pressure drop is too large, the downcomer can show
flooding (see below). The pressure drop can be split into three parts. As for packed
columns, a tray has a dry pressure drop, which would occur even if no liquid is
on the tray. The second part is the static pressure drop due to the height of the
liquid on the tray, which is at least equal to the weir height. The third part takes
the side effects into account, like liquid height over the weir, bubble formation
or spraying. A typical pressure drop of a tray is in the range 5–10 mbar. From this
structure, a minimum pressure drop of the tray can be estimated from the second
contribution:
∆pmin,tray = ρL ghW . (5.9)
If a calculated pressure drop is below that limit, there is a strong suspicion that
something is wrong⁸; either the pressure drop calculation or the assumption that
all trays are operating properly (e.g. strong weeping).
– Downcomer choke flooding:
If the friction losses in the downcomer are too large, the designed liquid flow can-
not go down to the tray below and will accumulate. The highest friction losses
occur at the downcomer entrance. The reason is vapor formation from the de-
gassing in the downcomer. Therefore, one common measure is the increase of the
downcomer width. As a rule of thumb, it should be at least 10 % of the column
cross-section area; however, this depends strongly on the individual situation.
A sloped downcomer is often useful, as it influences the friction losses at the
downcomer entrance selectively (Figure 5.26). For the ratio BFo/BFu, 2 : 1 is a rea-
sonable value. Downcomer choke flooding preferably occurs at relatively high
pressures (p > 7 bar) and/or high liquid rates.
Fig. 5.26: Sloped downcomer.
– Minimum downcomer residence time:

The residence time in the downcomer is strongly related to the downcomer choke
flooding. It should be large enough so that the liquid has enough time for de-
gassing. There are various recommendations; they refer to the apparent residence
time, which is defined by the ratio of the downcomer volume and the clear liquid
flow in the downcomer [95]. The minimum one is 3 s which should really be kept.
In this context, it should also be mentioned that the clear liquid height in the
downcomer, i.e. the area really filled with liquid and not with a froth layer, is usu-
ally less than 50 % of the tray spacing. There are correlations for the clear liquid
height; the difference to the apparent liquid height taking the froth layer into ac-
count as well is often significant and must be considered when the residence time
in the downcomer is evaluated. A similar criterion to the minimum residence time
is the maximum downcomer velocity. Calculated as the clear liquid velocity in the
cross-flow area of the downcomer inlet, it should be less than 0.06–0.18 m/s.
8 Exception: spray regime.

– Downcomer backup flooding:

As Figure 5.27 indicates, there is a certain condition for the discharge of the liquid
from the downcomer:
p2 − p1 = ∆ptray < ρL ghL,DC − ∆pfriction . (5.10)
If this condition is not fulfilled, no more liquid can be discharged to the tray below.
Certainly, the contribution of ∆pfriction is caused by the liquid load and is related
to the downcomer choke flooding, but often a pressure drop of the tray which is
too high is the reason for this type of flooding. This means that finally the down-
comer failure is caused by the vapor load. Therefore, in this case the increase of
the downcomer width as the typical measure against downcomer failure makes
things even worse; the increase of the downcomer cross-flow area reduces the ac-
tive area of the tray and therefore the vapor velocity and the pressure drop of the
tray increase. If possible, a reduction of the downcomer width might help or, of
course, an increase of the tray diameter. In many cases, the pressure drop for the
passing of the apron is the reason for downcomer backup flooding. In this case,
the downcomer clearance should be increased, i.e. the distance between the tray
and the lower end of the apron. Often, the outlet weir height has to be increased
as well to ensure a liquid seal. If the downcomer clearance is larger than the outlet
weir height, vapor could bypass the tray above by choosing the way through the
downcomer. In fact, this is not a strict criterion, as the liquid seal usually per-
sists during operation even if the outlet weir height is slightly lower, but design
programs might give a warning in this case. A small tray spacing can also be re-
sponsible for downcomer backup flooding. An increase in tray spacing might be
useful; for large tray spacings like TS = 600 mm downcomer backup flooding is
hardly possible.
Downcomer backup flooding is not likely in the spray regime.
p1
∆ptray
hL,DC
Fig. 5.27: Hydrostatic and pressure drop

p2 on a tray.
– Weir load:
To ensure a uniform flow of liquid over the weir, the weir load should be in the
range
2 m3 /(m2 h) < weir load < 60 m3 /(m2 h) . (5.11)
For large columns, weir loads up to 100 m3 /(m2 h) can be accepted. A common
measure to reduce the weir loads is the use of multipass trays for column diam-
eters d > 2000 mm (Figure 5.28). The increase of the column diameter is not ef-
fective; the throughput increases approximately with the square of the diameter,
whereas the weir length only increases linearly. The capacity of the weirs can then
become the limiting factor; large gradients could be the consequence, giving ex-
cessive weeping (see below) on one side of the tray and complete entrainment on
the other side (Figure 5.30). The multipass option provides more weir length on a
tray and can overcome this difficulty.
Fig. 5.28: Two-pass column.

At low liquid loads, the liquid flow on a tray will become inhomogeneous. The
weir load can be used as an indicator according to Equation (5.11)⁹. If it is below
the lower limit, the use of notched weirs might be an alternative, especially if large
fluctuations of the liquid load occur (Figure 5.29). At low liquid loads, only the
lower part of the notches is charged, giving a lower effective weir length. For the
V-notches, the weir length is increased steadily according to the increase of the
liquid load. They are used when large ranges for the liquid load have to be covered.
The blocked weirs with rectangular notches are used if the weirs are too long, e.g.
the central weirs in a two-pass column. In this case, the height of the spikes is
about as high as the froth layer on the tray.
– Minimum vapor load:
If the vapor load is too low, the liquid on the trays comes in contact with the vapor
in an irregular way. The vapor will then prefer the area near the weir, as there is
9 Another criterion is the height over the weir; it should be more than 5 mm.
Fig. 5.29: Weirs with rectangular and V-notches.

less liquid due to the gradient on a tray. On the other side of the tray, the height of
the liquid is larger and the vapor will avoid passing the tray there. The liquid is no
more prevented to go through the holes, which is called “weeping”. Weeping is
less dramatic than entrainment. In contrast to entrainment, the liquid goes to the
tray below as intended. A decrease of the tray efficiency can be the consequence,
as residence time on the tray is lost. Normally, 10 % weeping can be tolerated.
There are correlations for the weeping rate [95] or for the minimum vapor velocity
to avoid weeping at all. If weeping occurs, one must be careful, as the correlations
do not take into account that the tray openings for the vapor are occupied by the
weeping liquid. Therefore, the pressure drop can be larger than calculated. The
first measure to avoid weeping is the reduction of the hole diameter.
The worst way of weeping is the so-called vapor cross-flow channeling (Fig-
ure 5.30). At high liquid loads (weir load > 50 m3 /(m h)), large fractional hole
areas, high tray diameters and pressures p < 7 bar, the liquid can build up a
hydrostatic gradient after entering the tray at the downcomer apron. At the outlet
weir, there is much less liquid height than at the downcomer inlet. The vapor
chooses the way which has the lowest pressure drop and passes the liquid on
the tray near the outlet weirs, maybe with significant entrainment. On the other
hand, there is accumulation of liquid at the downcomer inlet, and weeping occurs
as the vapor avoids to go to this region. Entrainment and weeping occur on the
tray simultaneously. The weeping is detrimental to the efficiency in this case. The
weeping liquid bypasses two trays, it is directly transported from the tray inlet to
the outlet of the tray below [106].
– System flooding:
For system flooding, the same statements can be given as for packed columns
(p. 152). System flooding might occur for very large tray spacings (TS > 1000 mm),
or for dualflow trays with TS = 600 mm.
As a design strategy, the following guideline might be helpful:
– Estimate the column diameter with wG = 1 m/s.
– Downcomer choke flooding 󳨀→ enlarge downcomer.
– Downcomer backup flooding 󳨀→ enlarge active area.
– Entrainment/jet flood 󳨀→ enlarge tray spacing and/or active area.
– Always check sensitivities.
– If nothing succeeds 󳨀→ increase/decrease column diameter.
Fig. 5.30: Vapor cross-flow channeling.

For the first guesses, Table 5.1 could be useful.
Table 5.1: Reasonable guesses for tray data [112].
Vacuum Ambient pressure High pressure

Tray spacing 0.4–0.6 m 0.4–0.6 m 0.3–0.4 m
Weir length (Lw ) 0.5–0.6 d B 0.6–0.75 d B 0.85 d B
Weir height (hw ) 0.02–0.03 m 0.03–0.07 m 0.04–0.1 m
Downcomer clearance 0.7 hw 0.8 hw 0.9 hw
Bubble cap diameter (dbc ) 0.08–0.15 m 0.08–0.15 m 0.08–0.15 m
Bubble cap distance 1.25 dbc 1.25–1.4 dbc 1.5 dbc
Valve diameter (dv ) 0.04–0.05 m 0.04–0.05 m 0.04–0.05 m
Valve distance 1.5 dv 1.7–2.2 dv 2–3 dv
Sieve hole diameter (dh ) 0.01 m 0.01 m 0.01 m
Sieve hole distance 2.5–3 dh 3–4 dh 3.5–4.5 dh
Fractional hole area 10–15 % 6–10 % 4.5–7.5 %
Beyond the conventional design, tray columns can cause trouble which is usually not
expected. Regularly, foaming turns out to be a knock-out criterion for tray columns
[113]. There are few countermeasures:
– The consideration of a foam factor in the design just scales the residence time in
the downcomer. It could only be useful if the foaming tendency could be assigned
numerically. This is still a research topic [114].
5.4 Comparison between packed and tray columns | 171
– Mechanical foam deletion should be considered as an academic fantasy. To have

movable parts on a tray sounds expensive, and the author is not aware of any
success stories.
– Replacing a thermosiphon reboiler by a falling film evaporator or, respectively,
exchanging trays by packing might be successful, but it is quite an expensive trial
without guarantee.
– Finding the reason (usually an unknown substance in the ppb-region) or trying
to make use of the theory of foaming [115] has almost never been reported to have
solved a problem.
– Foam formation on the trays is only possible for low vapor loads in the bubble
regime. The foam is rapidly destroyed at transition to the froth regime, after the
vapor load has been increased [112]. However, foam formation in the downcomer
is not affected in this way.
– The only way to fight foaming right on the spot seems to be the use of an an-
tifoaming agent. As long as it can be tolerated in the bottom product, this is the
only strategy which has turned out to be successful in the long run. Finding an
appropriate antifoaming agent is a science on its own, but vendors usually have
specialists which can give valuable advice.
Another kind of problems with tray columns is vibration, which has been observed es-
pecially on one-pass sieve and valve trays with large diameters (> 2000 mm) at compa-
rably low vapor loads. Within hours, trays have been seriously damaged. Just increas-
ing the mechanical stability of the tray usually does not help. Instead, an increase of
the vapor load or the use of notched weirs should additionally be taken into account.
There are a few explanations in the literature [116–119]. It seems that the vapor enters
the tray above noncontinuously as jet pulses through the holes. When they synchro-
nize and actuate in phase, the resonance frequency of the tray might be struck [95].
Certainly, a thorough understanding of this phenomenon has not yet been achieved.
5.4 Comparison between packed and tray columns
The criteria for a comparison between packed and tray columns can be set up in the
following way.
– Pressure drop:
The pressure drop of packed columns is significantly lower than for tray columns,
as the vapor does not have to pass a liquid layer and the narrow holes, which
have a cross-flow area of max. 14 % of the active area. For well-designed pack-
ings, the pressure drop can be expected to be 1–2 mbar/m, whereas it is 5–10 mbar
per tray. Consider approximately two theoretical stages per m in a packed col-
umn and a Murphree efficiency of 67 % (i.e. three trays are two theoretical stages),
the pressure drop of a theoretical stage is lower by a factor of 7.5–30 in a packed
column. Lower pressure drops give lower pressures in the bottom of the column,
and at lower pressures the relative volatility is in most cases higher. Furthermore,
lower pressures correspond to lower temperatures, meaning less degradation of
the products, and it might be possible to use a steam with lower pressure. For high
vacuum columns, the use of packed columns is more or less obligatory.
– Theoretical stages per m:
The first guess is that there are two theoretical stages per m in a packed column,
usually there are slightly more. For a tray spacing of 500 mm, there are two trays
per m, which, however, represent about 1.3 theoretical stages due to their effi-
ciency. On the other hand, some space is always lost for collectors and distrib-
utors. In fact, it must be examined case by case where finally more theoretical
stages per m are obtained. The author would guess that in most cases the packed
column is advantageous.
– Flexibility:
Tray columns have a good turndown (2 : 1 – 4.5 : 1, according to tray type). So do
packings, but the distributors usually do not. Therefore, packed columns are def-
initely inferior to valve trays from that point of view. In tray columns, side draws
are easy to provide. Usually, they are designed on several trays, and optimization
is performed during operation¹⁰. For packed columns, side draws are a major con-
structive issue; they are related to the presence of a collector and a distributor.
– Fouling:
Packed columns are more sensitive to fouling, especially structured packings and,
more especially, wired-mesh packings. Sieve trays with large holes (> 10 mm) can
cope with fouling quite well. They can even handle certain amounts of solids.
– Foaming:
Foaming is a strong argument against tray columns. Columns with random pack-
ing can at least cope with limited foaming, whereas columns with structured pack-
ing perform worse [106]. For systems which are prone to foaming (aldehyde sys-
tems, caustic absorptions) this is a major item for the equipment choice.
– Column diameter:
Tray columns are well established for large column diameters. Multipass trays are
often used. There is also no objection to using packed columns for large diameters.
Sieve trays with large diameters can be subject to vibrations. For small column
diameters (< 800 mm) packed columns are advantageous, tray columns can only
be realized as cartridge columns (Figure 5.22).
10 Side draws must not be taken at the bottom of the downcomer, as gas bubbles might lead to choking
or pump cavitation.
– Loads:
Tray columns can cope with large variations for the liquid load. They are relatively
sensitive to variations of the vapor load. Random packed columns have difficulties
with low liquid loads, whereas structured packings have difficulties with high liq-
uid loads, as gas bubbles cannot be released in the narrow channels. If they are
captured, they will go down a long way. In contrast, random packings can eas-
ily get rid of gas bubbles, and tray columns transport them one tray down in the
downcomer. Aqueous systems are difficult for any kind of packing due to the high
surface tension. In contrast, trays have no problems with wetting at low liquid
loads or with aqueous systems.
– Residence time:
Trays provide a well-defined residence time so that they are advantageous when
desired chemical reactions are to occur in the column. On the other hand, packed
columns are advantageous to avoid undesired reactions. In this case, one should
take care that the reboiler has a small holdup and residence time as well.
– Heterogeneous systems:
For systems showing liquid-liquid equilibria, the design of packings is weakly
founded. Trays have no problem, as in the froth layer the separation of the two
phases does not take place.
– Safety:
In recent years, a lot of fires in structured packings have happened, and the reason
for this is the large surface per mass, as the thickness of the metal is just about
0.1 mm [120].
5.5 Distillation column control
In process engineering, typical control oriented modeling methods have not become
widely accepted, as it is the case in electrical and aerospace engineering. The control
engineering is not an own phase; instead, it is distributed over the life cycle of a plant
[121]. The main part of the control strategy is already fixed in the conceptual phase and
tested during the piloting. The draft of the control scheme is done according to the ex-
perience from other projects. In-depth knowledge of control engineering is requested
if the proposed strategy fails in operation or some optimization potential is presumed.
During the engineering, a systematic analysis of the control strategy hardly ever takes
place. Things might change with the upcoming training simulators and advanced pro-
cess control projects.
The strongest reason is that plant models are usually specific for a given plant.
Their development is a complex, large, and expensive effort and must be supposed
to be justified. Often, during the model development the knowledge on the process
increases in a way that ways are found to meet the targets with conventional process
engineering methods as well. The process models are not only extraordinarily com-
plex but also time-dependent due to catalyst aging or fouling on heat exchangers.
During the life cycle of a plant, many changes take place, such as capacity increases,
heat integration, improved catalysts, or changed product specifications. In these cases
it would be necessary to adjust the model and update the control parameters. The dis-
crepancy between the large effort for an adequate control engineering and the often
simple but successful solutions is one of the main reasons for the skepticism of prac-
titioners to a theoretically based control engineering.
Especially for distillation plants, a simple, intuitive control strategy is usually
chosen to maintain the demand for a sufficient and constant quality. Generally, there
are two types of variables to be controlled in distillation [96]. The control of column
pressure and liquid levels ensures that no accumulation of liquid (levels) and vapor
(pressure) occurs. Otherwise, a steady-state operation of the column would not be
possible. Maintaining the levels and the pressure just keeps the stable operation of the
column. For the product quality, the composition of the products has to be kept within
the specification limits. For this purpose, some kind of composition control must take
place. The special difficulty in distillation control is that the product concentration
can hardly be measured effectively and fast enough. Even gas chromatography as a
relatively fast analyzing method with a wide application range has too long response
times because of the distance between sample point and detector. In a control cycle, it
would be a great dead-time element. Further problems are high investment and main-
tenance costs and a possible phase separation of the sample, which could spoil the
result. In fact, there are few constructions of this kind¹¹, but the usual way for the qual-
ity control of the product uses the temperature profile as an indicator for the product
composition. The sensitivity of the temperature profile from the product concentra-
tion is exemplarily illustrated in Figure 5.31. The diagram is about the simulation of a
distillation column separating the binary system n-hexane/n-heptane on 50 theoreti-
cal trays at ambient pressure, where the reflux ratio has been varied. The three cases
refer to the overhead concentrations of n-hexane of 99.98 wt.%, 99.9988 wt.%, and
99.99984 wt.%, respectively. These purities each differ by one order of magnitude.
The boiling points of these overhead products can hardly be distinguished by mea-
surement in column operations and are not appropriate as a control signal. However,
the temperature profiles are significantly different. The largest differences seem to oc-
cur on stage 20. The temperature of this stage could therefore be used as a control
variable for the overhead concentration¹². There have been discussions on whether
vapor or liquid temperatures should be used for composition control. In fact, both
options work. The control by vapor temperatures show the faster response but might
be erroneous if weeping occurs.
11 e.g. oxygen analyzers.

12 Unfortunately this stage is relatively far away from the top of the column. Using this temperature
as a control variable might lead to a slow response behavior.
50
45
40 99.98%
99.9988%
35 99.99984%
30
Stage
25
20
15
10
5
0
340 345 350 355 360 365 370 375
T/K
Fig. 5.31: Temperature profiles at various overhead concentrations [89].

There are a lot of options for the control of a distillation column. Figure 5.32 shows an
often used one, which is discussed in the following section. In this example, the purity
of the overhead product is maintained by the control of the temperature (TC) on a
certain stage by manipulating the reflux amount. The distillate stream is the difference
between overhead and reflux stream. The outlet valve maintains the level in the reflux
drum (LC). This way, a steady state can be achieved. At the bottom of the column, the
steam flow is fixed (FC). The control of the bottom product flow is analogous to the
distillate flow control.
The column pressure is controlled (PC) by the cooling water flow to the condenser.
If the pressure is too high, it will be increased to condensate more and to lower the
pressure again¹³.
The control strategy in Figure 5.32 is not very fast anyway. To understand a control
scheme it is useful to follow the response of the column after something is varied. Con-
sider the case where the concentration of the feed varies, e.g. the light ends fraction
increases. The light ends will accumulate in the top section of the column. The level
controller of the reflux drum will open the distillate outlet. The control temperature in
the profile will drop. Then, the system answers with a reduction of the reflux amount.
This reflux must go down the column stage by stage, which takes some time. For fre-
quently occurring load variations, this control scheme is not really appropriate. Its
strength is to safely maintain the quality of the top product for constant load.
13 This kind of pressure control is relatively sluggish; alternatives are discussed below.
PC
LC
Feed
TC
FC
LC
Fig. 5.32: Control scheme of a distillation
column with top product quality control.
PC
LC
Feed FC
TC
LC
Fig. 5.33: Control scheme of a distillation
column with bottoms product quality control.
If the bottoms product quality has to be maintained, the arrangement in Figure 5.33 is
advantageous. In this case, the temperature control manipulates the steam, and the
change in the vapor phase affects the rest of the column rapidly, much faster than
the reflux change does in Figure 5.32. This scheme has a huge advantage in stripping
columns, where the distillate rate is small compared to the feed. It should be men-
tioned that the considerations about the response times mainly refer to tray columns,
whereas packed columns have a relatively small holdup and show pretty fast control
answers even for the option in Figure 5.32. In both schemes, the temperature control
can also be connected to the distillate or the bottom product outlet valve, while the
levels are maintained by the reflux flow or by the steam flow, respectively. All these
options have their pros and cons depending on the particular case, however the most
important rule is that small streams should not control a level, neither in the reflux
drum nor at the column bottom, as this results in extremely slow answers (Richard-
son’s law).
PC
LC
Feed FC
FC
LC
Fig. 5.34: Example for an energy balanced
control scheme.
Generally, we can distinguish two types of control schemes: the mass balanced con-
figuration (MB, Figures 5.32 and 5.33) and the energy balanced configuration (EB, Fig-
ure 5.34) [96, 122].
The mass balance control is the common way to control a column. The most im-
portant principle is that none of the product streams is flow-controlled. Otherwise, the
column would easily run out of the balance. If the flow of the product is fixed, it must
exactly fit to the corresponding fraction of the desired overhead components in the
feed. Consider a feed flow with 5000 kg/h low-boiler and 5000 kg/h high-boiler. If the
top product flow is fixed, it would have to match to the low-boiler content in the feed,
i.e. 5000 kg/h. If it is fixed to another value, the impurity of one product stream is un-
avoidable. In practice, it is neither possible to determine the exact fraction of a feed nor
are fluctuations avoidable. A coincidental 2 % deviation, e.g. 5100 kg/h low-boiler and
4900 kg/h high-boiler, would not be compensated, and 100 kg/h low-boiler would end
up in the bottom fraction. With a mass balance control scheme, both product streams
are controlled by other information, e.g. by the levels (Figures 5.32 and 5.33) or by a
temperature in the column profile. A number of configurations are possible as shown
above, they are well explained and discussed in [96].
The explanation above should rule out a control scheme fixing a product flow, but
in fact it makes sense for small product streams where the composition does not mat-
ter. Consider a case where there are 10 000 kg/h high-boiler and 10 kg/h low-boiler,
which are to be separated. The small overhead product stream will not have an influ-
ence on a level to be used for control purposes. An easy way to control this column
is to set the distillate flow to 20 kg/h. This implies that 0.1 % of the product are lost,
but the low-boiler should be completely removed, including some fluctuations in the
feed. The reflux ratio is a result of the steam flow, which can be related to a tempera-
ture in the profile or fixed like in Figure 5.34. This is an example of an energy balanced
configuration.
It might be surprising that the reflux ratio, which is one of the most important
parameters in column design, is seldom controlled directly. Usually, either the reflux
or the distillate flow are controlled, and the reflux ratio is only an operand.
It is also remarkable that there is no chance to control both top and bottom com-
positions to maintain them within their specification [96]. It seems obvious to set one
control temperature each in the stripping and in the rectifying section (Figure 5.35). In
fact, these control loops would seriously interact with each other. If for example the
fraction of low-boilers in the feed rises, both control temperatures decrease and cause
the corresponding reactions, maybe more steam flow to the reboiler for the bottom
concentration and a decrease of the reflux for the top product one. However, these
two reactions have completely different dynamic responses. While the reaction on the
steam increase is relatively fast, the effect of decrease of the reflux is pretty slow; it
is transported with the liquid and therefore coupled with the residence times of the
liquid on the particular trays. When the desired effect occurs, the increased steam
flow already requires an increased reflux, and a never-ending cycle starts. Therefore,
in conventional control schemes it is avoided to control both compositions.
PC
LC
Feed TC
TC
LC Fig. 5.35: Unstable control scheme using two
composition controllers for bottom and distil-
late products.
In all cases it often makes sense that streams fixed by flow control are related to the
flow of the feed stream (e.g. fixed reflux, fixed steam flow).
There are various options for pressure control. As in Figures 5.32 and 5.33, the
coolant flow through the condenser can be controlled by the pressure. An easy and
fast but not very elegant way is the feeding and venting of inert gases with a split
range control (Figure 5.35). Together with the inert gas, always a certain amount of
top product is vented; therefore, the column should be expected to operate steadily or
with low product concentrations in the vapor phase so that the losses are not harmful.
More options are explained in [96].
There is also an option to use pressure drops or temperature differences along
the column as input signals. It might happen that they have a larger significance than
5.6 Constructive issues in column design | 179
conventional signals, especially if the temperature itself is determined by undefined

components. However, these options can show instabilities in malfunction cases. If for
example the cooling water service fails, the temperature differences along the column
will become smaller. In Figure 5.33, the steam valve would fully open, which is the
worst reaction of all and will probably lead to safety valve actuation. A behavior like
this should at least be prevented by appropriate interlocks.
5.6 Constructive issues in column design
There are also some simple considerations in column design apart from component
separation. Similar to vessels, there are certain requirements for the liquid level at
the bottom. There must be several minutes of time for the operators to react if the
liquid level goes up and down between the particular levels (Chapter 9). If there is
a thermosiphon reboiler, it has to be checked whether the circulation takes place at
minimum liquid level. The nozzle for the vapor inlet from the reboiler must not be
flooded, and a sufficient distance must be kept to the lowermost tray or the packing,
respectively. Usually, the particular companies have their design guidelines where rec-
ommendations for the various issues are given.
There are several options for the bottoms design when a thermosiphon reboiler
is involved (Figure 5.36). While the first one is the simplest and the most common
one, a baffle can be placed in the bottoms region for different purposes. Option (2)
ensures that there is always enough liquid height to maintain an appropriate NPSH
value for the pump removing the bottoms product. The thermosiphon circulation is
well ensured with option (3) which maintains its driving force constant.
(1) (2) (3)
Fig. 5.36: Three options for the column bottom design.

Special care must be taken when a feed consists of vapor and liquid or when a su-
perheated feed flashes inside a column, meaning that pressurized liquid generates
vapor when it is expanded into a low-pressure column. A simple nozzle is only ac-
ceptable at low velocities with low vapor fractions. In other cases, some constructive
provisions must be made to enable the liquid to vanish in bottom direction, while the
vapor is smoothly directed to the top. Popular options are flash chambers (Figure 5.37).
In a flash chamber, the vapor can disengage in a defined way, while the liquid is col-
lected and guided to a liquid distributor below. Flash chambers can be located inside
(Type IN) or outside (Type A) the column. Type IN is appropriate for small flash vapor
amounts. For large ones, Type A is the better choice, which is in principle a small vessel
with a demister. For high velocity feed where the vapor is the continuous phase, vapor
horns are one of the favorite solutions. A tangential helical baffle forces the vapor to
follow the contour. It is closed at the top and open at the bottom. The liquid drops hit
the wall and run downward as requested [100].
(a) (b)
Fig. 5.37: Flash chambers for use inside (Type IN, (a)) and outside (Type A, (b)) the column.
Courtesy of Julius Montz GmbH.
5.7 Separation of azeotropic systems
Azeotropic systems cannot be separated by conventional distillation. Using applied

thermodynamics [11], there are a number of options to break an azeotrope and ob-
tain the components in their pure form. It is the art of the process engineer to suggest
the most appropriate one. A good description of the various separation processes for
azeotropes can be found in [8].
The easiest case is the separation of a heteroazeotrope. It splits in two-phases in a
decanter, and the two-phases can be worked up separately [8]. Pressure-swing distil-
lation is useful if the azeotropic concentration strongly depends on the pressure. This
is the case for relatively few binary systems. The most well-known one is tetrahydrofu-
rane–water (Figure 5.38). In the first column, the azeotrope THF–water with xTHF ≈ 0.8
is taken overhead at low pressure (p ≈ 1 bar), while pure water can be removed from
the process at the bottom. The overhead stream is condensed and compressed to a sig-
nificantly higher pressure (p ≈ 10 bar). At this pressure, the THF concentration of the
azeotrope is significantly lower (xTHF ≈ 0.6). Pure THF will remain at the bottom of the
second column when the azeotrope is taken overhead at high-pressure. It is recycled
to the first column. Overall, the outlets of this arrangements are water and THF with
arbitrary purity.
1 2
THF
P1 P2
H2O
Fig. 5.38: Pressure swing distillation for the
separation of the tetrahydrofurane–water
azeotrope [89].
THF © Wiley-VCH Verlag GmbH & Co. KGaA.
H2O Reproduced with permission.
Other examples are acetonitrile–water, methanol–acetone, ethanol–benzene, and

even ethanol–water [123]. The great advantage is that no additional substances have
to be introduced into the process.
There are four other main principles of azeotropic separation, which are il-
lustrated using the azeotrope ethanol–water (Figure 5.39). Ethanol–water is the
azeotrope which is split most often worldwide, with a capacity of 40 million tons
per year. It is quite an unpleasant one, as on the branch between the azeotrope and
the pure ethanol hardly any separation via distillation is possible (Figure 5.39).
– Azeotropic distillation (Figure 5.40):
After the azeotrope is obtained at the top of column K1, a substance is added
which forms a ternary azeotrope with ethanol and water. The most common op-
tions are benzene and cyclohexane (CHX), where the latter is nowadays preferred
2.4
2.0
p / bar
1.6
t = 100°C
1.2
0.8
0 0.2 0.4 0.6 0.8 1
x,yEtOH / mol/mol
Fig. 5.39: Ethanol–water azeotrope at t = 100 °C.
ETOHH2O
B7
FRESHCHX REC-CHX
S2 DECANTER
B6
S5 S7
S3 S4 K2
B5
FEED K1
S9 K3
ETOH
p = 1 bar
H2O p = 12 bar
p = 2.7 bar
Fig. 5.40: Azeotropic distillation of ethanol–water using cyclohexane.

due to the toxicity of benzene. With this ternary azeotrope, all the water can be
taken at the top in column K2, while pure ethanol is obtained at the bottom. The
ternary azeotrope can be split into two phases in the decanter (Chapter 6.1). The
upper phase consists of cyclohexane with small amounts of ethanol, which can
be directly recycled to column 2. The lower phase can be worked up in a further
distillation and recycled to the decanter or column 1, respectively.
– Extractive distillation (Figure 5.41):
The advantage of extractive distillation is that it is not necessary to start with
azeotropic concentration; an easily achievable preconcentration to approx. 90 %
is sufficient. In the first column K1 the water is washed down to the bottom with
a solvent which has lower activity coefficients with water than with ethanol. A
widely used one is ethylene glycol (1,2-ethanediol). At the top of column K1, the
ethanol is obtained with the desired purity. The bottom product is a mixture of
water and ethylene glycol. These components are separated in the second col-
umn K2. Ethylene glycol as the bottom product can be recycled and used again as
extractive agent in column K1.
EG REC-EG
B3
RECYCLE
ETOH
ETOHH2O K1
WATER
BOT1 K2
COOLER
B4
BOT S2 PURGE
Fig. 5.41: Extractive distillation of ethanol–water using ethylene glycol.

– Adsorption:
Especially for ethanol, an adsorption process has been developed where the re-
maining water in the azeotrope is removed with a molecular sieve. The process is
described in Chapter 7.2 (p. 215). The advantages are its robustness and simplicity.
Especially in the ethanol business these are two major items. A disadvantage is
the fact that it is necessary to start with azeotropic concentration and the com-
plicated control. The azeotrope contains 4 % water, with is quite a lot. It makes
it necessary to change the bed after a few minutes operation. The adsorber bed
must then be regenerated.
– Membrane:
Similar to adsorption, the water in the azeotrope can be removed with a mem-
brane (Chapter 7.1). The process strategy is the same, first, azeotropic composition
must be achieved, and then the water is removed using a multistep membrane
separation [124].
5.8 Rate-based approach
In the conventional simulation of a distillation or an absorption, it is assumed that the

liquid and the vapor phases which are leaving a stage are at complete equilibrium, i.e.
phase equilibrium, thermal equilibrium and mechanical equilibrium. Furthermore,
complete mixing and complete separation of the phases is assumed. In reality, these
assumptions are, of course, never fulfilled. To take this into account, efficiency factors
or HETP values are introduced so that realistic results can be obtained. However, one
must be aware that HETP might depend on the column diameter, the properties of the
substances, or the liquid and vapor flow rates. Tray efficiencies and HETP values are
never accurate; instead, they should more or less be interpreted as good guesses.
The so-called rate-based approach is an alternative. It considers the heat and mass
transfer between the phases which encounter each other in the column. It accounts for
the influence of throughput, equipment size, packing or tray properties and physical
properties of the fluids so that extrapolations are more reliable [125]. The heat and
mass transfer rates are determined by quantifying the temperature and concentra-
tion¹⁴ differences between the phases which are the driving forces of the separation.
The characteristics of the contacting device, i.e. the generated transfer areas, are also
taken into account. Thermodynamic equilibrium is still a very important piece of infor-
mation for the calculation, but it is only assumed at the interface between the phases,
referred to as vapor and liquid film in Figure 5.42. The mathematical details can be
found in [126]. Applying a rate-based approach, one should be aware of some pecu-
liarities which are often unexpected:
– The temperatures of vapor and liquid on a stage are generally different, as the two
phase do not reach equilibrium.
– In a multicomponent mixture, a component can diffuse in opposite direction to its
concentration gradient. This can happen in situations where the fluxes of the par-
ticular components are strongly coupled. The phenomenon has been thoroughly
described and experimentally proved in [127] and [128].
– The term of a theoretical stage is still used but no longer considered in the final
calculation. For packed columns, the packing is divided into so-called segments,
which have nothing to do with equilibrium stages or HETP; however, there are
rules of thumb to choose useful values for their height which are related to the
HETP values. The segment height should always be lower than the HETP. For ran-
dom packings, 10–12 times the size of the packing elements is a good approach,
for structured packings, HETP/2 is a reasonable choice. In general, the number of
segments should have no major influence on the calculation result, as long as the
choice is reasonable.
For tray columns, the trays theirselves are the entity.
14 to be correct: chemical potential differences.

5.8 Rate-based approach | 185
Lj–1 Vj
TLj–1 TVj
Heat transfer
Liquid vapor
Bulk liquid Bulk vapor
film film
Mass transfer Fig. 5.42: Rate-based approach.

Screen images of Aspen Plus® are reprinted
with permission by Aspen Technology, Inc.
Lj Vj+1 AspenTech® , aspenONE® , Aspen Plus® , and
the Aspen leaf logo are trademarks of Aspen
TLj TVj+1
Technology, Inc. All rights reserved.
Besides the phase equilibrium and the enthalpy description, the transport properties
are necessary for the calculation, i.e. viscosities, thermal conductivities, surface ten-
sions, and diffusion coefficients. As mentioned in Chapter 2.12, the viscosity of liquid
mixtures is not really accurate. The diffusion coefficients can be estimated very well
for gases, but for the liquid probably only the correct order of magnitude can be de-
termined [11]. Moreover, the mass transfer models are not accurate in any case, it is
a matter of experience to choose the best one, and even this is not a guarantee for a
correct representation of the system. The rate-based approach is not generally more
accurate than the equilibrium calculation. It connects a number of uncertain quan-
tities for the representation of the column, whereas the equilibrium model mixes all
of these influences together and represents it with one single uncertain value, i.e. the
HETP or the efficiency.
Nevertheless, there are a number of cases where the rate-based approach gives
significantly different results, and one should know when it makes sense to go beyond
equilibrium thermodynamics. One should be aware that the effort to switch to the rate-
based model in commercial process simulation programs is actually limited and is not
a reason for refusing this attempt; convergence has also been substantially improved
in recent years.
Rate-based calculations are more or less obligatory for absorption and desorption
processes, which are in most cases mass transfer limited. The efficiencies vary greatly
from component to component and from stage to stage, as well as in strongly non-
ideal systems. In absorption, the efficiencies are usually only 10–20 %, but also values
like 5 % are possible. In reactive distillation, the efficiency does not make sense at all,
when the main progress on a stage is the proceeding of the reaction and not of the
separation. Reactions with fast reaction rates might be mass-transfer limited. Trace
components, which have low mass transfer rates due to their low concentration, can
often not be adequately treated with an equilibrium calculation. As well, systems with
big gaps between vapor and liquid temperature can be heat transfer limited.
For illustration, here is a classical example. The absorption of traces of HCl from
exhaust air with water requires one single theoretical stage with an equilibrium
model, as the HCl is an electrolyte and will completely and immediately dissoci-
ate in water. In this case, it becomes clear that the use of a rate-based model is not a
matter of accuracy. The procedure which determines the removal of the HCl from the
air is the mass transfer in the gas phase by diffusion. It takes much more effort than
the absorption itself when the HCl has reached the boundary layer. For the correct
dimensioning the application of a rate-based model is obligatory, it will result in a by
far larger packing height¹⁵. Other measures are not appropriate, neither the increase
of the water amount nor the use of caustic soda, which makes the chemical absorption
just more irreversible.
The HCl does not know that the NaOH is waiting for it in the liquid!
Even the rate-based approach does not represent the total truth. There are a lot of sys-
tems which form aerosols in the vapor phase, e.g. sulfuric acid–water or the system
HCl–water which was just mentioned above. Aerosols are formed when, due to over-
saturation, condensation takes place directly in the vapor phase and not in contact
with the liquid phase. The droplets formed are not large enough to settle down into
the liquid, and they are not small enough to take part in the diffusion process. They are
entrained with the vapor phase, distorting the principle of distillation and absorption.
To explain the theory would go beyond the scope of this book. Good explanations can
be found in [129] and [130].
5.9 Dividing wall columns
Conventional distillation columns can separate a feed mixture into two product
streams with the desired concentration, as long as the design of the column is ap-
propriate. In case of a binary mixture, two pure component streams can be obtained.
In Figure 5.43, a side draw is taken from the column additionally. However, there is
no way to obtain a third product with any desired concentration. Consider a three-
component mixture with the light-end A, the heavy-end B, and the middle-boiler C. As
the feed (stream 2) is located below the side-draw, it can easily be achieved that there
is no heavy-end B in the side-product. However, light-end A will pass the stage where
the side-stream is taken, and certainly, part of it will be inevitably in the side-stream.
15 often several m of packing, if TA Luft must be reached.

5.9 Dividing wall columns | 187
INERTS
OVERH
COND
DIST REFLX
C1 SIDE Fig. 5.43: Column with side draw.

FEED
AspenTech® , aspenONE® , Aspen Plus® , and
BOT the Aspen leaf logo are trademarks of Aspen
INERTS
OVERH
COND
DIST REFLX
C1 SIDE Fig. 5.44: Dividing wall column principle.

FEED
AspenTech® , aspenONE® , Aspen Plus® , and
BOT the Aspen leaf logo are trademarks of Aspen
Dividing wall columns (Figure 5.44) are an alternative, which is increasingly becom-
ing established. Part of the column is divided by a separation wall which prevents
lateral mixing. The feed enters the column on the left hand side and is split into the
components A + C at the upper and B + C at the lower end of the separation wall.
At the top of the column there is a rectifying section, giving pure component A as top
product. Also, at the bottom the pure heavy-end B can be obtained in the conventional
stripping section. The right part of the column is fed with a mixture of A + C from the
top and B + C from the bottom. At an appropriate stage in the middle of the right-hand
side, pure product C can be withdrawn.
There is usually a distributor at the top of the divided section where it can be
controlled how much liquid is fed from the top to the particular column partitions.
Without a device, the vapor coming from the bottom is split in a way that the pressure
drop in both partitions is the same. Therefore, for a proper design the pressure drop
correlation used should work sufficiently well. Dividing wall columns represent the
highest degree of heat integration between columns; it is estimated that the energy
INERTS
OVERH
VR
COND
DIST REFLX
LL
C1LEFT SIDE
FEED VL C1RIGHT
LR
BOT
Fig. 5.45: Dividing wall column in process simulation.

savings amount to 20–35 % in comparison with an adequate conventional distillation

arrangement [131].
From the process simulation point of view, the dividing wall column can be repre-
sented by two independent columns which are linked by streams entering and leaving
the partition on the right-hand side (Figure 5.45). The stripping section, the partition
on the left-hand side and the rectifying section form the first column (C1LEFT). At the
upper end of the dividing wall, a liquid stream (LL) is withdrawn and led to the right
partition, while the complete vapor flow of the right partition enters the rectifying sec-
tion (VR). Analogously, part of the vapor from the left partition is taken as a side stream
to the bottom of the right partition (VL), while the whole liquid from the bottom of the
right partition is fed to the stripping section from the top (LR).
Calculating hydraulics in a dividing wall column is not really well-defined by the
commercially available programs. The only way the author knows is to choose a cir-
cular cross-section with equal area. This might be quite a good approximation for a
packed column, whereas for trays the passes for the liquid are considerably different.
Nevertheless, it is expected that the commercial programs will soon be supplemented
due to the increasing number of dividing wall column applications.
5.10 Batch distillation
Distillation can be performed either in the continuous mode or as a batch distillation

[123]. A continuous distillation is operated at steady state, meaning that the state vari-
ables do not change with time. The feed is continuously entering the column, while
top and bottom products are continuously withdrawn. In batch distillation, the feed
5.10 Batch distillation | 189
is filled into the bottom vessel of the column at the beginning of the operation (Fig-
ure 5.46 (a)) regular configuration). Depending on the time, various products can be
withdrawn at the top of the column. Side stream products and continuous feeds are
optional. There is usually no bottom product, the residue in the bottom vessel can
be removed from the column at the end of the distillation process. The state vari-
ables in the column change with time, the process is inherently unsteady [123]. Batch
distillation is often preferred to continuous distillation if relatively small amounts of
material which occur irregularly and possibly with changing composition have to be
separated. It is used extensively in laboratory separations and in the production of
fine and specialty chemicals, pharmaceuticals, polymers and biochemical products.
Batch distillation units are very flexible; they can usually handle different kinds of
products, and as a matter of principle, only one column is necessary to split a mixture
into its components unless azeotropes occur. As well, hydrodynamic calculations are
not as important as for continuous columns; if the column diameter does not fit, the
throughput can simply be distributed over a larger time, as long as it is in line with
the time schedule of the process (Chapter 3.3). Moreover, a batch product has its own
identity, i.e. it can strictly be controlled which feedstock a product comes from, which
is often important for quality control in the production of pharmaceuticals [123].
Essentially, there are two different kinds of batch distillation [132]:
– Operation with a constant reflux ratio, where the distillate composition changes
continuously. The final product concentration is an average value. At the begin-
ning, it is usually higher so that the product purity is above specification. At the
end, the reflux ratio is lower than requested, and it has to be taken care that
operation is stopped as long as the product concentration is in line with the spec-
ification. This is a considerable disadvantage, the successful operation can be
proved only at the end of the batch. If the final product turns out to be off-spec,
the batch has to be blended or rerun [123].
– Operation with a constant distillate composition, where the reflux ratio is varied.
This is normally the better approach, however, it requires a control mechanism
and is more complex. It might happen that the controller settings must be adjusted
during the process due to rough changes of the process conditions.
Of course, even both the reflux ratio and the product composition can be varied to
optimize the decisive criterion, e.g. batch time, product amount, or minimum cost.
While in the traditional batch distillation the feed is charged to the bottom, other
configurations might be useful. In an inverted batch column, the feed is charged to the
reflux drum at the top of the column. With this approach, it is easier to remove large
amounts of high-boiling components from the final product. Also, the middle vessel
configuration is generally more effective in terms of energy efficiency and product rate
[123]. All three configurations are illustrated in Figure 5.46.
(a) . (b) . (c) .

Qcond Qcond Qcond
Feed
Reflux Reflux Reflux
drum drum drum
+ Feed tank
Feed
Product Product
receivers receivers
Middle vessel
+ Feed tank
Reboiler
Feed + Feed tank
. . .
Qreboil Qreboil Qreboil
Product Product
receivers receivers
Fig. 5.46: Batch distillation configurations; (a) regular, (b) inverted, (b) middle vessel [123].
Batch distillations are more complicated to calculate than continuous ones, as the
MESH equations (p. 142) must be supplemented by a term describing the time-de-
pendent holdup on the various stages. Moreover, the change of the holdup on the
stages varies drastically with time¹⁶ so that the solution of this system of equations
is much more difficult. Alternatively, the batch distillation can be represented as a
series of continuous distillations where the holdups at the beginning and at the end
of the steps are taken as feed streams to or side products from the particular stages,
respectively [132].
16 At the beginning, the holdup is built-up, whereas at the end there are only slight changes in the
composition.
6 Two liquid phases
A thermodynamic sophistry says that it might happen that more than two liquid
phases can coexist in an equilibrium. Figure 6.1 shows a sketch of seven liquids
forming an equilibrium with the same vapor.
Vapor
Heptane
Aniline
Water
Perfluorokerosene
Phosphorus
Gallium
Mercury
Fig. 6.1: Seven liquid phases in equilibrium [133].
Fortunately, in technical applications these multi-liquid-liquid equilibria do not play

a major role, but there are a number of processes where two liquid phases occur.
Especially in extraction processes the two liquid phases are essential, as well as in
heteroazeotropic distillations, where two liquid phases form after condensation which
must be carefully separated for the continuation of the process.
6.1 Liquid-liquid separators
Usually, in a process the two liquid phases form a dispersion where one phase (dis-
continuous phase) is distributed in a continuous phase as droplets [134]. In technical
applications, in most cases an aqueous phase and an organic phase occur, where the
aqueous phase is the heavy one¹. In a liquid-liquid separator, the dispersion should
be transformed into two homogeneous phases. Figure 6.2 shows a horizontal liquid-
liquid separator. On the left-hand side, the dispersion enters the separator. As the
velocity decreases due to the enlargement of the diameter, turbulence and kinetic en-
ergy are reduced. A layer of droplets is formed, where droplets do not coalesce. On the
1 There are exceptions, for instance halogenated organic substances are often heavier than water.
192 | 6 Two liquid phases
right hand side, the coalescence takes place. Small droplets slowly coalesce. This can
be improved by internals. There are two kinds of these:
– internals which reduce the kinetic energy and distribute the liquid over the whole
cross-flow area;
– internals with a large surface where the droplets can coalesce (e.g. plates, random
packing, wire-mesh).
Fiber layers are recommended for droplet diameters between 1 and 100 µm. Plate in-
ternals are relatively expensive but useful if solid particles or surfactants are involved
or if the pressure drop should be minimized.
Dispersion Fine droplets Continuous phase
Inlet area Dense layer of disperse phase
Separated disperse phase
Fig. 6.2: Horizontal liquid-liquid separator without internals.
Vent
Siphon breaker
Upper phase flow
Feed
Adjustable
spool
piece
Lower phase flow Normally

closed
Fig. 6.3: Adjusting the phase boundary with a siphon [135].

6.1 Liquid-liquid separators | 193
Light phase out
Feed
Fig. 6.4: Liquid-liquid separator with a collector

Heavy phase out for the heavy phase.
Fig. 6.5: Horizontal liquid-liquid separator in chemical industry.
The directing of the liquid phases is an important item of the design. While in Fig-
ure 6.2 the apparatus is completely filled with the two liquid phases, it is also possible
that space for the vapor is left. A siphon can be used to adjust the height of the phase
boundary between the two phases (Figure 6.3), where the top height of the siphon can
be varied if it is designed as a spool piece [135]. For small amounts of the heavy phase,
a special collector can be placed at the bottom (Figure 6.4).
A theoretically founded design of liquid-liquid separators is not possible. Some
influences are clear; a large density difference, a low viscosity of the continuous
phase, and large droplet sizes are favorable to the sedimentation. However, there are
many phenomena which are yet unclear, e.g. the sedimentation of droplet clusters, the
droplet size distribution, and the coalescence behavior. But the most unpredictable
issue is the influence of surfactants. Even just traces can significantly change the
separation behavior, which often just turns the design procedure into a lottery. Also,
solid particles often tend to form a layer (crud) which can disturb the separation of the
two phases. The generally acknowledged procedure on the design of liquid-liquid sep-
arators driven by gravity is the one of Henschke [136, 137], which describes the transfer
of the results of a batch settling experiment in a standardized cell into a design of a
separator. Figure 6.6 shows the course of such an experiment, where the light phase is
the dispersed one. After the mixing of the two phases has stopped, the droplets start
rising upwards. The sedimentation curve indicates the range of the lower part which
is free of droplets. If the sedimentation of the droplets is faster than the coalescence
at the phase boundary, a layer of droplets is formed. The droplets coalesce at the
phase boundary, which is continuously shifted downwards. The extent of the clear
light phase is represented by the coalescence curve. The experiment is finished when
only half of the phase boundary is covered with droplets; this definition is necessary
to become independent of statistical effects caused by single droplets which coalesce
lately. The course of the particular curves can be used to adjust the Henschke model.
New approaches are under development, which consider internals as well.
Coa Coherent
lesc
enc disperse phase
Clo e cu
dis se-p rve
pe ack
rsi
on ed
Height
Sedimentation
zone e
rv
cu
n
tio
n ta Continuous phase
e
im
d
Se
Time tE
Fig. 6.6: Course of a settling experiment with the light phase as the dispersed one [138].
Often, when the phase separation is relatively fast, the dimensions of the separator are
determined by its function as a vessel, giving the plant operators time to react (Chap-
ter 9). For the designer, this is a lucky situation, but it has to be proved by experience
in any case.
6.2 Extraction
In liquid-liquid extraction, a substance (extractive) is removed from a solvent by an

extracting agent in the liquid phase which is not completely miscible with the solvent
[89]. Extraction has advantages in comparison with distillation if
– the separation factors are small, in the worst case at the azeotropic point;
– there are several components with significantly different boiling points which can
be separated simultaneously;
– substances with extremely high or low boiling points occur;
– the concentration of high boiling substance is low so that a very large part of the
mixture has to be evaporated;
– sensitive fluids must not be heated up.
The stream containing mainly the selective agent and the extractive is called the ex-
tract. The solution that has been cleaned from the substance is called the raffinate.
The physical foundation of the extraction is the liquid-liquid equilibrium between
the substances involved. A single equilibrium step is usually not sufficient. As for dis-
tillation and absorption, the separation effect can be increased by providing a number
of separation stages in a row. Again, columns are possible, where the solution and the
extracting agent are introduced at opposite ends of the column. The driving force of
the countercurrent flow is the density difference between the two liquid phases, and
therefore the light phase inlet is at the bottom, and the heavy phase inlet is at the top.
The design of extraction equipment should provide good mass transfer conditions,
i.e. a large contact area of the phases at a high degree of turbulence. One of the two
phases is split into droplets, forming the disperse phase. There are many criteria for
the choice of the disperse phase, which are sometimes contradictory. Often, the phase
with the larger mass flow is dispersed to get a large contact area. In packed columns,
the phase with the better wettability should be the continuous one. If the disperse
phase is wetting the packing, the droplets could coalesce and become larger with less
surface for mass transfer. Furthermore, the mass should be transferred from the con-
tinuous to the disperse phase. Moreover, flammable or poisonous substances should
be dispersed to lessen the hazardous potential. The final decision should be based on
experiments.
Criteria for the choice of the extracting agent are the extent of the miscibility gap
with the solvent, a high selectivity and a large capacity for the extractive. The separa-
tion of the extracting agent from the extract should be as easy as possible; extractive
and selective agent should have a large difference of the boiling points and should
not form an azeotrope. The density difference between the extracting agent and the
solvent should be large so that the separation of the two liquid phases is easy; other-
wise, there is also the option to achieve the phase separation by centrifugation. The
surface tension between the two phases is a relevant quantity; if it is too large, the for-
mation of small droplets is difficult, if it is too small, the separation of the two phases
becomes hard. There are also practical items like low price, small vapor pressure, so
that the losses by evaporation are small, high thermal and chemical stability, and low
viscosity, flammability, and toxicity.
Analogous to distillation and absorption, extraction can be described with an
equilibrium model or a rate-based model considering mass transfer. A comprehen-
sive description can be found in [97]. Compared with distillation and absorption, the
computational modeling of liquid-liquid extraction processes has much more uncer-
tainties. The dimensioning of equipment for extraction is hardly possible without per-
forming a pilot scale test.
Even the calculation of the phase equilibria causes problems. Unlike the repre-
sentation of vapor-liquid equilibria, the binary interaction parameters (BIPs) for the
NRTL- or the UNIQUAC equation obtained from binary phase equilibrium data cannot
be simply transferred to ternary and multicomponent mixtures, as already mentioned
in Chapter 2.5. Usually, they yield results which are only qualitatively correct. For a
reliable description of liquid-liquid equilibria, the BIPs have to be adjusted not only
to binary, but also to LLE data of ternary mixtures². Moreover, the temperature depen-
dence of the BIPs, which is especially distinct for systems showing strongly non-ideal
behavior, must be carefully regarded.
Example
A water stream contains 400 wt. ppm of tetrachloromethane (CCl4 ). The CCl4 content shall be reduced
significantly. Does it make sense to use a hexane stream as extractive agent, which already contains
1 wt.% of CCl4 ? The extraction process consists of a one-stage mixer-settler (Chapter 6.2.1) arrange-
ment and happen at t = 30 °C and p = 1 bar.
Solution
In fact, this option had been disregarded in a similar practical case, as it seemed that diffusion cannot
happen against the concentration gradient. However, there are two mistakes. First, the concentration
which accounts for mass transfer is the mole concentration. It does not change much, but the com-
parison must be made between 47 mol ppm in the aqueous phase and 0.6 mol% in the organic phase.
Second, the concentration measure which is decisive refers to the chemical potentials, i.e. the activi-
ties. For easy illustration, the activity coefficients at infinite dilution are taken. For CCl4 , they are:
∞
γaq = 10200 ,
∞
γorg = 1.23 .
Thus, it becomes clear that the product xCCl4 ⋅ γCCl4 is larger in the aqueous phase, and in the equi-
librium stage the CCl4 will be transported from the aqueous to the organic phase. An evaluation of
the proposed mixer-settler equilibrium stage yields quite a good result: The CCl4 content of the aque-
ous phase is reduced to 6 wt. ppm or, respectively, 0.7 mol ppm. However, note that afterwards the
aqueous phase is also saturated with hexane (33 wt. ppm, corresponding to 7 mol ppm).
2 Of course, data from quaternary and higher mixtures would be useful, but there is little available.
But even if the phase equilibria are well-known, a number of issues exist which
will not be solved in the foreseeable future. The surface tension between two liquid
phases, which determines the effort for the phase separation, cannot be predicted.
Small amounts of impurities can have a significant influence. To determine the res-
idence time necessary for the phase separation, experimental tests are necessary. If
rate-based models are applied, droplet sizes and their distribution for the calculation
of phase boundary surface, the wettability of the built-in components and the opti-
mum velocities cannot be predicted. Also, the diffusion coefficients in the liquid phase
have a large influence on the results, but their estimation is inaccurate [11]. Finally,
the choice of the disperse phase can be supported by theoretical considerations, but
at the end of the day, an experimental confirmation is required.
The scale-up of extraction columns is more difficult than for distillation or ab-
sorption columns. Columns with large diameters do not show the same separation
efficiency as laboratory or pilot scale columns. The reason is supposed to be a more
widely distribution of the residence time of the droplets, which has in principle a neg-
ative influence.
For the design of an extraction, it is obligatory to perform a series of experiments
[138]. These experiments comprise shaking trials to characterize the dispersion and
coalescence behavior, and laboratory, miniplant, or pilot plant runs as the basis for
the scale-up.
In distillation, absorption and extraction, it is essential that the phases which par-
ticipate in the separation process are exposed to each other intensively. Subsequently,
they have to be separated again. This is easy for distillation and absorption due to the
large density differences between liquid and vapor but much more difficult in extrac-
tion with small density differences of the two liquid phases. Therefore, the equipment
used for extraction must enable the two phases to separate after a certain contact time.
6.2.1 Mixer-settler arrangement
The simplest concept is the mixer-settler arrangement, where mixing and separation
take place at two different locations. In the easiest way, the mixer is a stirred vessel,
and the separator is another vessel providing residence time for the settlement of the
phases. Wire-mesh or packing elements can support the separation. If phase equilib-
rium and complete separation are achieved, one theoretical stage is realized. Mixer
and settler can as well be arranged in a more compact way (Figure 6.7), where also
countercurrent flow takes place.
The advantage of the mixer-settler principle is the easy scale-up by numbering-
up. The load range is large, and mixer-settler units are appropriate for extreme mass
flow ratios of the two phases. The height of mixer-settler units is low, however, the
floor space required is very large, as well as the liquid holdup.
Light
phase
Fig. 6.7: Mixer-Settler arrangement [89].

Heavy © Wiley-VCH Verlag GmbH & Co. KGaA.
phase Phase boundary Reproduced with permission.
6.2.2 Extraction columns
For extraction sieve tray columns and both random and structured packing columns
are used [97, 138]. In contrast to distillation, the sieve tray columns do not have a weir.
There is a downcomer or a pipe so that the heavy phase can get to the tray below. The
light phase accumulates below the tray above. The heavy phase coming down dis-
places the light phase, which is forced to go to the tray above across the sieve holes.
Therefore, the light phase is the disperse one, whereas the heavy phase is the continu-
ous phase. If the heavy phase should be the disperse one, a different construction must
be chosen with pipes leading to the tray above. In extraction, these sieve trays have
an efficiency of η = 10–30 %. The load range is relatively small. Another disadvantage
is the fact that a relatively large part of the column is used for accumulating the light
phase below the next tray. On the other hand, the construction is quite simple, and
the backmixing by dispersion is effectively prohibited. Also, the scale-up is easy. The
density difference between the liquid phases should be quite large (> 100 kg/m3 ).
A significant improvement of the efficiency can be achieved by pulsation. The liq-
uid in the column is vibrated by means of a piston pump (Figure 6.8). The amplitudes
of these vibrations are 6–10 mm, and the frequencies are between 50–150 min−1 . The
light phase passes the holes during the upstroke, and the heavy phase passes dur-
ing the downstroke. In this way, new phase contact areas are continuously formed.
Another option for the pulsation is the movement of the trays theirselves. Pulsated
columns have a good separation efficiency, but like the nonpulsated tray columns the
load range is small.
For packed columns, it is important that the packing is easily wetted by the contin-
uous phase. The disperse phase should not wet the packing; otherwise, the droplets
might coalesce, which lowers the interfacial area.
Often, extraction columns with rotating elements are used. The principle is that
both phases are thoroughly mixed by means of input of mechanical energy. Many
small droplets with a large surface are formed, and the mass transfer is improved.
The two liquid phases are separated in designated zones, and the axial back mixing
as one of the key problems of extraction columns is restricted. The drawbacks of these
columns are the high price and that they are prone to malfunction and attrition.
Phase
boundary
Heavy
phase
Sieve
tray
Pulsation
unit
Fig. 6.8: Pulsated sieve tray column [89].

Light phase © Wiley-VCH Verlag GmbH & Co. KGaA.
Fig. 6.9: Kühni column.

One of the most popular extraction column types with rotating elements is the Kühni
column (Figure 6.9). A turbine agitator produces a circulation flow with a high inter-
facial area between the liquid phases. Perforated discs provide for the separation of
the phases. The separation efficiency is high, there are up to 10 stages per m. Again,
the small load range is the main drawback.
The rotating disc contactor (RDC) has horizontal rotating discs on a shaft, which
provide the dispersion of the phases (Figure 6.10). A minimum viscosity of the phases
is necessary, as the dispersion is caused by shear forces. The back mixing is restricted
by the stator rings linked to the wall. However, for assembly reasons the inner di-
ameter of these stator rings is larger than the outer diameter of the rotating discs
Fig. 6.10: Rotating disc contactor (RDC).

so that the back mixing is not really prevented. RDCs can realize large throughputs
but only 0.5–1 stages per m. The asymmetric rotating disc contactor (ARDC) is a fur-
ther development. The shaft with the rotating discs is placed non-concentrically with
the column axis (Figure 6.11). Separation and transport of the phases takes place in
dedicated zones at the column wall which are separated from the mixing zone by ver-
tical plates. Compared to the RDC, the maximum throughput is a bit lower, but the
separation efficiency is much better (1–3 stages per m).
The hydraulic design of extraction columns is difficult. Details can be found in
[138] and [97]. The most important criterion for the determination of the column di-
ameter is the flooding point. Flooding is reached when the countercurrent flow can
no more be maintained, e.g. if the buoyancy of the dispersed light phase is not suffi-
cient to overcome the flow resistance of the droplets in the continuous heavy phase
Rotor
Separation Mixing
zone zone
Stator
Intermediate
plate
Fig. 6.11: Asymmetric rotating disc contactor (ARDC) [89].

coming down. The droplets of the light dispersed phase can be entrained downward
to the bottom outlet, or a phase inversion can take place if the droplets accumulate
and coalesce. A reasonable calculation of these phenomena is hardly possible, as the
droplet size distribution as the necessary information is not accessible.
As a strongly simplifying consideration, the layer approach can be used. The two
phases cover a part of the cross-flow area according to their holdup and move in op-
posite directions in countercurrent flow. The velocities are in the order of magnitude
of 1 cm/s.
6.2.3 Centrifugal extractors
Centrifugal extractors are a third type of equipment for liquid-liquid extraction. The
countercurrent flow and the phase separation are not achieved by gravity but by cen-
trifugal forces. The internals and the liquid routing provide an intensive mixing and
the subsequent phase separation after a residence time of a few seconds, giving high
throughputs and low holdups. Both investment and operation costs are high, centrifu-
gal extractors are mainly used in the pharmaceutical industry or if expensive solvents
are involved.
Light phase Heavy phase outlet

outlet
Heavy PC PC
phase
inlet Light phase
inlet
Phase
boundary
Fig. 6.12: Podbielniak extractor [89].

The principle of centrifugal extractors is that the extractor is rotating at a high speed.
By means of the centrifugal forces the heavy phase is forced to the outer wall, whereas
the light phase is displaced towards the rotation axis. The Podbielniak extractor is an
example. The phase separation is supported by concentrical perforated sheets. Coun-
tercurrent flow is achieved by feeding the light phase at the wall and the heavy phase
at the rotating shaft. Podbielniak extractors can have 3–5 theoretical stages.
7 Alternative separation processes
For thermal separations, the alternative processes membrane separation, adsorption,
and crystallization can solve some problems where the standard operations fail. They
are explained in the following sections.
7.1 Membrane separations
At the inlet of the membrane there is one stream you do not like. Downstream the membrane,
there are two of them.
(Hans Haverkamp)
Fortunately, this does not always apply. Membranes can be used successfully for ther-
mal separation problems, especially in combination with other processes. Figure 7.1
shows the principle of the membrane separation process and the denomination of the
streams. The membrane separates two spaces from each other. However, substances
can pass through the membrane and get to the other side. The stream having passed
the membrane is called permeate, the stream which has not is the retentate. For the
various substances, the permeability of the membrane is different, which is the basis
of the separation effect.
Feed Permeate
Fig. 7.1: Principle of the membrane separation process [11].
Retentate Reproduced with permission.
There are two different principles of membrane separation. The first type acts like
a sieve or a filter; small molecules can pass the membrane (permeate), whereas
larger molecules cannot (retentate). As membrane materials, glass-like polymers like
polyetherimide or polysulfone are used. Depending on the size of the retained parti-
cles, it is distinguished between microfiltration, ultrafiltration, and nanofiltration.
Nanofiltration is normally used for the treatment of aqueous systems. It separates
particles down to a particle size of 1 nm. The driving force is a pressure difference
up to 40 bar between both sides of the membrane. In many cases, nanofiltration is
also ion-selective. While monovalent ions can pass the membrane easily, bi- or multi-
valent ions are held back. Well-known applications are the removal of water hardness
(Ca-ions), the decoloring of waste waters from the textile and pulp industry, and the
desalination of waste waters.
204 | 7 Alternative separation processes
Ultrafiltration is operated with a pressure difference across the membrane be-

tween 3–10 bar. It can be used for separating highly molecular substances from a liq-
uid. Microfiltration is used for removing particles between 0.1–10 µm.
The mass transfer through these porous membranes can be explained with the
pore model. For a porous membrane, the size of the molecules or ions to be separated
and the pore size of the membrane are of the same order of magnitude. In this case,
the membrane separation is comparable to a sieve filtration.
Solubility membranes act in a different way. The mechanism for this separation
is the combination of solution and diffusion. On the high pressure side of the mem-
brane, a component is dissolved in the membrane polymer. It is then transported to
the other side of the polymer by diffusion, and desorbs at the low pressure side of
the membrane. The driving force of the procedure is the partial pressure difference
between the two sides of the membrane. For the permeability of a component through
the membrane, the product of the solubility in the membrane polymer and the diffu-
sion coefficient is decisive, according to
Di Si
ṅ i = ∆p i , (7.1)
l
where ṅ is the mole flow [mol/(m2 s)], D is the diffusion coefficient [m2 /s], S is the
solubility parameter [mol/(m3 Pa)], and l is the thickness of the polymer layer [m].
For example, pentane has a higher permeability through a silicon membrane than
nitrogen. Its diffusion coefficient in silicon is three times lower than that of nitrogen,
but its solubility is 200 times larger. Therefore, the permeabilities differ by a factor
of approx. 60. The quality of the separation depends mainly on the selectivity of the
membrane, which is defined as the ratio (D1 S1 )/(D2 S2 ) of two components to be sepa-
rated. Figure 7.2 illustrates the permeation behavior of various substances in different
membranes, with some remarkable and unexpected results.
Slow Fast
N2 CO He Ar O2 H2 CH4 C2H6 CO2 C3H8 C4H10 H2O
Silicon membrane
C4H10 C3H8 C2H6 CH4 N2 Ar CO O2 CO2 H2 He H2O
Polyetherimide membrane
C3H8 C2H6 C3H6 CH4 N2 CO Ar O 2 H2S CO2 He H2 H2O
Polyimide membrane
Fig. 7.2: Permeation behavior of various substances in different membranes.

To achieve high fluxes through a solubility membrane at sufficient separation effi-

ciency, it is necessary that the active layer for the separation is extremely thin [89].
The handling of thin materials is difficult; the solution is a so-called asymmetric mem-
brane. These membranes consist of a thin active layer (approx. 0.01–0.05 µm) con-
nected to a porous supporting layer (approx. 100 µm). The supporting layer achieves
the mechanical stability without contributing significantly to the mass transfer resis-
tance. The supporting layer can be made of the same material (phase inversion mem-
brane) or a different one (composite membrane). In some cases, an additional layer
made of polyacrylnitrile fibres is used to further increase the mechanical stability.
Examples of membrane materials are polyvinyl alcohol, cellulose or its deriva-
tives for organic membranes, whereas inorganic membranes can be made of sintered
metal powder, glass in a spongy structure, carbon, or ceramic. Organic membranes are
more widely used because of their low price and their mechanical stability. However,
inorganic membranes are thermally and chemically stable and have long durabilities.
Elastomer membranes preferentially let organic substances pass and have lower per-
meabilities for low-boiling gases like nitrogen, oxygen, or hydrogen.
Membranes are used as modules which provide relatively high mass transfer ar-
eas per volume. The most established ones are pipe modules, coil modules, and plate
modules (Table 7.1, Figure 7.3).
Table 7.1: Specific mass transfer areas of the particular membrane modules [8].
Module Specific area [m2 /m3 ]
Pipe module 25
Plate module 100–600
Spiral wound module 500–1000
Capillary module > 1000
Hollow fibre module approx. 10000
Feed Hose, capillary, hollow fibre
Feed
Permeate Retentate
Retentate
Coil modules Permeate
Pipe modules Feed
Membrane
Permeate Retentate
Plate modules
Fig. 7.3: Different kinds of membrane modules.

Courtesy of Prof. Dr. J. Gmehling.
A prediction of the membrane separation behavior is difficult and not state of the art
yet. The flow conditions on both sides of the membrane play an important role, as
they have an influence on the concentration profile. Empirical or semiempirical mod-
els are needed for the modeling of the mass transfer. Any extrapolation of these models
is difficult, so that experiments are definitely needed for both the choice of the mem-
brane and the design of the membrane separation process. From the qualitative point
of view, the statement can be given that high fluxes are only possible if the solubility in
the membrane is high. Therefore, polar membranes (e.g. polyvinyl alcohol) are appro-
priate for the separation of water, while hydrophobic membranes (e.g. polydimethyl
siloxane, PDMS) can be used for the separation of organic components from aqueous
solutions.
Besides the design problems, there is always the question about the durability of
the membrane. The experience is that in multicomponent mixtures there is usually
at least one substance which is detrimental to the membrane. Proof that the mem-
brane is stable can only be achieved by a long-term test. This issue and the design
effort are the reason why membrane processes are only used if distillation or other
unit operations are not appropriate. But membrane separations are an option in com-
bination with other operations, e.g. with distillation to overcome azeotropic points.
For waste water treatment, where small amounts of organic substances have to be
removed, membrane separations are a very popular choice. Also, membrane separa-
tions are used for the separation of gas mixtures, the recovery of salts from diluted
aqueous solutions, the desalination of sea water, or dialysis for patients with a kidney
disease. Table 7.2 gives an overview on the most important membrane processes, the
phases involved, and the membrane types. For reverse osmosis, pervaporation and
vapor or gas permeation the same membrane type is used, the difference is just the
phases involved.
We can distinguish between dead-end and crossflow filtration. In dead-end filtra-
tion, the flow goes through the membrane in a perpendicular direction. The filtered
particles are collected at the surface of the membrane and form a filter cake. In cross-
flow filtration, the flow direction is parallel to the membrane surface. If particles occur,
they might deposit on the membrane. A sufficient flow velocity must be provided to
reach an equilibrium between deposition and abrasion.
For reverse osmosis, semipermeable membranes are used, where in the ideal case
no transport of dissolved components (e.g. salts) takes place. On the other hand, it
should be fully permeable for the solvent itself. Because of the concentration differ-
ence, the solvent (e.g. water) goes through the membrane until equilibrium is reached.
This is the case when the hydrostatic pressure is equal to the osmotic pressure (os-
motic equilibrium). If the pressure on the side containing the dissolved components
is increased above this osmotic pressure, the process is inverted, i.e. the solvent con-
centration on this side is even decreased (reverse osmosis). The most well-known ap-
plication is sea water desalination. In any case, where a heavy end component has
to be removed from water, reverse osmosis should be taken into account to avoid the
Table 7.2: Technically important membrane processes [8].
Membrane process Type Driving force Phases Application

Microfiltration porous ∆p < 3 bar S/L removal of solid particles
from suspensions
Ultrafiltration porous ∆p < 10 bar L/L waste water treatment,
drinking water purification
Nanofiltration porous/ ∆p < 40 bar L/L treatment of aqueous solutions
dense and oil fractions
Reverse Osmosis porous/ ∆p < 80 bar L/L waste water treatment,
dense drinking water purification
Dialysis porous/ conc. diff. L/L kidney dialysis, acid recycling
dense
Electrodialysis dense electric field L/L removal of ions
from aqueous solutions
Pervaporation dense fugacity diff. L/V separation of azeotr. systems,
removal of unwanted traces
Vapor Permeation dense fugacity diff. V/V separation of azeotr. systems,
water removal in reactions
Gas Permeation porous/ fugacity diff. G/G separation of gas mixtures
dense
Osmosis Osmotic equilibrium Reverse osmosis
xi > 0 xi = 0 xi > 0 xi > 0

pB pB xi = 0 pB xi = 0
pB pA pA
pB = pA Δp = pB – pA = Π Δp = pB – pA > Π
Fig. 7.4: Osmosis, osmotic equilibrium, and reverse osmosis [11].

evaporation of large amounts of water. A good rule of thumb for the pressure used is
60 bar. Reverse osmosis as explained above is illustrated in Figure 7.4.
The equation for the osmotic pressure can be derived from chemical potentials
[11]. Referring to Figure 7.4 and setting the activity coefficient for the solvent to 1, one
gets
RT
Π = pB − pA = − ln x j , (7.2)
vL,j
where the index j denotes the solvent. Equation (7.2) is very convenient to use, as mass
balances from process simulation indicate the molar concentration x j of the solvent.
Nevertheless, Equation (7.2) is often rewritten as
Π = p B − p A = RT ∑ c i , (7.3)
i
where the index i denotes the respective solutes. The temptation to use Equation (7.3)
manually is great. However, one must take into account that the ionic species disso-
ciate. Therefore, the mole number of dissolved species is higher than expected and so
are the osmotic pressures.
Example
10000 kg/h of a 1 wt.% solution of sodium chloride is to be concentrated to 10 wt.% by a single evap-
oration step without vapor recompression. Alternatively, a reverse osmosis unit can be inserted up-
stream, where the pressure is restricted to p = 60 bar. The temperature is set at 300 K. Estimate
whether the reverse osmosis unit can save operation costs. The electric rate shall be 10 ct/kWh, the
steam costs shall be 20 €/t.
Solution
First, the steam demand without the reverse osmosis unit is estimated. The mixed stream consists
of 100 kg/h sodium chloride and 9900 kg/h water. To increase the concentration to 10 %, the water
amount must be reduced to 900 kg/h. Therefore, 9000 kg/h of water, more than 90 %, have to be
evaporated, requiring approximately the same amount of steam. A large effort is necessary to change
small concentrations.
For the membrane consideration, the mole compositions of the stream are considered:
9900 kg/h
nwater = = 549.54 kmol/h
18.015 g/mol
100 kg/h
nNaCl = = 1.711 kmol/h .
58.4425 g/mol
The latter split into 1.711 kmol/h Na+ and 1.711 kmol/h Cl− ions, giving the concentrations
549.54
xwater = = 0.9938
549.54 + 2 ⋅ 1.711
1.711
xNa+ = = 0.00309
549.54 + 2 ⋅ 1.711
1.711
xCl− = = 0.00309 .
549.54 + 2 ⋅ 1.711
Applying Equation (7.2) with Π ≈ 60 bar (a bit less, as the permeate must have a remaining overpres-
sure to be transported out of the membrane) and a specific liquid volume of vL,water ≈ 0.001 m3 /kg =
0.018 m3 /kmol, one gets after solving to x
vL,j 0.018 m3 /kmol

xRO,water = exp [−Π ] = exp [−60 ⋅ 105 Pa ] = 0.9576 .
RT 8.31447 J/(mol K) ⋅ 300 K
This is the minimum mole concentration of the solvent one can achieve in the retentate with reverse
osmosis. The mole concentrations of the ions are
xRO,Na+ = xRO,Cl− = (1 − 0.9576)/2 = 0.0212 ,
corresponding to the mass concentration

0.9576 ⋅ 18.015
xROw,water = = 0.933 .
0.9576 ⋅ 18.015 + 0.0212 ⋅ 22.99 + 0.0212 ⋅ 35.453
The retentate contains the 100 kg/h NaCl and, correspondingly, 1392.5 kg/h water¹. Therefore, another
492.5 kg/h water have to be removed by evaporation. For the pressure elevation, assuming a pump
efficiency of η = 0.7 the power can be calculated to be
̇ L Π/η = 10000 kg/h ⋅ 0.001 m3 /kg ⋅ 60 bar/0.7 = 23.8 kW .

P ≈ mv
Without reverse osmosis, the operation costs are
Cevap = 9000 kg/h ⋅ 20 €/t = 180 €/h . (7.4)
For the option with reverse osmosis and evaporation, the operation costs
CRO+evap = 492.5 kg/h ⋅ 20 €/t + 23.8 kW ⋅ 10 ct/kWh = 12.23 €/h (7.5)
can be assigned. Over one year (≈ 8000 h), the difference amounts to 1.34 million €, which should
rapidly pay off the investment costs of the reverse osmosis.
Pervaporation is different from the other membrane processes, as not only the mem-
brane separation but also a phase change takes place. A liquid feed stream enters the
membrane module and is split into a liquid retentate stream and a permeate stream
in the vapor state. By lowering the partial pressure on the permeate side the fugac-
ity difference as the driving force is increased. Nevertheless, the enthalpy of vapor-
ization has to be added; otherwise the temperature on the permeate side would be
significantly lowered, especially in multiple-stage membrane modules. Besides the
removal of organic compounds from aqueous solutions (or vice versa, according to
the polarity of the membrane), pervaporation is an attractive option for the separa-
tion of azeotropes in combination with distillation. As mentioned above, the separa-
tion in the membrane depends mainly on the solubility and on the diffusion through
the membrane, so that the separation characteristics can differ significantly from the
vapor-liquid equilibrium.
In gas permeation, in contrast to pervaporation the inlet stream is gaseous as
well. The mass transfer is proportional to the fugacity difference across the membrane.
Porous and dense membranes can be used. The main application is the recycling of
hydrogen in the ammonia and methanol manufacturing processes. Moreover, gas per-
meation is used for nitrogen enrichment of ambient air, natural gas drying, the sepa-
ration of ethylene and carbon dioxide and the separation of helium from natural gas.
1 Check: 1392.5/(1392.5 + 100) = 0.933.

In electrodialysis, the potential difference on both sides of the membrane can be

increased by applying an electric field if electrolytes have to be separated [139]. Ion-
selective membranes can support this process.
Much more information on membranes can be obtained from [140] and [141].
7.2 Adsorption
When solids get into contact with gaseous or liquid substances, interactive forces oc-
cur which can result in a way that these substances are bonded to the solid. This effect
is called adsorption. The strength of these bonds can differ from component to compo-
nent, which can be sufficient to achieve a selective separation. Especially microporous
solids with a high specific surface, corresponding to a high capacity, can be used as ad-
sorptive agents. For the separation, besides the different equilibrium load steric (sieve
effect) and kinetic effects (different diffusion coefficients) can be used as well.
Especially the development of improved adsorptive agents (e.g. molecular sieves)
and better regeneration techniques increased the relevance of adsorption as a thermal
separation process [89]. Whenever the separation factor of the vapor-liquid equilib-
rium is close to 1 (azeotropes, isomers), when difficult process conditions would have
to be realized (high or low temperatures) or only small amounts of impurities have to
be removed (waste water, exhaust air), adsorption has advantages in comparison with
distillation. On the other hand, adsorption always means that the process becomes
discontinuous. The adsorption unit is saturated after a time, and a regeneration has
to take place. During this time, a second bed (twin plant) must take over until regen-
eration of the first bed has been finished.
Adsorption is used for the drying of gases and solvents, for the removal of con-
densable components (CO2 , H2 O, hydrocarbons) upstream the air separation, for nat-
ural gas conditioning, separation of nitrogen or oxygen from air, separation of hydro-
carbon mixtures, and the treatment of waste water and exhaust air (Chapter 13.4.6).
A number of adsorbents have been developed for various applications. Due to
pores in their structure, they have enormous surfaces per volume (up to 1000–
1500 m2 /g), leading to a correspondingly large adsorption capacity. These mate-
rials are manufactured by degradation reactions of solids, where fluid reaction
products are formed and removed immediately. If the reaction temperature is be-
low the melting point of the solid, the crystal cannot sinter together, and the holes
and pores remain. The diffusion of the adsorbed substance inside the pores is usu-
ally the step which determines the necessary residence time. Examples for common
adsorbent materials are activated carbon, silica gel, clay gel and zeolites, where the
latter ones act as molecular sieves.
The adsorptive agent (adsorbent) should have a high selectivity and a high capac-
ity. Adsorption of water (except for drying purposes) and polymerizing components
must be avoided. As well, a low effort for the regeneration is desirable. There are
hydrophilic (e.g. silica gel, aluminium oxide, zeolites) and hydrophobic adsorptive
agents (e.g. activated carbon, carbon molecular sieves). Activated carbon is a very
inexpensive adsorbent which is used especially for the removal of hydrocarbons or
nonpolar components in general from waste water. Its mechanical stability is limited,
and it has a tendency to cause fires. On the other hand, activated carbon is so cheap
that regeneration can often be omitted; it can be directly sent to incineration.
Zeolites (molecular sieves) are crystalline aluminosilicates of alkali or alkaline
earth metals. They have defined cavities and pore diameters. The pore diameters
are between 0.3–0.8 nm. The well-defined structure can be used for the separation
of molecules of different size or shape, e.g. the separation of linear and branched
alkanes or m- and p-substituted aromates. A frequently applied option is the removal
of water from gases and solvents with a so-called KA zeolite (cation = K = potas-
sium, pore diameter is 3 or 4 Å). A very common application is the separation of the
ethanol/water azeotrope (see below).
To keep the dimensions of the adsorber low, adsorptive agents must have a large
surface. It is the inner surface which is decisive. An adsorbent particle is porous; it is
distinguished between macropores (d > 50 nm), mesopores (d = 2–50 nm) and mi-
cropores (d < 2 nm). The large specific surface is mainly caused by the great number
and good accessibility of the micropores. In Table 7.3, ranges for the specific surface
of the various adsorptive agents are given.
Table 7.3: Specific surfaces of various adsorptive agents [89].
Adsorptive agent Specific surface (m2 /g)

Activated carbon, general 300–2500
Activated carbon, narrow pores 750–850
Silica gel, wide pores 300–350
Aluminium oxide 300–350
Zeolites (molecular sieves) 500–800
Carbon molecular sieves 250–350
Adsorption processes for a particular component can be characterized by their adsorp-

tion isotherms, i.e. the relationship between the adsorbed amount and the concentra-
tion in the fluid phase at a certain temperature. There are five particular types, which
can be physically interpreted. These adsorption isotherms can hardly be estimated,
which means that the quality of an adsorption process cannot be predicted without
references or experiments. The adsorption equilibrium between the concentrations
of a component in the fluid phase (adsorptive) and in the phase at the surface of the
adsorbent (adsorbate) is decisive for the choice of the adsorbent and the design of the
adsorption column. The amount adsorbed per g adsorbent depends on the tempera-
ture, on the partial pressure or, respectively, the concentration, and the kind of the
adsorbent, including the manufacturing process (size of the inner surface) and the
history (aging, regeneration).
There are a number of equations for the isothermal adsorption equilibrium of pure
substances. Many of them are based on the Langmuir approach, where some sim-
plifications were made (homogeneous surface, no interaction between the adsorbed
molecules):
ni Ki pi
= , (7.6)
n i,mon 1 + K i p i
where n i,mon is the load for the limiting case of a monomolecular layer.
For the correlation of multicomponent adsorption isotherms, there is at least some
theory, which is similar to the procedure for correlating VLE. However, a prediction
method like UNIFAC is still missing. Furthermore, for technical applications the pa-
rameters characterizing the adsorbent like specific surface, pore distribution, crystal
irregularities, and interactions with the adsorbed species are often not reproducible,
not to mention the kinetics and mass transfer effects. More detailed information about
adsorption isotherms is given and derived in [142].
It is obligatory that adsorption equilibria be measured, and this is usually a large
effort. There are five types of adsorption isotherms (Figure 7.5) [89]. For type I, a
monomolecular layer is formed. This behavior can be described with Equation (7.6).
For types II and IV, more layers are formed, and condensation in the pores takes place.
For types III and V there is no tendency to form a monomolecular layer.
In technical applications, there are usually multicomponent mixtures, where the
components involved compete for the space on the adsorbent surface. There are sim-
ilar phase equilibrium diagrams as there are for vapor-liquid equilibria [89]. Even
azeotropic points occur. Adsorption is exothermic; as a first guess, it is a good ap-
proach to assume that the enthalpy of adsorption is approx. 1.5 times the enthalpy
of vaporization of the adsorptive [142]. Shortcut approaches and recommended con-
straints for the design of adsorbers are explained in [142].
In Figure 7.6, the adsorption procedure is illustrated. The feed enters the adsorp-
tion bed at the top of the column. The adsorption itself takes place in the adsorption
zone, which proceeds through the column towards the outlet of the adsorber with
time. Different components have different adsorption zones. The stronger the com-
ponent is adsorbed, the slower the saturation zone moves towards the outlet of the
adsorber. The component with the weakest adsorption can pass the adsorption bed
and be obtained in pure form.
When the saturation zone of the component to be removed approaches the end of
the adsorption bed, it is necessary to stop and switch over to a second adsorption bed
to avoid a breakthrough. Therefore, adsorption units usually consist of twin columns
(Figure 7.7). Meanwhile, as long as the second column is in operation, the loaded col-
umn must be regenerated. For the regeneration the flow direction is reversed. Regen-
eration can be done by increasing the temperature (temperature-swing adsorption,
TSA) or lowering the pressure (pressure-swing adsorption, PSA), by replacement with
6.0 30.0
5.0 I III
Benzene Bromine
4.0 20.0
n mmol/g
n mmol/g
3.0
2.0 10.0
Silica gel Silica gel
1.0 T = 303.15 K T = 352 K
0.0 0.0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
30.0 30.0
II IV
Water Water
20.0 20.0
n mmol/g
n mmol/g
10.0 Al2 O3 10.0 Silica gel
T = 303.15 K T = 313 K
0.0 0.0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
pi/psi
30.0
V Activated carbon
Water T = 298 K
20.0
n mmol/g
10.0
0.0
0.0 0.2 0.4 0.6 0.8 1.0
pi/psi
Fig. 7.5: Adsorption isotherm types [89].

front of
saturation
adsorption
zone
Fig. 7.6: Course of an adsorption with time [89].

purified vapor
from heater
loaded vapor
to condenser
Fig. 7.7: Typical arrangement of an adsorption twin plant.
another component or by lowering the partial pressure of the pollutant in the gas be-
ing in contact with the adsorbent. The latter is possible by flushing the adsorber with
an unloaded gas (Figure 7.7). Combinations are possible; one of the most often applied
procedures is the flushing with steam, where the partial pressure is lowered and the
temperature is elevated. Of course, the adsorptives then have to be removed from the
steam in a further step, which makes the whole process more complex. As usual, it
takes a large effort to remove the last traces of the adsorptives in the bed; therefore,
a residual load after regeneration is accepted, which in turn of course reduces the
capacity of the next adsorption cycle. After regeneration, some time should be taken
into account to get back to adsorption conditions, e.g. by cooling or repressurizing the
bed after TSA or PSA, respectively.
On a technical scale, adsorption has the disadvantage of being in principle a dis-
continuous process. It is aspired to operate the adsorption continuously in counter-
current flow like other thermal separation processes. However, countercurrent flow
can hardly be realized with a solid because of its attrition. Several attempts have been
made to overcome this difficulty. The most popular one is the so-called “simulated
moving bed” (SMB), invented by UOP (Universal Oil Products Inc., Des Plaines, Illi-
nois). The principle is explained in the following paragraph.
SMB is a continuous chromatography process in countercurrent flow with a bi-
nary system. The difference between adsorption and chromatography is that in chro-
matography the mobile phase achieves the desorption. Both components appear at
the outlet of the column, they are separated due to the different time they need for
passing the column. In fact, in SMB the adsorbent is not really moved. The movement
of the solid phase is replaced by changing the position of the inlet and outlet streams
in a cyclic way. An inlet stream is continuously split into two outlet streams, which
consist of purified components if the SMB is adequately designed. The difficulties of
the SMB are the mechanical complexity and the complicated design.
Figure 7.8 shows a case where the column is fixed. The feed enters the column in
the middle. Both components pass the column with different velocities and separate.
If the column itself moved in the opposite direction of the mobile phase with a velocity
that is between the velocity of the two components, the components appear to move
in different directions (Figure 7.9).
To get the SMB arrangement, the mobile phase has to flow in a closed loop. The
products are withdrawn at defined places with the exact volume flow. As mentioned
above, the thought experiment of the moving column is replaced by the movement of
the inlet and outlet nozzles (Figure 7.10).
Feed
Fixed column
Fig. 7.8: Normal chromatography arrangement

Mobile phase with a fixed adsorbent.
Feed
Column
Fig. 7.9: Separation due to the movement

Mobile phase of the column.
Mobile phase recycle

Feed
Column
Mobile phase Fig. 7.10: Principle of the simulated moving bed.
One of the most popular applications of adsorption is the dehydration of ethanol in

bioethanol production. Ethanol and water have an azeotrope (Figure 5.39), which can-
not be split into the pure components by simple distillation. On the other hand, there
is a strict specification for bioethanol concerning its water content. Therefore, other
techniques must be applied (Chapter 5.7). Pressure-swing adsorption (PSA) can be a
way, using a 3 Å molecular sieve. The ceramic pellets are shown in Figure 7.11. The wa-
Fig. 7.11: 4A molecular sieve pellets.

© Smokefoot/Wikimedia Commons/CC BY-SA 4.0.
https://creativecommons.org/licenses/by-sa/4.0/.
ter molecules can diffuse through the pores, whereas the larger ethanol molecules are
retained [143]. At the outlet of the adsorption bed, the water is more or less completely
removed. Due to the heat of adsorption, the temperature of the bed is strongly ele-
vated. The effect is even used for process control, as the temperature indicates where
the saturated zone is. Once the mass transfer zone reaches the outlet of the bed, re-
generation starts, and it is switched over to a second bed. Desorption is done by first
applying a vacuum to the tower. To remove the remaining water, the adsorbent bed
is purged with purified ethanol vapor in opposite flow direction, i.e. the vapor enters
the column from the opposite side at the bottom. Figure 7.12 shows the block diagram
of the process [143]. In column K1, distillation of the raw ethanol is performed. A va-
por stream close to azeotropic concentration (approx. 96 wt.%) is taken at the top of
the column. It passes the adsorber A1 from the top to the bottom. Downstream the
adsorber, it is condensed at the shell side of the falling film evaporator W2, serving
as a heating agent. The generated steam is transformed to a higher pressure by a jet
pump (Chapter 8.3) and used for direct steam heating (Chapter 13.1) in column K1. The
mixture of ethanol and water vapor from the regeneration of the adsorber bed A2 is
condensed in heat exchanger W1 and led back to the distillation column.
The dehydration of the ethanol–water azeotrope is a popular, but not really a typ-
ical application of molecular sieves. The amount of water being handled is quite large.
Normally, the water concentration of the streams being treated is a few hundred ppm;
here, it is approx. 4 %. Therefore, the bed is loaded rapidly, and the cycle times are
pretty short, just in the range of minutes. Often additional beds are used to provide
enough time for regeneration.
Ethanol / water
azeotropic
Ethanol
A2 vapor
Raw ethanol A1
W2
Steam
W1
K1
Water Ethanol
pure liquid
Fresh
water
Fig. 7.12: Block diagram for ethanol dehydration [143].
7.3 Crystallization
Crystallization should in fact not be called an alternative separation processes, as it is

in fact the oldest one. The common thing of the alternative processes in this chapter is
more or less that they cannot be designed on a theoretical basis, and that “something
solid” is involved.
In crystallization, we can distinguish between crystallization from a solution,
which is often applied for purifying inorganic salts, and crystallization from a melt,
which is often used for purifying organic substances. Like in distillation, in crystal-
lization energy for the cooling or evaporation of the solvent is necessary to create a
second phase. Because of the low density difference, the separation of the two phases
solid-liquid is not as easy as it is for the vapor-liquid separation in distillation. Also,
the transport of the solid phase is difficult. Often, the viscosity is high, making in turn
the mass transfer of the crystallizing component difficult.
Crystallization has advantages in comparison with the other thermal separation
options, especially distillation, if the components to be separated have a low thermal
stability or a low (or even no) vapor pressure. As well, it can have advantages if the
separation factor is close to 1, e.g. for azeotropes or for the separation of isomers. Crys-
tallization can be used to get extremely pure products. In most cases, crystallization
takes place in form of eutectic systems, and in this case the crystallizing component is
pure and can be obtained by melt crystallization with one separation stage. In practi-
cal applications, the separation of the solid and the liquid phase is not perfect, so that
inclusions of mother liquor will occur in the solid phase. This phenomenon depends
mainly on the crystal formation and growth. Regularly formed crystals are useful to
achieve a good separation of the two phases.
The decisive thermodynamic issue for the description of condensation is solid-
liquid equilibrium (SLE, Chapter 2.6). In the case of eutectic systems, a pure solid
phase is obtained, but it is a disadvantage that only part of this component can crys-
tallize. The remaining mother liquor has eutectic concentration and leads to mixed
crystals if crystallization is continued. This limitation can be overcome if crystalliza-
tion is combined with other thermal separation processes.
For the design of crystallizers an exact knowledge of the solid-liquid equilibria
with respect to temperature is necessary. For most of the salts, the solubility increases
with temperature. For some inorganic salts, the solubility can decrease with tempera-
ture. These are the so-called hardness components. Examples are gypsum (CaSO4 ) or
calcium carbonate (CaCO3 ).
Also, the kinetics of the seed crystal formation and of the crystal growth are im-
portant for the equipment design in crystallization.
An oversaturation is necessary for the formation and the growth of crystals. It can
be achieved in different ways, e.g. by cooling, evaporation of the solvent or depressur-
ization, which is another way of evaporating the solvent. Furthermore, crystallization
can be forced by the addition of a new component. Oversaturation by cooling has
advantages if the solubility increases strongly with temperature. If the temperature
dependence is less significant, an oversaturation by evaporation might be favorable.
For the seed crystal formation, there are several mechanisms. Crystals can be
formed at rough surfaces or impurities or by abrasion of small crystals from larger
ones. Many constraints like oversaturation and flow velocity have an influence. The
more the solution is subcooled, the more seed crystals are formed. However, due
to the increase of the viscosity with decreasing temperature the rate of seed crystal
formation decreases after passing through a maximum. The following relationship
between seed crystal formation rate r and oversaturation (∆c) has been found:
r = ∆c b , (7.7)
where b = 3–6.
A similar equation can be set up for the the crystal growth:
r = ∆c w , (7.8)
where w = 1–2.
The size of the crystals strongly depends on the degree of the oversaturation. Seed
crystal formation and crystal growth are competing processes. A large oversaturation
promotes the seed crystal formation, giving small crystals. Therefore, the oversatura-
tion has to be kept small, if large crystals are the target. The control of the oversatura-
tion is decisive for crystallization processes.
There are several options for the choice of equipment for industrial crystalliza-
tion processes. In suspension crystallizers, the crystals are dispersed in the solvent
or in the melt, respectively. The heat of fusion is transferred to the liquid. Suspension
crystallizers are operated continuously. It is tried to get separate zones for the oversat-
uration and the crystal growth. All oversaturation mechanisms can be applied. In the
following paragraph, evaporation is taken as example.
Usually, crystals are heavier than the mother liquor. To keep them in the suspen-
sion, there must be an upward flow in the crystallizer so that the crystals are located in
definite layers according to their size. There are various types of crystallizers to realize
this principle.
The one most widely used is the forced circulation crystallizer (FC). It is normally
operated under vacuum conditions. As can be seen in Figure 7.13, the suspension is
circulated with a pump through a heater, causing evaporation in the upper part of the
vessel. The concentration of the dissolved solids rises, and precipitation takes place.
The slurry can be continuously removed from the vessel. The FC crystallizer is appro-
priate if crystal size is not an issue. There is no mechanism to redissolve small crystals.
Larger crystals can be obtained with the DTB (draft tube baffle) crystallizer [144].
As the FC, it is operated under vacuum or a slight overpressure. It is provided with a
skirt baffle which forms a partitioned settling zone. Inside the baffle there is a vertical
draft tube, where the feed and the recycle are directed to (Figure 7.14). Outside the
skirt baffle, the mother liquor containing the small crystals is withdrawn and led to a
heater, where the small crystals have a chance to be dissolved again. At the top of the
crystallizer, vapor is generated, giving the desired oversaturation. The formed crystals
can settle down to the product discharge at the bottom. In comparison with the FC
crystallizer, the internal loop shows less attrition and crystal breakage; large crystals
which were formed are maintained.
The largest crystals are obtained in the Oslo type crystallizer, where the crystals
are grown in a fluidized bed (Figure 7.15). The growth is limited by the residence time.
There is again an external recirculation loop with a heat exchanger, where the temper-
ature is elevated. The loop reenters the crystallizer near the top. Evaporation can take
place, giving the oversaturation. The oversaturated solution is led to the bottom of
the crystallizer, where it first comes into contact with the larger crystals, so that these
crystals can further grow instead of forming small new ones. At the bottom, the prod-
uct is withdrawn. A classified bed is formed above, with the lowest concentration at
the nozzle for the recirculation outlet. In the Oslo type crystallizer, hardly any attrition
and crystal breakage occurs [144].
Layer crystallizers operate discontinuously. They have a cooled wall where the
crystallization takes place. They are used for melt crystallizations, either as falling
film crystallizers or as static crystallizers.
Falling film crystallizers work analogously to falling film evaporators (Figure 4.18).
The liquid runs down the inner side of the tube bundle. The tube bundle is cooled with
a heat transfer fluid on the shell side. The crystallization begins at the tube wall. The
melt is recycled, as long as the required amount of liquid has been crystallized. After
the liquid has run out of the apparatus, the solid layer is slightly heated up to remove
Vapor
Tangential entry
of circulation loop
Active
volume
Steam
Heater Product
Condensate discharge
Feed
Circulation
pump Fig. 7.13: Forced circulation crystallizer.
Agitator Vapor
system Settling zone
Skirt Mother liquor
baffle + fines
Draft Steam
tube
Active
volume Condensate
Product Feed
discharge
Fines
recycling
loop
Elutriation liquid Fig. 7.14: DTB crystallizer.
Vapor
Circulation loop Vaporizer
Steam
Weir baffles
Classified bed
active volume
Heater
Product
Condensate discharge
Feed
Circulation
pump Fig. 7.15: Oslo type crystallizer.
impurities in the surface layer. Finally, the whole crystal layer is melted and removed
from the heat exchanger. At static crystallization, cooling elements dip into the melt.
By varying the temperature the particular steps as described above can be carried out.
All these crystallizers work very reliably, as there are no moving parts or mechanical
devices for the removal of the liquid. However, the residence times are quite long,
giving large equipment volumes.
More information about crystallization can be obtained from [145] and [146].
8 Fluid flow engines
8.1 Pumps
There is no doubt that pumping is a science on its own. In most engineering units,
there is a “rotating equipment” department which is dedicated to the selection of the
appropriate pump in a thankworthy way. Otherwise, vendor companies usually give
assistance. However, a process engineer must be able to specify what the pump should
do in the process. The following chapter will introduce the fundamental terms for the
specification of a pump from a process engineering point of view. The explanations
refer to centrifugal pumps, which are the most common ones (80–90 % of all the pump
applications).
In process simulation, pumps usually do not play a decisive role. The pressure de-
pendence of the enthalpy of a liquid is neglected anyway, so the power consumption of
the pump is transferred into a slight temperature elevation, normally less than 1 K. The
exact arrangement of the source and the target vessel is determined later in the project,
as well as the necessary pressures and the pump characteristics, which is decisive for
the pump efficiency. Therefore, the power consumption of the pump is not accurately
calculated; at most, the result shows the correct order of magnitude. The calculation
of the power is performed according to the same scheme as for compressors, where it
is much more important. It is explained in Chapter 8.2.
A pump conveys a liquid from one piece of equipment, usually a vessel, to another
one. Between the two locations, a pressure and/or a height difference and the pressure
drop in the connecting line has to be overcome. Figure 8.1 shows an example of a
principle sketch of this situation.
For the arrangement in Figure 8.1, the pressure elevation by the pump ∆ppump can
be determined by the Bernoulli equation
p0 + ρgh0 − ∆pinlet line + ∆ppump − ∆poutlet line − ρgh1 = p1 , (8.1)
where the indices 0 and 1 refer to start and end of the whole line, respectively. The pres-
sure drops of inlet and outlet line comprise the line pressure drops (Equation (12.1)),
the pressure drops caused by special piping elements (Equation (12.18)) and the pres-
sure drops through control valves, which can often only be more or less set arbitrarily
(e.g. ∆pCV = 1 bar)¹. The terms for the kinetic energy are neglected, as the velocities
upstream and downstream the pump do not change very much. The pressure elevation
by the pump is usually converted into the delivery head Hpump
∆ppump
Hpump = . (8.2)
ρg
1 as long as the valve is not fully specified, which takes place in the detail engineering phase.
224 | 8 Fluid flow engines
pmax = 10 bar
H2 = 18 m pnorm = 9 bar
HLL = 3000
pmin = 8 bar
NLL = 2500
Leq2 = 30 m
LLL = 2000
p = 1 bar
H1 = 3 m
Leq1 = 6 m ∆p = 1 bar
Fig. 8.1: Example for a setup of a pumping process.
Analogously, the remaining terms in Equation (8.1) can be converted into heads. The
delivery head Hpump is a function of the volume flow; the pump characteristics curve
shows how the delivery head decreases with increasing volume flow (Figure 8.2). On
the other hand, with increasing volume flow the pressure drops of inlet and outlet
line increase, whereas the differences in head and the pressures p0 and p1 in the
equipment remain constant, which is called the plant characteristics. The situation
can be illustrated by drawing the pump and the plant characteristics in one diagram
(Figure 8.2). In a given arrangement, the operation point is defined by the intersec-
tion between the curves of pump and plant characteristics. Usually, this operation
point does not fit the requirements of the process. In this case, the plant characteris-
tics can be manipulated by throttling the flow with a control valve (Figure 8.2). If this
is not possible, the pump characteristics can be changed as well, e.g. by changing
the blade wheel diameter or by changing the number of revolutions per minute with
a frequency converter. It should be emphasized that this well-known construction is
Plant characteristics
with valve throttling
Pump characteristics
H Operation point
with valve
throttling Operation point
Static Plant characteristics
height
. Fig. 8.2: Typical curvatures of pump
Q and plant characteristics.
8.1 Pumps | 225
just an illustration; in practical applications, it is sufficient to define the requirements

(volume flow, delivery head) for the pump and provide a control device to adjust the
plant characteristics.
Besides V̇ and Hpump the so-called NPSH value (net positive suction head) is an
important operating parameter of a pump. It is relevant to avoid cavitation, which is
the worst failure of a circulation pump. Cavitation means that vapor bubbles occur
in the pump because of a local reduction of the static pressure below the saturation
pressure of the process liquid. These bubbles violently implode when they are trans-
ported into regions in the pump with higher pressures and therefore cause erosion,
often leading to the mechanical destruction of the pump. Moreover, the mechanical
stress on the pump impeller, the shaft, the seals and the bearings is increased.
The NPSH value indicates how far the medium inside a pump is away from its
saturation pressure in a static case, i.e. without movement in the pump. In the case of
Figure 8.1 the NPSH value would be calculated to be
ρw2
p0 + ρgh0 − ∆pinlet line − 2 − ps
NPSH = , (8.3)
ρg
meaning that NPSH is the difference between the total pressure inside the pump and
the saturation pressure of the liquid, transformed into a height. This NPSH value must
be greater than a minimum value², which has to be determined experimentally by the
vendor. A safety margin of 0.5 m should be kept. It depends on the type and construc-
tion of the pump and on the operating conditions. For boiling liquids with low flow
velocities (i.e. pressure drop and kinetic term are negligible), Equation (8.3) reduces
to
NPSH = h0 . (8.4)
In case of serious difficulties in maintaining the necessary NPSH value, special pumps
are available which need very low NPSH values and are even capable of conveying
liquids at their boiling point [147].
From the process engineering point of view, the specification of the pump can ex-
clude the pump itself, by just defining the states upstream and downstream the pump
using Equation (8.1). Then the pump specialist can chose an appropriate pump ac-
cording to his knowledge about the necessary ∆ppump and the corresponding mass
flow with its various physical properties. Usually, we distinguish between normal,
maximum, and minimum case. The minimum case should be defined by the most fa-
vorable conditions for the pump, i.e. the maximum level in the vessel upstream the
pump, the lowest mass flow (i.e. lowest pressure drop in the line) and the minimum
level in the target vessel, whereas the maximum case is just the other way round. It is
then up to the pump specialist to decide which pump type can cover this load range,
and which pump efficiencies are achieved.
2 called “NPSH value of the pump”.

Example
In the exemplary arrangement in Figure 8.1, methanol (ṁ = 12000 kg/h, t = 30 °C, ρ = 782 kg/m3 ) is
transferred from a vessel to a distillation column. Some items of the specification are evaluated:
(a) the necessary NPSH value of the pump;
(b) the normal pressure difference to be built up by the pump;
(c) the maximum pressure difference to be built up by the pump;
(d) the maximum power consumption of the pump.
In the sketch, the equivalent lengths of inlet and outlet line are given, meaning that all the bends,
elbows etc. are already included. The pipe diameters is d1 = 4′′ for the inlet line and d2 = 3′′ for the
outlet line. The pressure drop of the pipe can be calculated with Equation (12.1):
ρw 2 L
∆p = λ ,
2 d
where, for simplicity, for the friction factor λ a standard value of λ = 0.03 is used. For the valve, a
pressure drop of ∆p = 1 bar shall be assumed. The efficiency of the pump is η = 0.7.
Solution
First, the velocities in the pipes and the pressure drops are calculated:
4ṁ 4 ⋅ 12000 kg/h

w1 = = = 0.526 m/s
ρπd21 782 kg/m3 ⋅ π(4 ⋅ 25.4 mm)2
4ṁ 4 ⋅ 12000 kg/h
w2 = = = 0.935 m/s .
ρπd22 782 kg/m3 ⋅ π(3 ⋅ 25.4 mm)2
The pressure drops in the lines are
ρw 2 Leq1 782 kg/m3 ⋅ (0.526 m/s)2 6 m

∆p1 = λ = 0.03 = 192 Pa
2 d1 2 4′′
ρw 2 Leq2 782 kg/m3 ⋅ (0.935 m/s)2 30 m
∆p2 = λ = 0.03 = 4035 Pa .
2 d2 2 3′′
(a) For the NPSH value, the low liquid level (LLL in Figure 8.1³) is relevant. The static liquid head is
the sum of the height of the tangent line H1 and the low liquid level LLL. The saturation pressure
of methanol at t = 30 °C is ps = 0.219 bar. According to Equation (8.3) we get
ρw 2
p0 − ∆pinlet line − − ps
NPSH = h0 + 2
ρg
782 kg/m3 ⋅ (0.526 m/s)2
1 bar − 192 Pa − − 0.219 bar
= 5m + 2
3 2
782 kg/m ⋅ 9.81 m/s
= 15.1 m .
3 Numbers in technical drawings are in mm if no unit is given.

8.1 Pumps | 227
(b) The normal pressure elevation by the pump is calculated acc. to Equation (8.1), using the normal
liquid level (NLL) and the normal operation pressure of the column pnorm . For the outlet line,
the pressure drop of the valve must be added to the value obtained above. Solved for ∆ppump ,
Equation (8.1) reads
∆ppump = p1 − p0 + ∆pinlet line + ∆poutlet line + ∆pvalve + ρg(h1 − h0 ) ,
giving
∆ppump,norm = 9 bar − 1 bar + 192 Pa + 4035 Pa + 1 bar
+ 782 kg/m3 ⋅ 9.81 m/s2 ⋅ (18 − 5.5) m
= 10.0 bar .
(c) For the maximum pressure elevation, the low liquid level (LLL) and the maximum operation pres-
sure of the column pmax are taken as input. One gets
∆ppump,max = 10 bar − 1 bar + 192 Pa + 4035 Pa + 1 bar

+ 782 kg/m3 ⋅ 9.81 m/s2 ⋅ (18 − 5) m
= 11.04 bar .
(d) The maximum power consumption of the pump is

̇ pump,max
m∆p
Pmax = V̇ ∆ppump,max /η =
ρη
12000 kg/h ⋅ 11.04 bar
= .
782 kg/m3 ⋅ 0.7
= 6.7 kW
Most companies have their own guideline for the installation of a pump, depending
on reliability demands, the control philosophy, and the type of the pumps. Figure 8.3
shows an example with the most important features. Two pumps are installed in par-
allel, so that in case of failure of the operating pump a switch to the additional one can
be performed immediately, perhaps even automatically. It should be mentioned that
in this arrangement even the inlet pipe can be exposed to the high outlet pressure
generated by the pump. In case the nonoperating pump is not isolated by closing
the valves up- and downstream the pump, the operating pump will convey liquid
backwards through the nonoperating one and pressurize even its inlet line, which
is normally exposed only to the low inlet pressure. There is the so-called minimum
bypass line branching from the product line, ending in the vessel containing the feed
of the pump. The reason is that most pump types should not operate against a closed
valve. If the pressure in the outlet line exceeds a certain value, the control valve in
the bypass line opens so that further pressure build-up is inhibited. Furthermore, it
is prevented that the temperature of the system increases, as the motor power of the
pump is no longer removed, which could as well lead to damage of the system. Instead
of the control valve, an orifice can act as the restriction in the bypass line. This is only
acceptable for pumps with relatively low power consumption. The minimum bypass
is certainly an energy waste, as it simply reduces the pressure of a stream which has
just been built up. The orifice admits a bypass stream even at normal operation. For
large pumps, this would correspond to a significant energy waste, whereas for small
pumps it might be acceptable, as the cost of a control valve can be saved. Furthermore,
the typical safety sensors can be seen in Figure 8.3, i.e. vibration or, in this case, tem-
perature sensors, which are linked to an interlock that switches off the pump and, in
most cases, switches on the substitute pump.
Fig. 8.3: Typical for a pump installation.
The flow of a pump can in principle be controlled in two ways: First, the minimum
bypass can lead some of the flow back to the source vessel. As discussed above, this
option consumes electrical energy, as the pump simply provides its maximum flow
according to its characteristics. The second option, the use of a frequency converter,
is more elegant but also more expensive. It makes it possible to set the rotation speed
of the pump according to the demand for the volume flow.
8.1 Pumps | 229
There are three types of pumps.

– Centrifugal pumps:
Centrifugal pumps (Figure 8.4) have been thoroughly discussed above. The oper-
ation of centrifugal pumps is illustrated in Figure 8.5. The rotating impeller trans-
fers its rotational energy to the liquid, which is accelerated and discharged into
the casing due to centrifugal forces. When the casing area increases, the kinetic
energy of the liquid is converted to pressure. Centrifugal pumps are used for large
volume flows with moderate pressure heads. They are appropriate for low to mod-
erate viscosities. An undissolved vapor fraction up to 5–7 vol.% can be tolerated
in the liquid, however, with increasing vapor fraction the efficiency and the NPSH
value (cavitation!) are decreasing. Also, the solid content should be limited, 8 %
can be regarded as the maximum.
Fig. 8.4: Centrifugal pump

in standard arrangement [148].
Fig. 8.5: Functional principle of a centrifugal pump [148].

– Oscillating displacement pumps:

The most frequently used oscillating displacement pumps are piston pumps and
membrane pumps (Figure 8.6). In principle, they work discontinuously; for the
piston pump there is a well-defined intake stroke, where the piston generates an
underpressure to suck in the liquid to be conveyed. For the outlet stroke, the piston
Piston Pump
Outlet valve
Inlet valve
Hydraulic fluid
Outlet valve Membrane Pump
Membrane Inlet valve
Fig. 8.6: Operation modes of piston and membrane pump.
is moved back and generates an overpressure on the liquid which pushes it out
of the pump. Membrane pumps work in an analogous way, but the piston does
not get in direct contact with the conveyed liquid. Instead, the piston is actuat-
ing a working fluid which moves a membrane to and fro. Membrane pumps are
especially useful for corrosive fluids. Unlike centrifugal pumps, oscillating dis-
placement pumps are appropriate for moderate volume flows at high pressure
generation. The discontinuous operation can be overcome if necessary. The use of
several pump stages where the phases are displaced can yield a quasi-continuous
flow. Another option is the installation of a pressure vessel filled with pressurized
gas connected to the outlet line. The gas will be further compressed when the
outlet stroke takes place; during the inlet stroke, the gas expands and represents
an additional pressure source. The pump characteristics are completely different
from that of a centrifugal pump. After setting the repetition frequency, the volume
flow is determined, and the pressure obtained depends only on the plant charac-
teristics.
– Rotating displacement pumps:
Well-known rotating displacement pumps are gear pumps (Figure 8.7). The cogs
represent small compartments which are continuously filled with liquid at low
pressure and moved to the high pressure level. Comparably high pressure eleva-
tions up to 40 bar are possible. As for oscillating displacement pumps, the volume
flow is directly proportional to the rotation speed. Gear pumps are especially ap-
propriate for highly viscous fluids.
Fig. 8.7: Basic sketch of a gear pump.

© Duk/Wikimedia Commons/CC BY-SA 3.0. https://
creativecommons.org/licenses/by-sa/3.0/deed.en.
8.2 Compressors
Compressors, vacuum pumps, fans, and other fluid flow engines for the pressure ele-
vation of gases are widely applied in industry for the transport of fluids or for estab-
lishing a certain pressure to perform a reaction or a separation. In process simulation,
compressors cannot be regarded as a simple flash, as they cannot be specified by two
outlet variables. Instead, more information about the course of the change of state is
necessary. For most types of compressors, it can be assumed that they are adiabatic,
i.e. the heat exchange with the environment does not play a major role. Proceeding
from this assumption, the calculation route is illustrated by the adiabatic compres-
sion of a vapor. The changes in kinetic energy can be neglected in the energy balance.
The calculation is divided into the reversible adiabatic calculation and the integration
of losses.
1. Reversible calculation:
The reversible case characterizes the process that requires the lowest power con-
sumption. It is specified by the outlet pressure p2 at constant entropy. According
to the Second Law, the outlet temperature is calculated by the isentropic condition
s2 (T2rev , p2 ) = s1 (T1 , p1 ) , (8.5)
where the indices 1 and 2 denote the inlet and the outlet state, respectively. In
process simulation calculations, Equation (8.5) is directly evaluated with the cor-
responding equation of state to determine T2rev . A simplified calculation using
the ideal gas equation and assuming a constant heat capacity yields [11]
κ−1
T2rev p2 κ
=( ) , (8.6)
T1 p1
with
cid
p
κ= . (8.7)
cid
v
The specific power consumption for the reversible case is given by
w t12rev = h2rev (T2rev , p2 ) − h1 (T1 , p1 ) . (8.8)

2. Integration of losses:
The actual specific power consumption required is calculated with the isentropic
and mechanical efficiency:
w t12rev
w t12 = , (8.9)
ηth ηmech
while the power consumption of the process is
̇ t12 .
P12 = mw (8.10)
The isentropic efficiency ηth is an empirical factor which summarizes all the ef-
fects about the irreversibility of the process. ηth = 0.8 is often a reasonable choice.
It usually decreases with increasing pressure ratio. ηmech is the efficiency of the
energy transformation of the compressor engine (electrical to mechanical energy),
which is not related to the process flow. For large drives, ηmech = 0.95 can be used.
The outlet conditions of the flow are then calculated backwards via
h2rev (T2rev , p2 ) − h1 (T1 , p1 )
h2 (T2 , p2 ) = h1 (T1 , p1 ) + . (8.11)
ηth
Note that ηmech has been omitted. Knowing h2 , the outlet temperature T2 =
f(h2 , p2 ) can be calculated by an iterative procedure.
Example
Steam (5 t/h, p1 = 1 bar, t = 110 °C) is compressed adiabatically to p2 = 5 bar. The efficiency of the
compressor is given by ηth = 0.8, while the mechanical efficiency is supposed to be ηmech = 0.9. Use
a high-precision equation of state, e.g. [29].
Solution
The first step is always the reversible calculation using Equation (8.5). It gives the condition
s2 (T2rev , 5 bar) = s1 (383.15 K, 1 bar) = 7.4155 J/(g K) .
The solution is T2rev = 560.55 K = 287.4 °C. The power required for the reversible case is
̇ 2rev − h1 ) = ṁ [h(560.55 K, 5 bar) − h(383.15 K, 1 bar)]

W t12,rev = m(h
= 5000 kg/h ⋅ (3038.50 − 2696.34) J/g
= 475.2 kW .
To obtain the real consumption of the compressor, the efficiencies must be considered. One gets
W t12,rev
W t12 = = 660.04 kW .
ηth ηmech
For the outlet state of the steam, the mechanical efficiency has no influence; only the thermal efficiency
must be taken into account:
h2rev − h1 3038.50 − 2696.34
h2 − h1 = = J/g = 427.70 J/g ,
ηth 0.8
giving
h2 = 427.71 J/g + 2696.34 J/g = 3124.04 J/g .
At p2 = 5 bar, the outlet temperature T2 can be determined to be T2 = 601.91 K or t2 = 328.76 °C.
There are a number of different compressor types available, which differ in pressure
range, volumetric flow rate, and other requirements at operating conditions like pro-
cess safety, physical properties, or environmental conditions. Moreover, it has to be
considered whether the fluid contains drops or particles and whether there are com-
ponents which tend to polymerize.
Certainly, if the achievable compression ratio is too low, several compressors can
be combined into a “multistage compressor”, usually with intermediate coolers or di-
rect liquid injection to reduce the gas temperatures and, correspondingly, the volume
flows.
The main compressor types are:
– Piston compressors:
Piston compressors work according to the same principle as piston pumps (Fig-
ure 8.6). Compression ratios up to 6 : 1 per stage can be achieved. There are no lim-
itations for the volume flow, but relatively small ones are preferred (≈ 200 m3 /h).
The lubrication of the compressor is always a major item. Care should be taken
that no process components can accumulate in the lubricant, which would reduce
the effectiveness and make it necessary for the lubricant to be exchanged often.
Alternatively, dry-running compressors can be used if appropriate. Of course there
are special requirements for the inlet and outlet valves. The main disadvantage of
piston compressors is their high maintenance effort. The flow pulsation can cause
vibration and structural problems due to the unbalanced forces, making heavy
foundations necessary. Figure 8.8 shows a so-called hyper compressor, used in
the LDPE process for compressing ethylene from approx. 300 bar to 3000 bar in
two stages. The space demand is huge; a hyper compressor often takes up a whole
hall and needs extremely strong fundamentals. In the middle of the picture there
is the driving shaft with its coupling to the motor and the lubrication unit, caus-
ing the movement of the plungers on the right hand and the left hand side. The
arrangement of the plungers is symmetric to avoid unbalanced forces as far as
possible. The nozzles below the plungers are the gas inlets and outlets.
– Membrane compressors:
Membrane compressors have a principle similar to membrane pumps (Figure 8.6).
They are appropriate for small volume flows and achieve compression ratios up
to 20 : 1 per stage. Like piston compressors, the flow pulsation causes problems,
which makes it necessary to replace the membrane regularly.
Fig. 8.8: Two-stage hyper compressor.

© 2016. Burckhardt Compression AG.
– Screw compressors:
Screw compressors are displacement compressors, where the medium is enclosed
in a chamber which is continuously shortened, causing the compression of the gas
(Figure 8.9). Compression ratios of 4.5–7 : 1 can be achieved, the range of the vol-
ume flows is reported to be 300–60 000 m3 /h. Screw compressors are not sensible
to small amounts of liquid or dirt in the gas stream; on the contrary, liquid is often
introduced to reduce the temperature elevation. No valves are involved, which is
a great advantage in comparison to piston compressors. Furthermore, screw com-
pressors have a high efficiency and a wide range of applications.
Fig. 8.9: Screw compressor rotor.

© MAN Diesel & Turbo SE.
– Rotary piston compressors:

Rotary piston compressors are displacement compressors which are usually ap-
plied for vacuum generation. The rotary pistons and the shell form moving cham-
bers which force the gas to the pressure side. The compression ratios are compa-
rably low (1.8–2 : 1), while the volume flows are not restricted (100–80 000 m3 /h).
The function principle is analogous to the rotary vane pump (Figure 8.18).
– Turbo compressors:
1. Radial turbo compressors:
Radial turbo compressors are the corresponding compressor to centrifugal
pumps (Figure 8.5). The compression ratio is limited (2–4 : 1), while the vol-
ume flows (5000–150 000 m3 /h) can be considerably high. However, larger
or smaller volume flows cause construction problems which are not easily
solved. The wide operation range and the high reliability are the advantages
of radial turbo compressors; their drawbacks are that they are sensitive to re-
duced flow rates, changes in gas composition, and dynamic problems.
2. Axial turbo compressors:
Axial turbo compressors (Figure 8.10) are appropriate for very large volume
flows up to 1 200 000 m3 /h and compression ratios up to 8 : 1 [149]. The ex-
tremely large capacity and the high reliability are the main advantages of this
compressor type, and the main disadvantage is the limited turndown. The
smaller the volume flow, the more ineffective this compressor type is. A lower
bound for axial compressors is approximately 60 000 m3 /h.
Fig. 8.10: Axial turbo compressor rotor.

© MAN Diesel & Turbo SE.
Gas
Gas/Liquid
Gas mixture
Seperator
Liquid ring
gas pump Outlet for
liquid
Circulating liquid
Service liquid Fresh liquid
Fig. 8.11: Liquid ring compressor arrangement.

Courtesy of Sterling Fluid Systems Holding GmbH.
– Liquid ring compressors (Figure 8.11):

Liquid ring compressors (LRCs) are compressors in which the service fluid forms
a liquid ring, which itself both acts as a compressor and as a seal between suction
and discharge side. The shaft of the impeller is placed eccentrically so that the
cells containing the gas formed by the impeller and the liquid become smaller
and smaller, which results in the compression. Moreover, the compressor power
which heats up the gas is transferred to the liquid, and thus the compression can
be regarded as isothermal. The peculiarity of LRCs when compared to other com-
pressor types is that the medium to be compressed comes into direct contact with
the service liquid. Normally liquid ring compressors operate as displacement com-
pressors. If the vapor condenses due to the compression or due to the temperature
decrease after contact with the liquid, the compression is supported by absorption
or condensation. The liquid is circulated; cooling is necessary as the heat gener-
ated by the compression (and possibly by the condensation/absorption) increases
the temperature of the liquid. The strength of LRC compressors is that any medium
can be compressed, there is no restriction for the medium to be explosive, toxic
or carcinogenic [150]. The max. compression ratio for one stage is 3 : 1 (in vac-
uum operation up to 7 : 1), the volume flows range from 1–20 000 m3 /h [150]. LRCs
are very robust and inexpensive, but they have a low efficiency (ηth ≈ 0.35). In
case the liquid absorbs components from the gas stream it has to be continuously
worked up to avoid accumulation.
The characteristic curves of axial and radial turbo compressors and displacement
compressors are outlined in Figure 8.12.
140
Displacement
Axial turbo compressor
120 compressor
% Pressure
Radial turbo
100
compressor
80
60 80 100 120 Fig. 8.12: Qualitative characteristic curves

% Volume flow of different compressor types.
Mechanical vapor recompression is one of the main tasks of compressors. For this pur-
pose, the thinking for application is different. Figure 8.13 shows a typical arrangement.
Electric
power Feed
Steam
Condensate
Fig. 8.13: Mechanical vapor recompression

Product arrangement.
The vapor used as heating agent for evaporation is the generated vapor itself. However,
this vapor would condense at most at the same temperature as the evaporation itself
takes place, and only if a pure substance is vaporized⁴. Because of pressure drops in
the line and in the nozzles and because of the boiling point elevation the condensation
temperature of the vapor is certainly lower than the boiling temperature of the liquid.
Nevertheless, the vapor can in fact be used as a heating agent if its condensation tem-
perature is elevated. This can be achieved by increasing its pressure, and this is what
mechanical vapor recompression can do.
4 which would not make any sense.

In terms of thermodynamics, the power required for the compression is used to

elevate the temperature level of the heat of condensation of the vapor. This means
that the mechanical power is at least not lost; it is used for heating the medium as
well. However, the power from the electric current is turned into heat, which is a de-
valuation.
The blowers used for mechanical recompression are usually simple industrial fans
(Figure 8.14). They accomplish a compression ratio of approx. 1.3–1.4 per stage. This
corresponds to an elevation of the dew point of 8–10 K. Usually, this should be suf-
ficient to cover the boiling point elevation of the product, especially if plate heat ex-
changers are used, which need only low driving temperature differences. If not, there
is also the option to use an arrangement of two or even three blowers in series. Blow-
ers are highly standardized. The power can be adjusted by manipulating the rotation
speed, which is usually achieved with a frequency converter. The costs for mainte-
nance are low, especially compared to other compressor types. Special care must be
taken for the design of the bearings and for the shaft seals.
Fig. 8.14: Blower for mechanical vapor recompression.

© Piller Blowers & Compressors GmbH.
Example
Saturated water vapor coming from an evaporator (0.8 bar, 93.5 °C) is compressed by a blower with a
pressure ratio of 1.35. What will be its condensation temperature?
Solution
The outlet pressure of the blower is pnew = 0.8 bar ⋅ 1.35 = 1.08 bar. The corresponding condensa-
tion temperature of the outlet stream is then 101.8 °C, meaning that the compression has achieved a
boiling point elevation of 8.3 K.
The energy consumption of the blower is usually moderate. Many t/h of water vapor
can be recompressed, saving the same amount of fresh steam (e.g. 100 t/h in the ex-
ample above with a power consumption of approx. 2 MW).
The inlet of a compressor should in general have a vapor fraction of 1, because
liquid droplets can cause erosion to the impeller. It is necessary to carefully check
what the demands of the compressor are in this area. The vapor-liquid separator can
often be designed for a maximum droplet size (Chapter 9); however, there is no way
to predict the amount of droplets. The various types of compressors have different
susceptibilities to droplets; the least sensitive ones are in fact the blowers for vapor
recompression, where liquid is even injected beyond the saturation level on purpose
to continuously clean the impeller and avoid deposits on the impeller, which could
lead to imbalances.
Getting into the two-phase region during the compression becomes more probable
[11] the larger the molecule is⁵. According to Equation (8.6) for the ideal gas case, the
temperature rises during compression. As
κ−1 1 cid
v R
= 1 − = 1 − id = id , (8.12)
κ κ cp cp
one can easily see that the exponent decreases with increasing molar isobaric heat ca-
pacity, which means that the temperature elevation during compression is lower, the
larger the molecule is⁶. If the rise of the boiling temperature is larger than the temper-
ature elevation during compression, drops are formed inside the compressor, which
might be seriously detrimental to the impeller [11]. If the molecule has at least four
C-atoms, one can almost be sure that droplet formation will occur (“wet fluids”), for
molecules with less C-atoms, the feed stream will remain gaseous (“dry fluids”) [151].
5 This paragraph refers to pure components as representatives.

6 cid
p on molar basis increases with the size of the molecule, as more vibration options can be acti-
vated.
8.3 Jet pumps
A jet pump is an alternative to a compressor for the compression of gases, the genera-
tion of vacuum or for the transportation of liquids or even bulk materials. Its advantage
is its high reliability, caused by the fact that a jet pump has no moving parts. Also, it
is not sensitive to fouling or corrosion and appropriate for large volume flows [152].
The energy is transferred by a fluid under high pressure, e.g. steam, compressed air
or water and other liquids. The principle of jet pumps is explained in Figure 8.15. The
energy is supplied by the motive steam on the left-hand side. It makes use of the prin-
ciple of the Laval nozzle, where supersonic velocities can be reached in a tube [153].
The motive steam passes a narrowest crossflow area (Chapter 12.1.3), where the speed
of sound is reached. Downstream it is further accelerated in the diffusor to supersonic
velocity. Due to the acceleration, the static pressure is lowered below the pressure of
the suction stream. The suction stream is taken in and mixed with the motive steam
in the first part of the diffusor. In the second part, the flow is slowed down again, and
the pressure rises. Finally, at the outlet the flow has a pressure between the pressure of
the suction and the motive steam. The motive steam has expanded while the suction
stream has been compressed; the jet pump works like an equivalent system where the
motive stream is expanded in a turbine, while the power obtained is used to run a
compressor for the suction stream.
Motive
flow (1) Motive nozzle
Head Outlet diffuser
Steam
chest Inlet diffuser Mixed
discharged
flow (3)
Suction Fig. 8.15: Scheme of a jet pump.
flow (2) © 2015, Körting Hannover AG.
One of the most popular applications of jet pumps is the compression of low pressure
steam to a more useful pressure using high-pressure steam as the motive steam. As
long as the motive steam has a pressure which has to be reduced anyway, no operation
costs are related to the compression. For assessing whether a jet pump might be useful
it is necessary to know how much of the motive steam is needed. The use of jet pumps
in thermal vapor recompression has been shown in Chapter 3.2.
Clearly, the final decision on the necessary amount of the motive stream should
be made by the vendor. Nevertheless, some first guesses can easily be made. The min-
imum possible amount of the motive stream is determined by the entropy balance.
According to the Second Law, the entropy of the outlet stream (3) must be larger than
the sum of the entropies of the suction (2) and the motive stream (1):
8.3 Jet pumps | 241
ṁ 1 s1 + ṁ 2 s2 ≤ (ṁ 1 + ṁ 2 )s3 , (8.13)
while at the same time the First Law
ṁ 1 h1 + ṁ 2 h2 = (ṁ 1 + ṁ 2 )h3 (8.14)
must be obeyed, provided that the velocities at the nozzles do not have a major con-
tribution to Equation (8.14). From Equations (8.13) and (8.14), one can get a first idea
of the order of magnitude of the motive steam ṁ 1 .
More realistic values can be obtained considering the efficiency [152]
ṁ 2 [h(T3 , p3 ) − h(p2 , s3 )]
ηjet = ≈ 0.2 … 0.4 , (8.15)
ṁ 1 [h(T1 , p1 ) − h(p3 , s1 )]
which refers to the above mentioned turbine/compressor analogy.
Example
1000 kg/h water vapor at t2 = 130 °C, p2 = 2 bar shall be used at p3 = 4.5 bar. How much motive steam
(t1 = 190 °C, p1 = 11 bar) is necessary (a) using Equation (8.13) for the reversible case and (b) using
Equation (8.15) with an efficiency of ηjet = 0.4 for a more realistic case?
Solution
(a) First, the specific enthalpies and entropies are determined. Using the high-precision equation of
state [29] one obtains
h1 = 2796.6 J/g s1 = 6.5868 J/(g K)
h2 = 2727.3 J/g s2 = 7.1797 J/(g K) .
Estimating ṁ 1 = 1000 kg/h, h3 can be determined to be h3 = 2762.0 J/g according to Equa-
tion (8.14). The corresponding specific entropy s3 (p3 , h3 ) turns out to be 6.8997 J/(g K), which is
larger than the value obtained from Equation (8.13), 6.8832 J/(g K). Thus, the estimation for ṁ 1
was too high. After several iterations, the result is ṁ 1 = 916.3 kg/h, giving h3 = 2760.4 J/g and
s3rev (p3 , h3 ) = 6.8962 J/(g K), which is obtained with Equation (8.13) as well.
(b) The enthalpies involved in Equation (8.15) are
h(T1 , p1 ) = 2796.6 J/g

h(p3 , s1 ) = 2630.0 J/g .
Again, an iterative solution is necessary. Estimating ṁ 1 = 2000 kg/h, one gets h(T3 , p3 ) =
2773.5 J/g according to Equation (8.14) and h(p2 , s3 ) = 2627.4 J/g. ηjet is then determined to
be 0.438, indicating that the estimation for ṁ 1 was too low. After some iterations, ηjet = 0.4 is
obtained with ṁ 1 = 2194 kg/h, giving h(T3 , p3 ) = 2774.9 J/g and h(p2 , s3 ) = 2628.7 J/g.
The operational characteristics of a jet pump can be summarized in the diagram ac-
cording to [152] (Figure 8.16).
ṁ2 = f(p3)
at p2 = const.
p1 = const. p2 = const.
plim (e.g. 10 mbar)
ṁ2 = f(p2)
at p2 ≤ plim
ṁ2 kg/h
Fig. 8.16: Operational characteristics

of a jet pump [152].
p2 p3 Reproduced with permission.
In this diagram, the suction stream for a given jet pump is depicted as a function of
suction pressure and outlet pressure, both as coordinates on the abscissa. The pres-
sure p1 of the motive steam remains constant. The left border line refers to the suction
pressure. There is a minimum suction pressure to obtain a suction flow at all. Above
this suction pressure, the suction flow increases continuously with the suction pres-
sure, with a typical sharp bend in the curve shape. Moving to the right-hand side of the
diagram, the pressure on the abscissa refers to the outlet pressure. The outlet pressure
can be varied over wide ranges without an effect on the suction stream (see horizontal
lines). At a certain outlet pressure, which slightly depends on the suction pressure, the
suction flow is seriously affected and drops down rapidly. The goal is to avoid getting
into this region. For a given jet pump, the flow of the motive steam can only be in-
creased by increasing its pressure. This measure has little effect on the suction stream,
but the critical outlet pressure is moved to higher values of the outlet pressure [152].
Jet pumps are very often used as vacuum pumps, which is analogous to the use
as vapor compressors. The pressure ratio is usually between 15–20. The main pres-
sure range where jet pumps are used is between p = 0.1–100 mbar. For this purpose,
several stages are used [154]. Between the stages there are condensers to get rid of
the condensables as far as possible to reduce the load of the jet pumps. A very good
explanation of the design criteria of jet pumps can be found in [155].
8.4 Vacuum generation
Vacuum is divided into the following pressure ranges:

– rough vacuum: 1 –1000 mbar;
– medium vacuum: 10−3 –1 mbar;
– high vacuum: 10−7 –10−3 mbar;
– ultra-high vacuum: < 10−7 mbar.
Before introducing the particular options for vacuum generation, some general
thoughts might be useful. In process engineering, the main part of vacuum gener-
ation is covered by condensation. This works in the following way [154].
Consider a vessel which contains water vapor at p = 1000 mbar, t = 110 °C (Fig-
ure 8.17). At the beginning, the piston at the top is completely movable so that the
content of the vessel is in mechanical equilibrium with the environment. Next, the
piston is fixed, and the whole vessel cooled down to t1 = 20 °C, while the steam par-
tially condenses. In this case, the pressure will drop down to the vapor pressure of
the new temperature p s (t1 ) ≈ 23 mbar. Thus, using simple condensation, a vacuum
of 23 mbar has been generated. A better vacuum can be generated if the temperature
is further lowered, for instance 12 mbar at 10 °C.
p = 1 bar
t = 110° C
Fig. 8.17: Condenser used as vacuum generator.
Things are different when inert gases are involved. Consider the case above, where
the partial pressure of water is only pwater = 950 mbar and the remaining 50 mbar
are caused by an inert gas, e.g. nitrogen. This causes the vacuum to deteriorate sig-
nificantly. While the water is still condensed until it develops its vapor pressure at
t1 = 20 °C, the inert gas will remain gaseous. Neglecting the volume of the conden-
sate and the small solubility of the inert gas, its partial pressure will only decrease
due to to the temperature decrease according to the ideal gas law, giving
293.15 K
pinert ≈ 50 mbar ⋅ ≈ 38 mbar .
383.15 K
The overall vacuum will then be p = 23 mbar + 38 mbar = 61 mbar, by far worse than
the 23 mbar previously obtained. This is a typical problem in vacuum process engi-
neering. Most vapor streams to be condensed contain a certain fraction of inert gas.
After condensation, the fraction of the inert gas will have increased, as in the exam-
ple above from xinert = 1000
50
= 0.05 to xinert = 38
61 ≈ 0.62. To avoid accumulation of the
inert gas, the remaining vapor mixture has to be removed, and this is essentially what
vacuum pumps are for. As mentioned above, the fraction of condensables is decreased
with decreasing temperature. Therefore, it is useful to remove the inert gas at the cold-
est part of the condenser, i.e. at the inlet of the cooling agent. This is another reason
to realize countercurrent flow in the condenser, apart from maintaining the driving
temperature difference.
The condenser is the more economic part of vacuum generation; in [154] it is
demonstrated with an example that condensation should be used as far as possible,
and the vacuum pump should only remove the remaining gas.
There are certain types of vacuum pumps:
Jet pumps have been explained above in Chapter 8.3. They are the most inexpen-
sive alternative, as they contain no movable parts. On the other hand, their flexibility
is limited. Liquid ring compressors (Chapter 8.2) can also be used as vacuum pumps
and are the most widely used types of vacuum pumps in chemical industry. They
are robust, simple and inexpensive, but not very efficient concerning the energy con-
sumption. They can achieve a vacuum down to 30 mbar. Also, piston and membrane
compressors can also be utilized as vacuum pumps [154].
Rotary vane pumps are the most common types if it is ensured that all gases
sucked in can be transported by the vacuum pump without condensation. Otherwise,
besides a worse performance the lubrication oil of the pumps will be spoilt by dilu-
tion or by forming an emulsion with insoluble liquids as water. Compatibility with
condensing substances is a strong criterion to distinguish whether an application is
appropriate for rotary vane or for liquid ring pumps. The function principle of a rotary
vane pump is illustrated in Figure 8.18. Inside the stator, an excentric rotor is rotating.
In the center of the rotor, a spring pushes two vanes apart so that they get contact with
the wall of the stator and form two separated chambers at inlet and outlet. At the inlet
side, the volume of the corresponding chamber increases, so that substance from the
recipient is sucked in. At the outlet, the volume of the chamber has decreased, which
pushes the gas to the outlet line. Rotary vane pumps can achieve high vacuum down
to 10−3 mbar. There are similar types working according to the displacement principle
(rotary piston pump, roots blower).
For the generation of high and ultra-high vacuums, oil diffusion and turbomolec-
ular pumps are used. Both need a considerable prevacuum to keep the load low. Oil
diffusion pumps (Figure 8.19) have the same functional principle as jet pumps. A
high-boiling oil with a low vapor pressure is heated up electrically so that it finally
evaporates. The vapor exits the pump through a special system of nozzles with ve-
locities above speed of sound, and, correspondingly, low local pressures. The gases
0°
3
4
3
90° 270°
2
1
180°
Fig. 8.18: Rotary vane pump. 1: stator, 2: rotor, 3: vanes, 4: spring.

© Rainer Bielefeld/Wikimedia Commons/CC BY-SA-3.0. https://creativecommons.org/licenses/by-
sa/3.0/deed.en.
Recipient
Cooling
To
jacket
pre-vacuum
Pump oil
Heater
Fig. 8.19: Scheme of an oil diffusion pump.
rapidly diffuse to the oil. After the oil has been condensed at the wall, they are removed
by the pre-vacuum pump. The advantage of oil diffusion pumps is their reliability, as
they have no moving parts. The can produce even an ultra-high vacuum down to
10−10 mbar. Care should be taken that no oil can flow into the recipient.
Turbomolecular pumps consist of a series of rotor/stator pairs which act as small
compressors. The rotation speed is in the range 300–400 m/s or 10 000–90 000 rpm.
It is possible to achieve ultra-high vacuum down to 10−10 mbar.
For the dimensioning of vacuum pumps reasonable values for their capacity are
necessary. It is mainly determined by the leakage rate, which can be estimated accord-
ing to empirical rules of thumb. There are several approaches:
– For flange connections, the leakage rate can be estimated to be 200–400 g/h per m
gasket length. Special measures (tongue and groove face flange, surface treatment
of the gasketing areas, use of special gaskets) can reduce the leakage rate down
to 50–100 g/(h m).
– According to the maximum mass flux density (Chapter 14.2), through an opening
of 1 mm2 approx. 0.83 kg/h can flow. Surprisingly, this value does not depend on
the vacuum pressure. As long as the pressure ratio (p. 358) between environmen-
tal pressure and recipient is above the critical pressure ratio of approx. 2, meaning
that the vacuum is at least below p = 500 mbar, the air intake is independent of
pressure, and the maximum flow is determined by the speed of sound in the nar-
rowest cross-flow area.
– The leakage air flows can be determined according to Table 8.1 [156].
Table 8.1: Recommended values for leakage rates [156].
Connection type
Equipment volume Flange Flange and welded Welded or special gaskets
(m3 ) Leakage rate (kg/h)
0.2 0.15 – 0.3 0.1 – 0.2 < 0.1
1 0.5 – 1 0.25 – 0.5 0.15 – 0.25
3 1 – 2 0.5 – 1 0.25 – 0.5
5 1.5 – 3 0.7 – 1.5 0.35 – 0.7
10 2 – 4 1 – 2 0.6 – 1.2
25 4 – 8 2 – 4 1 – 2
50 6 – 12 3 – 6 1.5 – 3
100 10 – 20 5 – 10 2.5 – 5
200 16 – 32 8 – 16 4 – 8
500 30 – 60 15 – 30 8 – 15
9 Vessels and separators
Vessels are normally the simplest pieces of equipment in a process, as long as they
are not used as reactors. They have several functions in a process. The most important
one is the decoupling of two process parts, which is especially important in the startup
phase when different parts of a plant are starting operation independently from each
other. Other functions are the separation of vapor and liquid by gravity or the provi-
sion of suction head to achieve a sufficient NPSH value for a pump (Chapter 8.1). If the
only function of a vessel is the storage of raw materials or product, it is called a tank.
For mixing the content of the vessel, agitators are used. For a quick overview on
this topic, the book of Baerns et al. [8] is recommended.
Figure 9.1 shows a typical PID representation of a horizontal vessel. The vessel
is equipped with several nozzles for inlet and outlet streams, measurements for tem-
perature, pressure and liquid level. There is a man hole on the right-hand side. An
extension at the bottom enables a possible second liquid phase (probably water) to
separate even if the amount of this phase is low. A vortex breaker at the liquid outlet
nozzle prevents waterspouts to be formed. Two of the feed inlets are designed as dip-
pipes to prevent electrostatic charging. Furthermore, a dip-pipe ensures that vapor
backflow is not possible. The liquid in the vessel forms a liquid seal so that vapor
from the vessel cannot flow back through these lines if no liquid is delivered.
Fig. 9.1: PID representation of a typical vessel.

248 | 9 Vessels and separators
The calculation of the volume of vessels is a bit complicated, due to some apparently
well-meant simplifications. In fact, some guidelines define that the volume of the ves-
sel is the cylindrical volume calculated with the outer diameter of the vessel and the
length of the cylindrical section. Specially shaped heads ensure that there are no sharp
edges where solids could deposit. Half-sphere heads need too much space and have a
bad accessibility. Flat heads would provide more space for nozzles, but the tensions re-
quire larger wall thicknesses. Therefore, flat heads are rarely used for pressure vessels
or for large vessels. The volume of the two heads is neglected, which is conservative,
but partly compensated for due to the fact that the outer diameter is used instead of
the inner one. Also, it is possible to calculate an exact volume, e.g. using the volume
of an ellipsoidal head
Vel. head = 0.1298 d3i . (9.1)
Before beginning to evaluate the vessel volume, it should be clarified which conven-
tion is to be used in the particular project. For the definition of the liquid level, the rela-
tionship between liquid volume and liquid level is trivial for vertical but sophisticated
for horizontal vessels. With the liquid height shown in Figure 9.2, the relationship is
LD2 2h D
VL = arccos (1 − ) − L ( − h) √ Dh − h2 , (9.2)
4 D 2
where L is the cylindrical length of the vessel. When the volume demand of a vessel is
calculated, it should be considered that only part of it can be used as working volume,
usually 65–80 %. The vessel must not be completely filled with liquid, due to sudden
pressure rise when the temperature slightly increases.
h Fig. 9.2: Cross-section of a horizontal vessel

partly filled with liquid.
The dimensions of a vessel are usually determined by providing response time for the
operator. In Figure 9.1 various levels are indicated. LN is the normal level which is
aspired to be kept by the level controller. If the level drops or rises too much, an alarm
goes off to attract the attention of the operator. In Figure 9.1, these levels are called
LAL (level alarm low) and LAH (level alarm high), respectively. The level switch LSHH
(level switch high high) usually actuates an interlock which prevents the vessel from
being overfilled. LSHH is chosen in such a way that a certain part of the vessel volume
siphon
breaker
Fig. 9.3: Example of an overflow nozzle.
remains free, e.g. 10 %. Similarly, the LSLL (level switch low low) sets off another in-
terlock, which e.g. might protect the pump transferring the liquid from the vessel by
switching it off. Between the alarm and the automatic response of the process control
system there must be enough time for the operator to react, e.g. 1 min for an activity in
the control room or 5 min for an activity in the plant area. This defined period of time
mainly determines the size of the vessel; the volumes between LAL and LSLL, and LAH
and LSHH, respectively, must correspond to the feed and the withdrawal during the
specified response time for the operator.
The last line of defense against excessive vessel filling is the overflow nozzle,
which is often used in atmospheric tanks [157]. Usually, there is a pipe attached lead-
ing to the bottom. In case the tank is blanketed at the top, this overflow nozzle must
not be an opportunity for the blanket gas to escape from the tank. This is the reason
why the pipe is led down below the liquid level on the internal side (Figure 9.3). Also,
a siphon breaker makes sense for preventing the overflow stream from emptying the
vessel after the liquid level has dropped below the threshold for the overflow nozzle
[157]. A siphon breaker is a piece of pipe located at the highest point of the piping and
connected to atmosphere.
For the determination of the nozzle sizes, the following rules of thumb can be
applied:
In the general case with a two-phase entry, the inlet nozzle should obey the con-
dition
ρav w2av ≤ 1500 Pa , (9.3)
where the average velocity and the average density can be calculated according to
ṁ V + ṁ L
ρav = (9.4)
V̇
V̇
wav = (9.5)
A
ṁ V ṁ L
V̇ = + , (9.6)
ρV ρL
with A as the cross-flow area of the nozzle. The vapor outlet nozzle should have the
same size as the adjacent pipe as long as the condition
ρV w2V ≤ 3750 Pa (9.7)
is maintained. The recommended velocity is wV = 10 m/s. For the liquid outlet nozzle,
the criterion is
ρL w2L ≤ 400–900 Pa . (9.8)
The velocity should be kept below wL = 1 m/s. Even for low flows, the minimum di-
ameter should be 50 mm.
Another function of vessels is the separation between vapor and liquid phase.
There are several kinds of vapor-liquid separators; the vessel is a so-called gravity
separator, making use of the principle that vapor and liquid droplets have a different
density. Separators of this kind are usually built as vertical vessels; although horizon-
tal ones are possible [158], they hardly occur in chemical industry. A simple equilib-
rium of forces between weight, buoyancy and flow resistance leads to the limiting gas
velocity for a given droplet diameter d [158, 159]
4 g(ρL − ρV )dlim
wV = √ , (9.9)
3 ρV cw
where the cw correlation according to Brauer [160]
cw = 24/Re + 4 Re−0.5 + 0.4 (9.10)
can be used. The Reynolds number is defined as
wV dρV
Re = . (9.11)
ηV
It should be noted that in Equation (9.10) the values for the physical properties must
be taken for the flowing phase, that is, the gas which is flowing around the droplets.
The velocity must be determined iteratively, as it is also part of the Reynolds num-
ber. The iteration is quite easy with the mathematical methods of current comput-
ers; nevertheless, Equation (9.9) has always been subject to simplifications. Some are
usually justified; there is no major objection to use only the first summand in Equa-
tion (9.10) for the laminar region where Re < 0.25, which gives a direct relationship be-
tween droplet diameter and limiting velocity. However, it was experienced that many
companies derive their own correlations, which are restricted to their special appli-
cation cases and filled with empirical factors. This cannot be recommended, as the
applications are changed more often than the correlation.
The meaning of Equation (9.9) is that for a given vapor velocity wV , which results
from the diameter of the vessel, droplets larger than dlim are separated, while smaller
ones are not. In fact, any application case has a droplet distribution, and it will not
Table 9.1: Recommended values for the limiting droplet diameter [159].
Application Limiting droplet diameter

Standard 0.2 mm
Compressor or turbine inlet 0.15 mm
Dryer inlet, prevent loss of solvent 0.1 mm
Not decisive for process 0.35 mm
happen that the separation is performed in such a rigorous way. Therefore, the inter-
pretation that 50 % of the droplets with the limiting droplet size will be separated from
the gas flow makes more sense. From process engineering experience, rules of thumb
can be given for what a reasonable limiting droplet size might be in a particular case
(Table 9.1).
Nevertheless, one should be aware that the limiting droplet diameter is not ex-
actly what the engineer needs. Although the concept is more or less accurate from
the mathematical point of view, the recommended values are just a rough guideline.
Its limitation becomes clear when it has to be specified how much liquid is entrained,
e.g. to specify the COD value (chemical oxygen demand, p. 323) of a condensate. There
is currently no way to determine the amount of liquid droplets and their size distribu-
tion for a given arrangement. Figure 9.4 shows the main dimensions of a vessel without
demister.
The efficiency of the droplet separation can be increased with a so-called demis-
ter, a wire mesh layer placed in the vapor space of the separation vessel (Figure 9.5).
In contrast to the gravity separator, a high vapor velocity is advantageous in that the
droplets hit the wire mesh and do not pass around the wires. Therefore, demisters need
lower vessel diameters, making them more inexpensive than gravity separators. How-
ever, one must check that no fouling or even polymerization of the separated droplets
occurs.
From experience, there are some recommendations for the design of vessels which
are used as vapor-liquid separators with or without demister. They should be applied
together with the recommendations concerning the residence time given above. Rec-
ommended values for the dimensions are given in Figures 9.4 and 9.6. The height of
a demister is between 100 and 150 mm. Larger heights only slightly improve the sep-
aration but also cause an additional pressure drop proportional to their height. The
design velocity can be set as
ρL − ρV 0.5
weff = 0.7K ∗ [ ] , (9.12)
ρV
where the default value for the constant is K ∗ = 0.11 m/s. For high pressure or high
vacuum, K ∗ = 0.06 m/s should be used. The velocity should not go below wmin =
0.3 weff . It can be expected that the limiting droplet size is between 3 and 5 µm. Fig-
ure 9.6 shows the main dimensions of a vessel with demister.
Vapor outlet
HV = 1.2 D, min. 0.9 m
Feed
H
Hnozzle = 0.3 D,
min. 0.3 m
HL
Fig. 9.4: Sketch of a vertical separator

Liquid outlet without a demister.
Fig. 9.5: Wire mesh demister.

© ENVIMAC Engineering GmbH.
Vapor outlet
Demister
HV = 1.1 D, min. 1 m
Feed
H = 0.3 D,
Hnozzle
min. 0.3 m
HL
Fig. 9.6: Sketch of a vertical separator

Liquid outlet with a demister.
Fig. 9.7: Knock-out-drum and cyclone [161].

Several other types of droplet separators are used in the chemical industry, such as
baffle separators (“knock-out drum”) or cyclones (Figure 9.7), where the droplets are
settled out by centrifugal forces. More details can be found in [161].
In case the vessel has an agitator, it is strongly recommended to provide baffles to
prevent that the liquid is rotating in a whole and forms spouts due to the centrifugal
forces. The H/D ratio for stirred vessels is normally in the range 1–1.5, for large volumes
there is a tendency to use slender and high vessels, as the power required for agitation
is proportional to d5 .
Many applications require heating or cooling of the vessel content. The standard
approach is to perform the heat transfer across the vessel wall. The simplest way is
that the heating or cooling agent is transported in half-pipe coils. It is inexpensive,
but in pressurized vessels there is a poor heat transfer because of the wall thickness.
Because of the welding seams, only approx. 2/3 of the vessel wall can be used as heat
transfer area. With a jacket, more area can be provided, but the design is a bit more
complicated. Often, it is not effective to transfer the heat across the vessel wall. Coils or
other internals can be placed inside the vessel, getting significantly more heat transfer
area and, possibly, a better heat transfer coefficient. The most effective way are exter-
nal heat exchangers, which are not restricted in their dimensions by the vessel itself.
In this case, however, a pump is necessary to operate the cycle.
10 Chemical reactions
10.1 Reaction basics
For the author, this will be the most difficult chapter of the book. The topic easily jus-
tifies books on their own, e.g. [8] or [162], and can definitely not be covered in only
a few pages. The goal of this chapter is only to explain the most important technical
terms so that the reader is able to follow the discussions of practitioners.
The extent of a chemical reaction is characterized by the conversion (X). It must
be defined which of the reactant components the conversion refers to. Then, the con-
version is
number of reacted moles of reference reactant
X= .
number of moles of reference reactant at the beginning of the reaction
In contrast, the yield (Y) of a reaction refers to a product of the reaction. It must be
taken into account that the maximum number of product moles depends on the stoi-
chiometry:
number of product moles formed
Y= .
number of reactant moles ⋅ stoichiometric ratio
The stoichiometric ratio is defined as
number of product moles in reaction equation
stoichiometric ratio = .
number of reactant moles in reaction equation
The quality of a reaction can be characterized by the selectivity (S). It is defined as
number of product moles formed
S= ,
number of converted reactant moles ⋅ stoichiometric ratio
where again the numbers refer to a special product and reactant, respectively.
Example
In the process of oxychlorination of ethylene by hydrogen chloride and oxygen, giving 1,2-dichoro-
ethane and water, there is a reaction network of several competing reactions. The most important
ones are:
(1) C2 H4 + 0.5 O2 + 2 HCl 󳨀→ CH2 Cl–CH2 Cl + H2 O;
(2) C2 H4 + HCl 󳨀→ CH3 –CH2 Cl;
(3) C2 H4 + 2 O2 󳨀→ 2 CO + 2 H2 O;
(4) C2 H4 + 3 O2 󳨀→ 2 CO2 + 2 H2 O.
The (fictitious) conversions of ethylene with respect to the particular reactions shall be 90 % (1),
3 % (2), 2 % (3), 3 % (4). The rest of the ethylene will remain¹. The stream entering the reactor consists
of 100 mol/h C2 H4 , 190 mol/h HCl and 200 mol/h O2 .
1 In the real oxychlorination, the selectivity for 1,2-dichloroethane and the surplus of ethylene are by
far larger [163].
256 | 10 Chemical reactions
Calculate the overall conversions of ethylene, HCl, and O2 and the yields and selectivities of
CH2 Cl–CH2 Cl with respect to ethylene and to HCl.
Solution
In Table 10.1, the composition of the stream after the reaction is calculated. Any stoichiometric calcu-
lation should be checked for consistency in the atom balance. In this case, at the inlet there are
– 100 ⋅ 2 = 200 C-atoms;
– 100 ⋅ 4 + 190 = 590 H-atoms;
– 190 Cl-atoms;
– 200 ⋅ 2 = 400 O-atoms.
At the outlet there are

– 2 ⋅ 2 + 90 ⋅ 2 + 3 ⋅ 2 + 4 + 6 = 200 C-atoms;
– 2 ⋅ 4 + 7 + 90 ⋅ 4 + 3 ⋅ 5 + 100 ⋅ 2 = 590 H-atoms;
– 7 + 90 ⋅ 2 + 3 = 190 Cl-atoms;
– 142 ⋅ 2 + 4 + 6 ⋅ 2 + 100 = 400 O-atoms. 󳨀→ o.k.
The overall conversions are

100 − 2
X C2 H4 = = 98 %
100
190 − 7
XHCl = = 96.3 %
190
200 − 142
X O2 = = 29 % .
200
The yields of 1,2-dichloroethane referring to ethylene or, respectively, HCl can be calculated to be
90
YC2 H4 Cl2 ,C2 H4 = = 90 %
100
90 2
YC2 H4 Cl2 ,HCl = = 94.7 % .
190 1
Finally, the selectivities with respect to ethylene and HCl are
90
SC2 H4 Cl2 ,C2 H4 = = 91.84 %
98
90 2
SC2 H4 Cl2 ,HCl = = 98.36 % .
183 1
The speed of chemical reactions vary. While the corrosion of iron is slow and can
take years, other reactions take minutes or hours, and reactions like the neutraliza-
tion of acids and bases take place instantaneously. Performing reactions in industrial
processes often requires increasing or slowing down the reaction rates. Therefore,
knowledge of the influences on reaction kinetics is essential for a process engineer.
An in-depth knowledge in reaction kinetics can be found in the above mentioned text-
books; for the explanation of the basics only homogeneous reactions are regarded.
The reaction of the formation of ammonia can be used as an example [164]:
N2 + 3 H2 󳨀→ 2 NH3 .
Table 10.1: Calculation of the mole numbers at the reactor outlet.
Comp. Inlet Reac. (1) Reac. (2) Reac. (3) Reac. (4) Outlet
C2 H4 100 −90 −3 −2 −3 2
HCl 190 −180 −3 0 0 7
O2 200 −45 0 −4 −9 142
CH2 Cl–CH2 Cl 0 90 0 0 0 90
CH3 –CH2 Cl 0 0 3 0 0 3
CO 0 0 0 4 0 4
CO2 0 0 0 0 6 6
H2 O 0 90 0 4 6 100
The reaction rate in this case can be set up as
dcNH3
= kcN2 c3H2 , (10.1)
dτ
where c is the volume concentration (c-concentration)
c i = n i /V = x i ⋅ ρ . (10.2)
The c-concentration is not popular in chemical engineering, as the density and, sub-
sequently, the c-concentration itself are temperature-dependent. Nevertheless, for the
calculation of reaction rates it is essential². Equation (10.1) is called a formal kinetics
equation, which means that the stoichiometric coefficients in the reaction equation
are the exponents of the c-concentration. This is the easiest approach; however, it
is not necessarily correct. In fact, for ammonia synthesis the kinetics is much more
complicated [162].
The factor k in Equation (10.1) is the reaction rate factor. A usual approach for its
temperature dependence is
k = k 0 exp (−EA /RT) , (10.3)
where EA is the activation energy of the reaction. The often cited rule of thumb that
a temperature increase of 10 K gives an increase in the reaction rate by a factor of 2–3
should not be used for calculating, but rather to illustrate how temperature-sensitive
reaction rates are.
In principle, any reaction has a reverse reaction which actually takes place to a
certain extent. In many cases, the reverse reaction can be neglected, e.g. for combus-
tion reactions. One can hardly imagine that CO2 and H2 O really react back and form a
hydrocarbon and oxygen. However, a lot of cases can also be found where the reverse
2 Note that c, n, x, and ρ all refer to moles, i.e. the units are mol/l or mol/m3 , mol, mol/mol, and
mol/m3 , respectively.
reaction is important, e.g. for all esterification reactions. At a certain stage, the con-
centrations of the participants of the reaction stay the same, as long as temperature
and pressure keep constant. From the overall view, it seems that no reactions take
place any more. In fact reaction and reverse reaction both happen, but with the same
reaction rate so that an equilibrium is formed. The most famous equilibrium reaction,
where all the aspects can be explained, is the ammonia reaction:
N2 + 3 H2 󴀕󴀬 2 NH3 .
In equilibrium, both reaction rates³ are equal, i.e.

dcNH3
= k1 cN2 c3H2 − k−1 c2NH3 = 0 . (10.4)
dτ
After rearranging Equation (10.4), an equilibrium constant K can be defined as
2
k1 cNH3
K= = . (10.5)
k−1 cN2 c3H
2
Equation (10.5) is called the law of mass action.

The condition for the reaction equilibrium can also be derived from thermody-
namics, starting from the chemical potential. A comprehensive explanation can be
found in [11]. It ends up with a slightly different expression for the equilibrium con-
stant. For the ammonia reaction, one would find
(fNH3 /fNH
0
)2
K= 3
, (10.6)
(fN2 /fN02 )(fH2 /fH02 )3
with f as the fugacity and f 0 for the fugacity at a standard state. The fugacity itself is
f i = py i φ i , (10.7)
with φ as the fugacity coefficient. The fugacity coefficients can be calculated using
the equations of state proposed in Chapter 2. Equation (10.6) can also be written in the
form
p −2 yNH3 φNH3
2 2
K=( ) , (10.8)
p0 yN2 y3H2 φN2 φ3H2
where the exponent of the pressure term is calculated from the stoichiometric coeffi-
cients (2 − 1 − 3 = −2). Discussing Equations (10.5) and (10.6), the following statements
can be given:
1. Le Chatelier’s principle:
In Equation (10.8), it can be seen from the exponent of the pressure term that
there is a tendency that the ammonia concentration increases with increasing
3 written with formal kinetics for illustration.

pressure⁴. According to the principle of Le Chatelier, the system counteracts the

effect of the pressure increase by lowering the mole number in the mixture.
2. There is a formal discrepancy between Equation (10.5) and Equation (10.8) as long
as the term with the fugacity coefficients is not negligible. In fact, the kinetic ap-
proach does not consider nonideal behavior. This means that one must be aware
that reaction kinetics calculated with the c-concentration do not reach the ther-
modynamic equilibrium defined in Equation (10.8). It would be helpful if the fu-
gacities were used as the concentration measure in reaction kinetics, but this is
not really wide-spread in chemistry.
3. Even if the fugacity term can be neglected, the equilibrium constant should be
evaluated using Equation (10.8). The determination of the coefficients from the
reaction rate will not really be accurate enough.
4. The importance of taking the vapor phase nonideality into account is illustrated in
Figure 10.1, where an attempt is made to reproduce the equilibrium conversion at
t = 450 °C with two equations of state of different quality as a function of pressure.
As can be seen, it is advantageous to use an accurate equation of state instead of
the ideal gas law. The VTPR equation of state can represent the influence of the
pressure more or less exactly, whereas the ideal gas law produces unacceptably
large deviations. It should be noted that the data is more than 80 years old; how-
ever the agreement of the two data sources and their plausibility indicate that they
can be considered to be reliable.
Analogously to Equation (10.6) the equilibrium constant for a reaction in the liquid
phase like
CH3 COOH + CH3 OH 󴀕󴀬 CH3 COOCH3 + H2 O
can be derived as [11]

(xCH3 COOCH3 γCH3 COOCH3 ) (xH2 O γH2 O )
K= . (10.9)
(xCH3 COOH γCH3 COOH ) (xCH3 OH γCH3 OH )
Again, a kinetic approach like Equation (10.5) cannot reach equilibrium. This problem
can only be overcome if the activities are used to describe the concentration, which is
not widespread. Also, heterogeneous reactions can be described with a correspond-
ing equilibrium approach. In this case, the mass transfer can also have a significant
influence on the reaction rate [8, 162].
The equilibrium constant K can be estimated from thermodynamics using the
standard Gibbs energies of formation ∆g 0f , again according to the stoichiometric coef-
4 A bit more slowly: With increasing pressure, the pressure term decreases due to the negative expo-
nent. Neglecting the pressure dependence of the φ-term, the concentration term must increase to keep
K constant. This is only possible by increasing the NH3 concentration and, correspondingly, decrease
the concentrations of the reactants N2 and H2 .
0.9
Equilibrium conversion X
0.6
Exp. data:
Larson, A.T., Dodge, B.F.,
J.Amer.Chem.Soc. 1923, 45, 2918
0.3 Gillespie, L.J., Beattie, J.A.,
Phys. Rev. 1930,36,743
Ideal gas
VTPR
0
0 200 400 600 800 1000
p/atm
Fig. 10.1: Influence of the real gas phase behavior on the equilibrium conversion
of the ammonia reaction.
ficients, e.g. for the ammonia reaction Equation (10.8)
RT ln K = −(2 ∆gNH
0
3
− 3 ∆gH
0
2
− ∆gN
0
2
), (10.10)
where the ∆g 0 values for the various components refer to the ideal gas and can be
obtained from [11]:
T T
T T cid
p
∆g0i (T, p0 = 1 atm) = ∆g0f,i + (1 − ) ∆h0f,i + ∫ cid
p dT − T ∫ dT . (10.11)
T0 T0 T
T0 T0
The effect of temperature on a chemical reaction is complex. First, the equilibrium

prefers the endothermic reaction at high temperatures and the exothermic reaction
at low temperatures. On the other hand, the reaction rates strongly increase with in-
creasing temperature, often in a way that a reaction can take place at all only at high
temperatures. To mention the ammonia reaction again, it is an exothermic reaction,
and therefore low temperatures are preferred. However, to obtain useful reaction rates
the temperature must be sufficiently high, so a compromise must be found. For the
ammonia reaction, in chemical industry a temperature of 400–500 °C is used. As the
number of moles decreases when ammonia is formed, high pressures (150–250 bar)
drive the equilibrium to the ammonia side.
A catalyst is a substance which increases the rate of a chemical reaction without

being consumed. Often, only very low amounts of a catalyst are sufficient to achieve
a significant effect. If more than one reaction is possible, a catalyst can promote the
desired one. A catalyst does not change the chemical equilibrium of a reaction; this
means that it promotes the reaction itself as well as the reverse one. For reaction ki-
netics, the catalyst concentration must often be regarded as well. The most popular
approach is Michaelis–Menten kinetics [162], which assumes that the reactant and
the catalyst form a complex that further reacts to the product and the catalyst. The
reaction rate is
creactant
r = k ccat , (10.12)
kMM + creactant
with kMM as the Michaelis–Menten constant.
Besides the equilibrium, the enthalpy of reaction is also often of interest. It is de-
fined as the heat released when a reaction takes place at constant temperature and
pressure. The equation is pretty simple:
∆h0R = ∑ hproducts − ∑ hreactants . (10.13)
The reason why this equation is so simple is that the real work has been done before
with the rigorous calculation of the enthalpy, using the enthalpy of formation as the
starting point (Chapter 2.8). Equation (10.13) automatically covers all the influences of
temperature, pressure, and real phase behavior. An often cited approach for the calcu-
lation of the temperature dependence using polynomials for the cid p and summarizing
the coefficients of the various temperature terms (e.g. [11]) is mainly an academic ap-
proach; in practice, its application is awkward. It is restricted to ideal gas applications
and, moreover, it is required that all cid
p functions are given as polynomials, which is
rarely the case and which is by far not the optimum choice for the correlation.
∆h0f and ∆g0f can be taken from data tables (e.g. [41, 42]) or be estimated using
group contribution methods [11]. However, the problem of differences between large
numbers must always be taken into account. Relatively small errors in determining
∆h0f and especially ∆g0f can lead to significant errors in estimating the enthalpy of
reaction or the equilibrium constant, respectively. Therefore, the results of estimation
methods should be handled with care. The following example gives an impression
about the sensitivity of ∆h0f .
Example
Calculate the enthalpy of reaction for the hypothetical isomerization reaction of 1,1-dichloroethane
(11DCE) and 1,2-dichloroethane (12DCE)
CHCl2 –CH3 󴀕󴀬 CH2 Cl–CH2 Cl
at t = 25 °C in the ideal gas state. The enthalpies of formation are [41]
∆h0f (11DCE) = −130120 J/mol

∆h0f (12DCE) = −126780 J/mol .
Consider that these values might have an error of ± 1 %. What is the possible range of the enthalpy of
reaction?
Solution
The expected enthalpy of reaction is
∆h R = −126780 J/mol + 130120 J/mol = 3340 J/mol .
The maximum possible enthalpy of reaction is
∆h R = −126780 ⋅ 0.99 J/mol + 130120 ⋅ 1.01 J/mol = 5909 J/mol ,
whereas the minimum one is
∆h R = −126780 ⋅ 1.01 J/mol + 130120 ⋅ 0.99 J/mol = 771 J/mol .
This means that in this case an error of ± 1 % causes a deviation of ± 77 %.
Chemical reactions are often considerably exothermic. To control the temperature,

the heat removal must be sufficient even in the worst case. Otherwise, things will
continue to spiral downward. If the heat removal cannot compensate the heat of reac-
tion, the temperature in the reactor will rise. With increasing temperature, the reaction
rates increase exponentially, causing more heat of reaction which cannot be removed
again, leading to a further temperature rise … and so on (runaway reaction). Accord-
ing to Equation (10.3), the heat generation increases exponentially with temperature,
while the heat removal due to cooling increases with temperature only linearly. In a
short time, large amounts of heat can be generated which can possibly end up in an ex-
plosion or in the destruction of the reactor. Often, degradation reactions are promoted
which spoil the product. It is essential that the occurrence of temperature peaks must
be avoided [8]. Especially in fixed bed reactors temperature peaks can occur locally
and then show the behavior described above.
For reactor design, it must be known how sensitive the reactor is when the op-
eration conditions like reactant concentration or temperature are slightly changed.
Criteria have been developed [8] which make it possible to assess whether a runaway
reaction is possible or not.
Usually, multiple reactions take place in a reactor in parallel. The design must be
performed in a way that the desired ones are supported. For this purpose, there are
some options available.
10.2 Reactors
The reactor is always the heart of a chemical plant, and the choice of the type of the
reactor has a great influence on the amount and the quality of the product. On the
10.2 Reactors | 263
other hand, the number of choices is manifold, and only the most typical ones can be
discussed here. Comprehensive discussions on reactor types can be found in [8] and
[162].
First, one can distinguish between discontinuous and continuous reactors. In the
discontinuous mode (batch), all the reactants, solvents and catalysts are fed into the
reactor, usually a vessel. The reactor is agitated to achieve that the holdup is homo-
geneous concerning temperature and composition. The composition of the holdup
changes with time. Batch reactors are flexible, they can handle different products and
the residence time can be chosen arbitrarily. Their disadvantage is that there are dead
times for filling and emptying the reactor and for the adjustment of the temperature.
Batch reactors are useful for multipurpose plants and for products with small plant
capacities. It is difficult to handle fast or strongly exothermic reactions.
Continuous reactors have constant feed and product streams. All parameters like
temperature or pressure are kept constant. The product quality can be kept constant
due to the constant conditions and the option of automation. There is a tendency for
reactor volume becoming smaller, as there are no dead times. Contrary to batch re-
actors, their flexibility is limited, and only small variations of temperature, feed flow,
and feed quality can be tolerated. In construction, care must be taken that inlet and
outlet nozzles are not too close together. Otherwise, shortcut flows will occur, and a
larger part of the reactants does not take part in the reaction.
Also, reactors can be operated in a semicontinuous mode. This can mean that in a
continuous reactor one of the reactants is fed batchwise, or that in a batch reactor one
of the products is continuously removed. For the latter case, esterification reactions
are a well-known example, e.g.
C6 H13 COOH + C10 H21 OH 󴀕󴀬 C6 H13 COOC10 H21 + H2 O .
Water is the substance with the lowest boiling point and can be easily removed from
the reactor by evaporation. The advantage is that the equilibrium is shifted to the ester
side, and 100 % conversion can be achieved. In the semicontinuous mode, strongly
exothermic reactions can be handled by adjusting the feed of one of the reactants in
a way that the heat of reaction can be removed.
Many reactor types can be characterized by three simplified ideal reactors (Fig-
ure 10.2).
Fig. 10.2: Stirred tank reactor

and tubular reactor.
– Ideally mixed batch stirred tank reactor:

The mixture of reactants is filled into the reactor and perfectly mixed during the
entire reaction time. During the reaction time no substance is removed or added.
The reactor can be heated or cooled.
– Ideally mixed continuous stirred tank reactor (CSTR):
The continuously operated stirred tank reactor is a continuously operated vessel
where the reaction takes place. There is an average residence time of the reaction
mixture, simply given by
V
τres = . (10.14)
V̇
The actual residence time can differ for the various components. In the ideal con-
tinuous stirred tank reactor, it is assumed that reactants and products are mixed
instantaneously. There is no concentration or temperature gradient. This is usu-
ally not an advantageous assumption, as reactant molecules which have not re-
acted are instantaneously transported from the inlet to the product stream. In
contrast to intuition, a real continuous stirred tank reactor is superior to the ideal
one, as the transportation of non-reacted reactants from the inlet to the outlet
of the reactor takes some time, so that all reactant molecules have the chance to
react. The CSTR is appropriate for fast reactions, where the time needed for the
reaction is considerably smaller than the average residence time.
– Plug-flow tubular reactor (PFR):
The tubular reactor is a line with a continuous flow, meaning that the mass flow
does not change over the length of the reactor. It is essential to assume plug flow
for simplification. Temperature and concentration are constant over the cross-
flow area, but vary continuously along the length coordinate due to the progress of
the reaction. The characteristics of the plug-flow tubular reactor are analogous to
the batch stirred tank reactor; in both cases, the reaction takes place in a given vol-
ume, where all substances are completely mixed, without interacting with other
volumes or, respectively, volume elements⁵. In real tubular reactors, dispersion
causes axial mixing of the substances, and heat conduction and the influence of
possible radial heating or cooling cause a temperature profile.
– Mixed forms:
The ideal continuous stirred tank reactor and the plug-flow reactor represent the
limiting cases concerning reactor behavior. In the CSTR, there is complete mixing
of reactants and products without a temperature profile. The reactant concentra-
tion is always low due to the instantaneous mixing, whereas the product concen-
tration is high. In the tubular reactor, the reactant concentration decreases from
high values at the inlet to low values at the outlet, with the products showing
the opposite profile. There are combinations of CSTR and PFR showing the inter-
5 The statement is only valid as long as the the volume does not change during the reaction.
10.2 Reactors | 265
Feed Product
Fig. 10.3: CSTR cascade.
Tubular reactor
Feed Product
Recycle
Pump Fig. 10.4: Plug-flow reactor with recycle.
mediate behavior, i.e. the CSTR cascade (Figure 10.3) and the PFR with recycle
(Figure 10.4). With an increasing number of reactor elements, the CSTR cascade
approaches the behavior of the tubular reactor, whereas the plug-flow reactor with
recycle can show the same properties as a single CSTR.
There are also reactor types available for heterogeneous reactions (Figure 10.5). To
illustrate which aspects have to be considered in reactor design, gas-liquid reactions
will be exemplarily reviewed. The other types are thoroughly discussed in [8].
For gas-liquid reactions, it is assumed that the reaction takes place in one of the
phases. Therefore, one of the components has to change the phase, and it must be
transported from the phase boundary to the bulk of the phase where it can take part in
the reaction. There is a complex dependence of the reaction rate on mass transfer, heat
Inlet
. .
. G G
L
. . Catalyst
L L
(solid)
.
Catalyst L
(solid) Heat
. . transfer
G G medium
Fixed
Bed
Suspension Bubble Fluidized Tubular
Reactor Column Bed Reactor
Fig. 10.5: Reactor types for heterogeneous reactions.

transfer, reaction kinetics, solids distribution, mixing, and gas solubility. Usually, a
complex modeling performed by a specialist takes place before the reactor is designed.
It is a key question how the energy needed for the distribution of the gas is introduced
into the reactor. Mainly, three types can be distinguished, i.e.:
– by agitating the reactor content, where the agitator disperses the gas and mixes
it with the liquid. Most effective is the hollow stirrer, where the gas is sucked in
through the hollow shaft, as a low-pressure region is formed when the agitator
is rotating. Note that the power demand for the agitator strongly depends on the
rotation speed, but even more on the diameter of the vessel:
P = Ne ⋅ n3 ⋅ d5 ⋅ ρL , (10.15)
where Ne is the so-called Newton number, characterizing the particular agitator.

– by compressing the gas. An example is the bubble column (Figure 10.5). The gas
is led into the reactor at the bottom through a nozzle holder, which disperses the
gas to small bubbles to increase the mass transfer area between gas and liquid.
Different types of circulation can be set up. Installing perforated plates, even a
cascade can be realized.
– by a liquid pumparound. The dispersion is achieved in a two-component jet. By
accelerating the liquid in the jet, again a low-pressure region is formed which
sucks in the gas.
Concerning the energy input, agitators and liquid pumparounds are more effective
than gas compression. For all options, high viscosities can cause problems.
An interesting option for performing reactions determined by the equilibrium is re-

active distillation. Its great advantage is that reaction and separation are carried out
in one piece of equipment. If the reaction is exothermic, the enthalpy of reaction can
easily be used for vapor generation. Because of the vapor-liquid equilibrium, there is
a given limit for the temperature, so the danger of undesired side reactions or even
runaway reactions is small. However, the main and outstanding advantage is that the
conversion can exceed the conversion given by the reaction equilibrium. If the boiling
points are appropriate, one of the products can be removed from the reactants due to
the distillation effect, driving the equilibrium to the product side. A well-known exam-
ple is the esterification reaction of methanol with acetic acid, giving methyl acetate.
As methyl acetate forms azeotropes with water and with methanol, the purification of
the ester produced is rather complex and requires several columns in the conventional
path [8]. With reactive distillation, only one single column is necessary (Figure 10.6).
The design of reactive distillation columns is much less certain than it is for con-
ventional distillations. There is a complex interplay between phase equilibria and
reaction kinetics, where the kinetic equations must be formulated using activities in-
stead of concentrations (Chapter 10.1). The performance of trays and packings with
respect to the chemical reaction can hardly be predicted. Laboratory tests should be
10.2 Reactors | 267
Methyl acetate
Acetic acid
Reactive part
Methanol
Fig. 10.6: Reactive distillation for the production

H2O of methyl acetate.
performed, and if there are doubts about the scale-up, experiments in larger equip-
ment might be useful.
One of the drawbacks of reactive distillation is that heterogeneous catalysts are
difficult to exchange. Figure 10.7 shows the Katapak packing (Sulzer). It is a structured
packing where the catalyst is put into small bags to remain stationary. The exchange
of the catalyst is a great effort and must completely be done manually. Homogeneous
catalysts usually cause much less trouble and can often be removed by distillation.
However, the column diameter is still determined by the residence time for the reac-
tion. The simultaneous optimization of hydrodynamics cannot take place, and from
that point of view the column has often a severe underload. Another option is the use
of external reactors (Figure 10.8). They can be used for slow reactions as well, as the
residence time in the reactive zone can be adjusted by conventional reactor design.
The scale-up is easy, as there is no coupling between reaction and hydrodynamics.
The catalyst exchange is easier, as well as the design. An extensive piloting is usually
not necessary.
For standard reactors, process simulators offer some options for the representa-
tion.
– The simplest and most widely used one is the stoichiometric reactor. As input, the
stoichiometry of the various reactions and the conversion referring to one of the
reactants are necessary. There is the option that the conversion factors refer to the
amount at the reactor inlet or that the reactions take place in a sequence, where
the conversion refers to the amount at the beginning of this reaction.
– The yield reactor is an even more simple one. The outlet concentration can be
specified, the mass flow is kept constant. The disadvantage is that alchemy can
be simulated with this option, it is no problem to specify that water is turned into
gold. It is often used when the reactions taking place are more or less unknown,
e.g. in fermenters in biotechnology.
Fig. 10.7: Sulzer Katapak.

Packing
Vapor
Collector
Liquid
External
reactor Distributor
Packing
Liquid
Fig. 10.8: Reactive distillation
with an external reactor.
– There are several options to define an equilibrium reactor. In the first option, the
reactions taking place must be specified. Either the equilibrium constants can be
given, or the simulator calculates the equilibrium using the Gibbs energies (Equa-
tion (10.11)). Multiple phases can be considered as well.
– Alternatively, only the possible reaction products can be listed without defining
the reactions theirselves. The program can then find the corresponding minimum
of the Gibbs energy.
– Tubular reactors (as plug flow reactors), continuous and batch stirred tank reac-
tors can be specified using reaction kinetics.
11 Mechanical strength and material choice
Even a process engineer should know that the crack in the sausage on the grill is in longitudinal
direction.
(Olaf Stegmann)
The statement refers to a discussion where it was assumed that the worst case for a
pipe was a sudden break through the cross-flow section. In fact, this is very unlikely
to happen, as Figure 11.1 indicates. When the pipe is pressurized, the equilibrium of
force in longitudinal direction is¹
πD2
σa πDs = p , (11.1)
4
giving
pD
σa = , (11.2)
4s
where σa is the mechanical tension in the axial or longitudinal direction. In circum-
ferential direction, the equilibrium of forces yields (Figure 11.1)
σt ⋅ 2sL = pDL , (11.3)
giving
pD
σt = , (11.4)
2s
with σt as the mechanical tension in the tangential or circumferential direction. Thus,
the tension in the circumferential direction is twice as large as in the longitudinal
direction, and as long as there is no predetermined breaking point, the pipe would
burst in the circumferential direction at overpressure, causing a longitudinal crack
(Figure 14.11). If it bursts at all. There is a guideline statement called “leak before
burst”. The established pressure vessel standards require designs which favor “leak
before burst”, allowing the fluid to escape and reducing the pressure before the dam-
age is growing so large that a complete fracture takes place. Usually, the weakest parts
against pressure load are the gaskets, and one can easily imagine that these will show
leakage first.
Equation (11.4) is the so-called boiler formula. It is the foundation of many formu-
las for mechanical stability calculations for vessels, where the elastic limit is inserted
for the tension. In the particular technical guidelines, Equation (11.4) is supplemented
by terms describing manufacturing uncertainties, corrosion allowances, safety factors
and the influence of welding seams. Also, different ratios between outer and inner di-
ameter are regarded [149].
1 The wall thickness s is assumed to be small compared to the diameter.

270 | 11 Mechanical strength and material choice
axial tangential
s
σa σt
D p p
σa σt
Fig. 11.1: Illustration of the vessel formula.
Before the mechanical stability can be considered, for each piece of equipment and the
adjacent piping the design temperatures and design pressures² have to be assigned.
These are the conditions where a safe operation is definitely possible without restric-
tions, even if they are both reached at the same time. In fact, as standard wall thick-
nesses are used in the manufacturing process, the equipment will be able to cope with
even higher pressures, since not the required standard but the next larger standard
wall thickness will be taken. It has to be proved that the equipment can withstand
the design pressure, usually by pressurizing with liquid water. The test can in most
cases not realize both design pressure and design temperature at the same time; there-
fore, a conservative temperature correction is performed, giving a test pressure higher
than the design pressure at the lower temperature. Also, for high pieces of equipment
(e.g. columns) this kind of test systematically yields higher pressures than requested,
as the whole apparatus is pressurized to the test pressure, but the lower part is ad-
ditionally exposed to the hydrostatic pressure. In operation, pressure vessels are in
most cases protected by pressure relief devices such as safety valves or rupture discs
(Chapter 14.2), which actuate at the design pressure. Again, the definition of design
pressure turns out to be soft, as a safety valve starts to open at the design pressure
but fully opens at a pressure which is 10 % above so that again the design pressure is
systematically exceeded.
In the design basis, a rule must be set up how the values for design temperature
and design pressure can be determined. From process simulation, the normal operat-
ing conditions are known. The design conditions are related to these values; often by a
factor (1.1–1.25) for the design pressure, and by an offset for the design temperature. In
some cases, the operation conditions are not decisive. For example, the design temper-
ature of heat exchangers must refer to the highest possible temperature of the hot side.
In case the apparatus is cleaned by steamout³, it can happen that the corresponding
steam determines the design conditions.
2 Design pressures refer to overpressures, and the unit is therefore “MPag” or “barg”, where the “g”
stands for “gauge”, i.e. overpressure. In contrast, if absolute pressures are meant, the letter “a” for
“absolute” can indicate that absolute pressures are meant, e.g. MPaa or bara.
3 See Glossary.
Fig. 11.2: Ductile fracture and brittle fracture.

© BradleyGrillo/Wikimedia Commons/CC BY-SA 3.0.
© Sigmund/Wikimedia Commons/CC BY-SA 3.0.
https://creativecommons.org/licenses/by-sa/3.0/deed.de
Process equipment can be exposed to low temperatures. The materials of construction

can undergo a transition from ductile to brittle behavior at low temperatures, which
increases the risk of brittle fracture [165, 166], one of the most critical damages. Mate-
rial failure should be of the ductile type, meaning that plastic deformation takes place
before complete destruction. This gives at least some time to react. In brittle fracture,
the material failure occurs suddenly. Figure 11.2 shows the difference between brittle
fracture and ductile fracture. It can easily be seen that for the brittle fracture no defor-
mation takes place. Therefore, the specification of the equipment should include an
indication of the minimum design metal temperature (MDMT).
The case which is most often relevant for the determination of the MDMT is the
so-called auto-refrigeration case. If a low-boiling substance is stored in a vessel under
pressure in the liquid state, it will evaporate in case of pressure relief. The tempera-
ture of the liquid will follow the boiling point curve. For example, a vessel containing
liquid propylene at p = 10 bar will cool down to t = −48 °C, the boiling temperature
of propylene at p = 1 bar, if a complete pressure relief down to ambient pressure takes
place. If there is a mixture in the vessel, the boiling temperature varies according to the
vapor-liquid equilibrium, possibly ending up at the boiling temperature of the high-
est boiling component or azeotrope. However, it usually cannot be guaranteed that
at pressure relief only the component with the highest boiling point is in the vessel.
Therefore, in most cases the lowest boiling point is relevant for the design. Certainly,
one should not relate the MDMT with the design pressure of the vessel. Especially
272 | 11 Mechanical strength and material choice
for the case of pressure relief described above, it becomes clear that the pressure to
which the vessel is exposed strongly decreases with the temperature. Therefore, the
minimum allowable temperature (MAT) should be defined as a function of pressure.
Construction engineers should at least be provided with single temperature-pressure
pairs for the various cases to avoid overdesign.
It is often underestimated that an overpressure from outside can be critical for the
mechanical stability of equipment or piping. This occurs if the equipment is evacuated
according to the process conditions or if the equipment is emptied with a vacuum
pump. If such conditions occur, it should be indicated in the specification, e.g. by an
additional design pressure pDes = −1 barg. Things become more complicated if the
pressure exposure from outside is higher than 1 bara. In double pipes, steam on the
annular side might have a considerably higher pressure than the product in the inner
tube. This should be indicated in some way on the datasheet, e.g. by specifying the
overpressure from outside with a negative sign as in the vacuum case. An indication
like pDes = −16 barg will hopefully attract the attention of the vendor or cause at least
a further inquiry so that a damage as shown in Figure 11.3 can be prevented.
Fig. 11.3: Damage of the inner tube of a double pipe due to wrong
specification.
Mechanical stability is highly related to the use of appropriate materials. Also, the
chemical stability must be guaranteed. To withstand aggressive chemicals, a number
of special materials are available. Often, the quality of the surface is also an issue, e.g.
in bio-processes, where microorganisms tend to stick to rough surfaces.
In chemical plants, the materials used can be divided into metals (steel, alu-
minium, nickel, titanium), nonmetallic materials (ceramics, graphite, glass), and
polymers. The choice of the materials should be performed by a specialist. Critical
issues for the choice of the material are:
– strong organic or inorganic acids;
– sour gases, especially bromine and chlorine;
– fluoride, chloride and bromide ions;
– caustics, especially at high temperatures;
– hydrogen, which can diffuse through many materials and is explosive over a wide
concentration range.
Corrosion data can often be found in the literature. If sufficient information is not
available, a material test can be helpful where a piece of metal is exposed to the
medium for a certain time. This procedure is easy but time-consuming. The terms
for the identification of certain steels is a bit confusing, as different guidelines use
completely different systems. For example, the well-known V2A steel is called 1.4301
according to the material number (DIN EN 10088-1), X5CrNi18-10 according to the
chemical composition and 304 according to the AISI standard. A comprehensive
compilation of all the issues concerning the material choice can be found in [167].
12 Piping and measurement
A tube has an excavation which is as long as the tube itself.
(Reportedly from a handbook of construction)
The transport of gaseous and liquid substances between the different pieces of equip-
ment is achieved in pipes, which form a major part of a chemical plant. The piping
engineers are the ones with the most interaction with other activities; for instance,
a detailed piping cannot be performed unless the nozzles of the vessels have been
put in place. A stepwise working procedure with continually increasing generation of
information is typical for piping. While piping activities are more or less concentrated
in the detail engineering, the basis is in fact provided by process engineering, as the
piping dimensions are mainly determined by the process. This chapter focuses on the
process aspects of piping and other items like valves and measurement devices which
are related. The important terms concerning the planning of the piping are explained.
12.1 Pressure drop calculation
12.1.1 Single-phase flow-through pipes
Pressure drop calculations in pipes are standard tasks in process engineering. The
following chapter explains the common equations which are widely used in process
engineering. It should be mentioned that all equations refer to Newtonian fluids. For
non-Newtonian flow, where the viscosity depends on the shear forces, other proce-
dures are necessary [168]. With the tube length L and the friction number λ, the rela-
tionship between pressure drop and velocity is given by Equation (12.1):
ρw2 L
∆p = λ , (12.1)
2 dh
where dh is the hydraulic diameter. For different geometries, it is defined by
A
dh = 4 , (12.2)
U
where A is the cross-flow area and U the circumference. For circular tubes, Equa-
tion (12.2) yields of course the inner tube diameter dh = d. For circular rings, Equa-
tion (12.2) gives
dh = da − di , (12.3)
with da as the outer and di as the inner diameter. For a rectangular cross-section with
two different side lengths s and w, the hydraulic diameter is
2 sw
dh = . (12.4)
s+w
276 | 12 Piping and measurement
Fig. 12.1: Moody diagram.

© S. Beck and R. Collins, University of Sheffield/Wikimedia Commons/CC BY-SA-3.0.
https://creativecommons.org/licenses/by-sa/3.0/deed.en.
Using the Reynolds number

wdh ρ
Re = , (12.5)
η
the friction factor λ as a function of the Reynolds number can be determined with the
well-known Moody diagram (Figure 12.1). For laminar flow (Re < 2320), there is a strict
theoretical relationship independent of the surface roughness of the pipe according
to the Hagen–Poiseuille law [149],
64
λ= φ. (12.6)
Re
The factor φ is 1 for circular tubes. For noncircular cross-sections, the factor can be
determined using Tables 12.1 and 12.2 [149].
For the turbulent flow, the roughness of the inner pipe surface plays a major role
and is represented by the roughness k, which can be interpreted as the size of sand
grains on the surface. Guide values are given in [149]. For the turbulent flow in a
smooth pipe (Re ≥ 2320, k Re < 65 dh ), the friction factor can be determined by the
Table 12.1: Circular ring.
da /di 1 5 10 20 50 100
φ 1.5 1.45 1.4 1.35 1.28 1.25
Table 12.2: Rectangular cross-section.
s/w 0 0.1 0.3 0.5 0.8 1.0

φ 1.5 1.34 1.1 0.97 0.9 0.88
formula of Prandtl and v. Kármán [149]:

−2
Re√λ
λ = [2 lg ] , (12.7)
2.51
which has a theoretical background [169] and is the only one valid in the whole range.
Its disadvantage is that it must be solved iteratively. There are two popular approxi-
mations, the Blasius equation
0.3164
λ= , (12.8)
Re0.25
valid for 2320 < Re < 100000, and the formula
0.309
λ= , (12.9)
[lg (Re/7)]2
which can be applied in the whole range [149]. Figure 12.2 shows that it is really worth
to take care of the application ranges.
For rough surfaces, the formulas of Colebrook [149]:
2.51 0.27 −2
λ = [−2 lg ( + )] (12.10)
Re√λ dh /k
for Re ≥ 2320, 65 dh < k Re < 1300 dh ; and Nikuradse [149]:
λ = [2 lg (3.71 dh /k)]−2 (12.11)
for Re ≥ 2320, k Re > 1300 dh can be used.
0.030
0.025
0.020
λ 0.015
0.010
0.005
0
104 105 106 107 108 Fig. 12.2: Comparison between Equations
Re (12.8) (· · ·), (12.7) (—), and (12.9) (- - -).
During the process engineering of a whole plant, the diameter of a large number of
lines has to be sized, which requires an excellent documentation and workflow. It is
a good approach to compute the friction factor Equations (12.6)–(12.11) in an EXCEL
file and make a list of the lines to be sized. A quantity which can easily be interpreted
is the pressure drop per 100 m tube length according to Equation (12.1). Varying the
tube diameter, the calculated pressure drops can be compared with the correspond-
ing values which are usually defined in the guidelines of the particular companies.
Also, it can be checked whether the velocities in the pipe are reasonable, e.g. 1–2 m/s
for liquids and 10–20 m/s for gases. Figure 12.3 gives a rough orientation. All cases
regarded have in common that for larger diameters higher velocities can be allowed,
as in this case the friction with the tube wall plays only a minor role. It should also be
mentioned that diameters below 2′′ are rarely applied due to static reasons.
20
Steam
18 Nitrogen
(170° C,
(20° C,
6 bar)
Recommended velocity / m/s
16 20 bar)
Ethylene (50° C, 5 bar)
14
12
10
8
6
4
2 Water (40° C, 6 bar)
0
0 2 4 6 8 10 12 14 16 18 20
Pipe diameter/inch
Fig. 12.3: Recommended velocities in a pipe.
It is worth thinking over the meaning of this procedure. It should be clear that the
guideline values for the pressure drop are recommended values. They are compro-
mises between the additional investment costs for pipes with larger diameters and
higher operation costs due to larger pressure drops for pipes with smaller diameters.
Anyway, the tube does not fail if the recommended values are exceeded. It makes
sense to accept them if no other requirements for the pipe exist, e.g. special pres-
sure drop constraints if the compressor or the pump is limited or if the guidelines
for safety valves must be obeyed (Chapter 14.2). They can be completely ignored if
the line leads to a valve where the pressure is significantly lowered anyway. There
is often an overreaction when the recommended value is slightly exceeded. When the
next diameter is chosen, it often happens that the pressure drop goes down to very low
values, indicating that the tube is by far overdesigned. This becomes clear when Equa-
tion (12.1) is written with the mass flow and, for simplicity, the Blasius Equation (12.8):
ṁ 2
ρ ( ρA ) L
−0.25
∆p = 0.3164 Re . (12.12)
2 d
With A = πd2 /4 one gets
wdρ 4ṁ 4ṁ −1
Re = = = d (12.13)
η πdη πη
and, after summarizing constant terms,

L
∆p = C1 d0.25 C2 d−4 = Cd−4.75 . (12.14)
d
This illustrates the dramatic dependence of the pressure drop on the tube diameter,
as well as the following example.
Example
A cooling water stream (ṁ = 50000 kg/h, t = 28 °C, p = 6 bar) is pumped to a consumer unit. The
recommended maximum pressure drop is ∆p = 0.2 bar per 100 m. Calculate the appropriate tube di-
ameter. The tube shall be hydraulically smooth.
Solution
First, the corresponding physical property data are ρ = 996.46 kg/m3 and η = 0.8324 mPa s [29].
The first approach will be a 4′′ tube, giving d ≈ 101.6 mm (1′′ = 25.4 mm). Then, the Reynolds
number is (Equation (12.13))
4ṁ −1 4 ⋅ 50000 kg/h

Re = d = = 209099 . (12.15)
πη π ⋅ 0.8324 mPa s ⋅ 101.6 mm
According to Prandtl/v. Karman (Equation (12.7)), the friction factor is evaluated iteratively to be λ =
0.0155. Then, using the velocity
ṁ 50000 kg/h
w= = = 1.72 m/s , (12.16)
ρA 996.46 kg/m3 ⋅ π (101.6 mm)2
4
the pressure drop per 100 m is equal to (Equation (12.1))
996.46 kg/m3 (1.72 m/s)2 100 m

∆p100 m = 0.0155 = 0.225 bar . (12.17)
2 101.6 mm
This is slightly larger than the recommended 0.2 bar/100 m, at a reasonable velocity. Increasing the
tube diameter to 6′′, which is the next standard nominal diameter, the resulting pressure drop per
100 m becomes ∆p = 0.03 bar, which is by far lower. Probably, it is the more reasonable decision to
stay at d = 4′′.
12.1.2 Pressure drops in special piping elements
For special piping elements, the pressure drop is usually calculated via
ρw2
∆p = ζ . (12.18)
2
A number of ζ -values according to [149] is listed in Appendix B.
12.1.3 Pressure drop calculation for compressible fluids
For gas flows, the procedure described in Chapter 12.1.1 is only valid if the flow can
be regarded as incompressible. This can be checked with the Mach number (Ma), the
ratio between the actual velocity and the speed of sound. The velocity should stay
below 30 % of the speed of sound:
Ma = w/w∗ < 0.3 . (12.19)
The speed of sound can be expressed as [11]

∂p
(w∗ )2 = −v2 ( ) . (12.20)
∂v s
With a pressure-explicit equation of state, it can be determined via [11]
T ∂p 2 ∂p
(w∗ )2 = v2 [ ( ) −( ) ] (12.21)
c v ∂T v ∂v T
with
v
∂2 p
cv = cid
v +T∫( ) dv . (12.22)
∂T 2 v
∞
For ideal gases, Equation (12.20) gives
(w∗ )2 = κRT (12.23)
p /c v , where c p and c v are assumed to be constant and not a function of

with κ = cid id id id
temperature. The speed of sound is a limiting velocity for a gas flow in a pipe. Except
the case that the pipe has the shape of a Laval nozzle with a minimum in the cross-
flow area (Figure 12.4), thermodynamics can ensure that the speed of sound is not
exceeded [153].
Subsonic Supersonic
Throat
high pressure low pressure
Convergent Divergent
section section Fig. 12.4: Shape of a Laval nozzle.
Due to the relatively high velocities for compressible flows, large pressure drops occur,
often changing the state variables of the stream significantly. Due to the pressure loss,
the density is reduced. This means that at constant mass flow the volume flow and
therefore the velocity increases, which in turn produces further increased pressure
drops. The density further decreases, and finally the pressure drop is by far increased
in a vicious circle. This behavior can be summarized in the following striking formula:
“Pressure drop causes pressure drop”
This is an important issue, especially for the design of the outlet lines of pressure relief
devices (Chapter 14.2).
For compressible flow, the pressure drop calculation is demonstrated in the fol-
lowing example. Because of the changes of the fluid state variables, it is useful to
divide the tube into small increments and calculate them sequentially one by one,
where the state is updated at the inlet of each segment. The procedure can be per-
formed in an EXCEL file or with the help of a process simulator, which usually offer
the incremental calculation as an option.
It is important to know that a conventional pressure drop calculation with a given
mass flow usually yields to a solution which is not realistic. At the end of the pipe, an
incompressible fluid must end up with the given outlet pressure. If the pressure drop
is too large, the conclusion is that the mass flow cannot be realized; due to choking, it
will be less. If the pressure drop is too low, the fluid will expand in a way that the inlet
pressure is lowered to an extent where the outlet pressure is met. The same holds in
principle for a compressible flow; however, the mentioned expansion is coupled with
a significant change of the state. Furthermore, another restriction is that the speed
of sound cannot be exceeded in a pipe. If it is reached upstream the pipe outlet, it is
clear that the mass flow assumed is too high and choking takes place. At most, speed
of sound can be reached directly at the pipe outlet. In this case, the outlet pressure
will not be met; instead the fluid will expand directly after it has left the pipe. The
following example should illustrate this. It might look a bit exotic, however, cases like
this occur in outlet lines of rupture discs (Chapter 14.2).
Example
A nitrogen flow (ṁ = 30000 kg/h, p = 100 bar, t = 20 °C) enters a line (L = 50 m) which ends at a
header with p = 1 bar. Choose an appropriate diameter for the line. The tube is hydraulically smooth.
Solution
Various diameters are tested with a tube increment length of ∆L = 0.1 m. The results are illustrated in
Figures 12.5 and 12.6.
120
8"
100
2"
p/bar 80
60
1"
40
20
0
0 10 20 30 40 50
Fig. 12.5: Pressure courses without expansion
L/m at the inlet.
120
100
80
p/bar
60
1"
40
2"
20
8"
0
0 10 20 30 40 50
Fig. 12.6: Pressure courses with expansion
L/m at the inlet.
– The 1′′ line is too narrow for the given mass flow. At the inlet, the Mach number is already 0.38.
Due to the “pressure drop causes pressure drop” effect, the fluid continuously expands, and the
velocity rises increasingly. After almost 14 m, speed of sound is reached and choking takes place
(Figure 12.5). The given mass flow cannot pass the pipe as assumed.
– For a 2′′ line, things are more difficult. Starting with p = 100 bar at the inlet, the pressure drop
of the line is ∆p = 6 bar, giving p = 94 bar at the outlet, and the Mach number increases very
smoothly from Ma = 0.09 at the inlet to Ma = 0.1 at the outlet (Figure 12.5). The conclusion is
that the pipe diameter is sufficient; however, one should imagine what really happens in the pipe;
especially, the design pressure of the pipe (Chapter 11) might be interesting. For this purpose, it
is assumed that an adiabatic expansion happens at the inlet, where the velocity change is not
neglected in the First Law:
1 1
h1 + w12 = h2 + w22 . (12.24)
2 2
In an iterative procedure, it is found out that an expansion to p = 33.7 bar at the inlet gives a
possible result. At the outlet of the pipe, speed of sound is reached, while the pressure only
drops to 9.1 bar (Figure 12.6). The fluid will rapidly expand after leaving the pipe.
– While the 4′′ and the 6′′ line show a similar behavior, the 8′′ line works in a different way. Again,
the inlet pressure of p = 100 bar will yield a pressure drop which is by far too low (∆p = 7 mbar,
Figure 12.5), and expansion will take place. Iteratively, an expansion to p = 1.49 bar at the inlet
can be determined. At the pipe outlet, p = 1 bar is reached, while the Mach number Ma = 0.6
indicates that the velocity is below speed of sound (Figure 12.6).
12.1.4 Two-phase pressure drop
The evaluation of the pressure drop of a one-phase flow is a quite exact one with a
well-defined theory behind it. Things become much more complicated when a second
phase comes into play. Especially vapor-liquid flows have a great technical impor-
tance. The pressure drop of a two-phase flow is characterized by the friction between
the phases, which is hardly predictable. This friction causes the pressure drop to be
higher than expected; it is usually underestimated, as even the apparently most con-
servative assumption of a pure vapor flow is not on the safe side, as Figure 12.7 shows.
The pressure drop of water in the two-phase region at p = 1.1 bar is considered for
various vapor fractions. It can be seen that the obvious approach of averaging the
pressure drops of the vapor and the liquid flow systematically underpredicts the two-
phase pressure drop. For high vapor fractions, the two-phase pressure drop exhibits
a well-defined maximum; even the assumption of a pure vapor flow as mentioned
above yields lower pressure drops. As explained below, the horizontal and the verti-
cal upward and downward flows have to be distinguished for the calculation of the
two-phase pressure drop.
The best calculations for the two-phase pressure drop are probably the ones used
in the commercial heat transfer programs, as they are decisive for the design of ther-
mosiphon reboilers. However, to the knowledge of the author, they have not been pub-
lished. The most popular published correlations are the ones of Lockhart-Martinelli
[170], Friedel [171–173], and Beggs-Brill [174]. In the following, the Friedel method is
exemplarily explained, which is considered to be the most reliable one because of its
large database. However, errors up to 50 % might still occur. The method determines
the two-phase factor R2Ph , which represents the ratio of the pressure drops of the two-
25
100% Vapor flow
Averaging between vap. and liq. flow
Pressure drop per 100 m/mbar
20 Upward flow acc. to Friedel

Downward flow acc. to Friedel
15
10
0
0 0.2 0.4 0.6 0.8 1
Vapor fraction
Fig. 12.7: Typical curvature of the two-phase pressure drop with respect to the vapor fraction.
Calculated with the Friedel equation.
phase flow and of a one-phase liquid flow with the same mass flow:
∆p2Ph = R2Ph ∆pL . (12.25)
∆pL is the pressure drop according to Equation (12.1), where the total mass flow of the
two-phase stream is replaced by a fully liquid stream:
̇
(m/A) 2 L
∆pL = λL . (12.26)
2 ρL dh
With the Reynolds numbers for both phases j = L, G

̇ h
md
Rej = (12.27)
A ηj
and the vapor mass fraction

mG
x= (12.28)
mG + mL
an auxiliary quantity A∗ can be calculated as
ρL λG
A∗ = (1 − x)2 + x2 . (12.29)
ρG λL
It must be emphasized that in fact for ṁ the total mass flow has to be used. In con-
trast to the single phase flow, there is no discontinuity for λ j at the transition laminar/
turbulent but a continuous transition region.
For a circular cross-flow area, the friction factor is
λ j = 64/Rej for Rej ≤ 1055 (12.30)
and
−2
Rej
λ j = [0.86859 ln ] for Rej > 1055 . (12.31)
1.964 ln Rej − 3.8215
For geometries differing from the circular cross-flow area the following changes have
to be considered:
– For a rectangular cross-flow area, the hydraulic diameter (Equation (12.2)) must
be used:
ṁ
Rej,rectangle = Ψ dh /η j , (12.32)
A
with
2 11 s s
Ψ= + (2 − ) , (12.33)
3 24 w w
where s is the length of the shorter and w the length of the larger side. The other
steps are analogous to the circular cross-flow area.
– For circular ring cross-flow areas there are the relationships
λ j = 64/Rej for Rej ≤ 1055 (12.34)
and
−2
λ j = [2 lg (Rej √ λ j ) − E] for Rej > 1055 . (12.35)
Equation (12.35) must be solved iteratively. E can be determined according to the

following table.
di /da 0 0.05 0.3 0.8 1.0

E 0.8 0.932 0.961 0.968 0.97
Between the values for di /da , linear interpolation can take place. For the evalua-
tion of the Reynolds number Re, again the hydraulic diameter (Chapter 12.1) must
be used.
With the Froude number

ṁ 2
FrL = 2
(12.36)
A2 gd h ρL
and the Weber number

ṁ 2 dh
WeL = (12.37)
A2 σρL
R2Ph can be determined to be
– for the horizontal and the vertical upward flow:
R2Ph = A∗ + 3.43 x0.685 (1 − x)0.24 (ρL /ρG )0.8 (ηG /ηL )0.22
(12.38)
(1 − ηG /ηL )0.89 Fr−0.047
L We−0.0334
L ,
– and for the vertical downward flow:
R2Ph = A∗ + 38.5 x0.76 (1 − x)0.314 (ρL /ρG )0.86 (ηG /ηL )0.73
(12.39)
(1 − ηG /ηL )6.84 Fr−0.0001
L We−0.087
L .
The Friedel equations are valid for the whole vapor fraction range 0 < x < 1. As limit-
ing cases, one obtains the corresponding equations for the single-phase equations for
vapor and liquid, except the transition region from laminar to turbulent flow. In case
the influence of the roughness of the tube wall is not negligible (k ReG < 65 dh ), the
result should be compared with the one for the pure vapor flow. One should always
be aware that due to the relatively high pressure changes the vapor fraction along
the tube might vary significantly. In these cases, it makes sense to divide the tube into
small increments and evaluate the pressure drop increment by increment, as it is often
necessary for compressible flow as well (Chapter 12.1.3). As for the compressible flow,
the statement “Pressure drop causes pressure drop” holds. Current process simulators
usually offer an opportunity to specify such a calculation.
For the flow through piping elements, only the pipe elbow is sufficiently dis-
cussed:
For the 90°-elbow, Muschelknautz [175] specifies the following procedure:
2.2
B =1+ (12.40)
λ dL (2 + r/d)
ρL
R2Ph = 1 + ( − 1) [B x (1 − x) + x2 ] (12.41)
ρG
with r as the elbow radius. Accordingly, one obtains for the pressure drop
̇
L (m/A)2
∆p2Ph = λ R2Ph . (12.42)
d 2 ρL
Equations (12.40) und (12.41) are only valid for the 90°-elbow.
For other piping elements, the only remaining option is to define an equivalent
tube length according to
dh
L=ζ (12.43)
λ
and calculate the pressure drop along this artificially defined pipe. λ is then calculated
using Equations (12.6), (12.7), (12.10), or (12.11), depending on the conditions.
Besides the pressure drop, the flow pattern of a vapor-liquid flow is important. They
are shown in Figure 12.8, taken from [176]. For vertical upward flow, the patterns are
– Bubble flow (A):
A large quantity of bubbles is present which are almost homogeneously mixed
with the liquid. The liquid phase is still wetting the whole tube wall.
– Slug flow (B):
Very large bubbles are formed which have a length that is by far larger than the
diameter of the tube. When they end, they are followed by liquid flow with low
vapor content. At a tube bend or a transition piece, this liquid will hit the tube
wall and cause mechanical damage with time. Slug flow should be avoided.
– Chaotic flow (C):
Large and small bubbles are randomly distributed.
– Wispy annular flow (D):
The liquid is predominantly distributed around the tube wall. Vapor and swarms
of droplets are in the tube core.
– Annular flow (E):
The liquid is almost entirely distributed around the tube wall, only few droplets
are suspended in the vapor flow in the tube core.
104 101 Flow direction

Annular flow (e) Bubble flow (a)
100 a
103
b
FD
TD 10–1 c
KD[–]
102 d
Slug flow (d) e
Wavy flow (c) 10–2
101 KD
10–3
Stratified flow (b)

100 10–4
10–3 10–2 10–1 100 101 102 103 104
X[–]
105 A B C D E
Flow direction
104
Annular flow (E) Annular flow
103 wispy (D)
ig,0[kg/m s2]
102
Chaotic Bubble flow (A)
101
flow (C)
Slug flow (B)
100
10–1 4
100 101 102 103 10 105 106
2
il,0[kg/m s ]
Fig. 12.8: Flow patterns of vapor-liquid two-phase flow for horizontal (upper pictures) and vertical
upward flow (lower pictures) [176].
For horizontal flow, the patterns are (Figure 12.8)

– Bubble flow (a):
The vapor phase forms small bubbles. Due to the influence of gravity, they are
distributed in the liquid in the upper part of the tube.
– Stratified flow (b):
The vapor phase is in the upper part, the liquid phase is in the lower part of the
tube. There are no waves at the phase boundary.
– Wavy flow (c):
Similar to stratified flow (b), but with waves at the phase boundary.
– Slug flow (d):

Similar to wavy flow, but the waves can occupy the whole cross-section. There is
an increased occurrence of bubbles in the liquid phase and droplets in the vapor
phase. Again, there is the danger of mechanical damage for the tube.
– Annular flow (e):
The tube wall is fully wetted, but the liquid ring formed in the cross-section is
asymmetric, with more liquid at the bottom than at the top of the tube. The vapor
phase is in the center of the tube, with many liquid droplets.
In the diagrams in Figure 12.8 the coordinates of the X- and Y-axis are defined as fol-
lows.
For the vertical flow:
ṁ 2 (1 − x)2
i l,0 = (12.44)
A2 ρL
ṁ x2
2
i g,0 = 2 , (12.45)
A ρV
and for the horizontal flow:

0.5
( ∆p
∆x )L
X= 0.5
, (12.46)
( ∆p
∆x )V
where the ∆p values refer to the situation where, respectively, the vapor and the liquid
phase would occur alone and occupy the whole cross-section area. ṁ is the total mass
flow. For the horizontal flow, the boundary between bubble and slug flow refers to the
ordinate
( ∆p
∆x )L
0.5
TD = [ ] . (12.47)
(ρL − ρV ) g
The coordinate for the boundary between annular and wavy flow is
̇
mx
FD = 0.5
. (12.48)
A ((ρL − ρV )ρV dg)
The coordinate for the boundary between annular and wavy flow is
ṁ 3 x2 (1 − x)
KD = . (12.49)
A3 (ρL − ρV )ρV gηL
For the choice of the diameter, one should take care that periodical shocks due to slug
flow should be avoided. It can be avoided by choosing lower pipe diameters, however,
this option is often limited and the diagrams are not excessively accurate. Further-
more, small diameters give large pressure drops, which is often the limitation. Thus,
the usual strategy is to avoid two-phase flow as far as possible, e.g. by placing the
expansion valves directly upstream the vessel or, if possible, at a low point.
12.2 Pipe specification | 289
12.2 Pipe specification
Besides the inner diameter of a pipe, which is decisive for the pressure drop calcula-
tion, there are of course a large number of other items to be specified for a pipe. To
keep the overview, so-called pipe classes are defined, which differ from company to
company. In these pipe classes, a large part of the information about a pipe can be
predefined, such as:
– Pressure rating:
The design conditions of a pipe are normally predefined by the adjacent pieces
of equipment. In the pipe class denomination there is usually a code which en-
sures sufficient design conditions. Also, the influence of the temperature on the
mechanical stability is considered and defined.
– Fluid code:
A certain abbreviated code gives qualitative information about the fluid going
to the pipe, both concerning the substances involved and the design conditions.
There are sophisticated pipe class systems where the fluid code carries the com-
plete information about the pipe.
– Piping material:
The piping material is indicated in the piping denomination code, either explicitly
or via the fluid code mentioned above.
– Gasket type:
The gasket type can be indicated in the fluid code. Also, the necessary pipe con-
nections (flange, welded connection) can be defined.
– Insulation:
There are a number of insulation types for a pipe which have to be distinguished:
– None:
Insulation is not needed for streams where no exorbitant temperatures or
other dangers occur. An example is cooling water.
– Heat insulation:
An insulation must be defined (material, thickness) which avoids heat losses
in the pipe. The insulation material must be thermally stable and effective at
the required temperature.
– Cold insulation:
An insulation must be defined (material, thickness) which can keep the
stream in the pipe at the cold temperature required. The insulation material
must be effective at the required temperature. Often the ingression of air
humidity into the insulation material must be avoided.
– Personnel protection insulation:
An insulation is provided which is thermally not effective, but prevents that
members of the staff touch the pipe. This could cause injuries if it is hotter
than approx. 60–70 °C. Most companies have guidelines where personnel pro-
tection insulation is required.
– Electrical tracing:
Electrical tracing is required if there is a danger that the fluid is no more
pumpable at cold environmental temperatures, e.g. water at temperatures
below 0 °C. Electrical tracing is relatively expensive.
– Jacket tracing:
A double pipe is manufactured where a heating agent (steam, hot water) is
used to keep the temperature in the inner part of the pipe.
Pipes are assembled at environmental conditions. During operation, they will often be
exposed to elevated temperatures, and their lengths will increase due to thermal ex-
pansion. If these expansions are not compensated for, large mechanical tensions will
occur which can possibly cause damage to the gaskets, the pipe fittings, and the pipe
itself. Small changes in length can be compensated by the elasticity of the material,
for major expansions special compensating elements are necessary.
For pipes operating at high pressure, bend elements are used. The most popular
one is the U-bend (Figure 12.9), which can compensate the tensions by deformation.
In application, one should not forget the high-point vent or the low-point drain to
avoid accumulation of gases or, respectively, liquids. Another option is the bellow ex-
pansion joint (Figure 12.10). Between two flanges, a bellow pipe can equalize the pipe
expansion. A guide tube inside can prevent the bellow from being polluted, however,
in this case only axial expansions can be compensated.
Fig. 12.9: Expansion loop.

12.3 Valves | 291
ΔX
Δl
Δ0
Axial Angular Lateral Fig. 12.10: Bellow expansion joint [177].
movement movement movement © Hydrocarbon Processing.
12.3 Valves
Valves are used in piping systems to control flowrates, pressure or temperature, to

simply turn a flow off or on, or to separate two pieces of equipment [178]. Regarding
their function, they can be divided into isolation valves and control valves. The differ-
ence is that isolation valves are actuated by an operator, and their states are “open”
or “closed”, whereas control valves are operated automatically and it is decisive that
a certain intermediate state between “open” and “closed” can be continually main-
tained.
12.3.1 Isolation valves
Isolation valves must reliably isolate two sections of the pipe against each other, even
after a long operation time [179]. Leakage to environment must be avoided due to fire
danger or emission control. There are several kinds of valves which have their partic-
ular pros and cons. They are further explained in [179] and [178].
1. Globe valves (Figure 12.11):
Globe valves can be used for a precise flow control. They do not have a dead
storage and close tightly. Their disadvantage is the high pressure drop across the
valve, caused by two 90° turns inside the valve.
2. Ball valves (Figure 12.11):
Ball valves can be fully opened with practically no additional pressure drop. They
can handle solids and are appropriate for automation for use as control valve.
The leakage is very low, and ball valves can be operated at high temperatures and
pressures. The disadvantages are the remaining liquid holdup in the valve due
to a large dead storage. Electrostatic problems might occur, so some precautions
should be taken if flammable liquids are handled.
3. Gate valves (Figure 12.11):
Gate valves are designed to be fully open or fully closed. In case they are fully
open, they do not show an additional pressure drop. Lubricants are not neces-
sary. Gate valves are tight and open and close slowly so that fluid hammering is
avoided. Their disadvantage is that gate valves do not have a gradual valve char-
acteristics. They are more or less either open or closed, they are not appropriate
for use as control valve. In the partially open state, the valve can start vibrating,
which leads to damage with time [178].
4. Membrane valves (diaphragm valves, Figure 12.11):

Membrane valves are completely tight; however, their pressure drop is consider-
able, and they are not appropriate for high temperatures and pressures or dirt.
The mass flow control is not gradual. Due to its tightness, it is considered to be
suitable for special cleanliness demands; so it is very popular in pharmaceutical
applications.
1 2 3 4
Fig. 12.11: 1 = Globe valve, 2 = Ball valve, 3 = Gate valve, 4 = Membrane valve.
© KSB Aktiengesellschaft.
5. Plug valves (Figure 12.12):

Plug valves have also a dead storage, and the additional pressure drop is also very
low as well as the leakage. The disadvantages are the high turning moment for
operation and the possible contamination of the product with lubricant. They are
usually not used for control purposes.
6. Butterfly valves (Figure 12.12):
Butterfly valves have a low pressure drop. They are tight, have no leakage to en-
vironment and no dead storage. They open gradually and are appropriate for use
in control applications. Maintenance is easy. Excentric butterfly valves are even
appropriate for high pressures and temperatures. The main disadvantage is that
the disc and the shaft are in the flowpath of the fluid. Highly abrasive media will
erode the disc, and it is difficult to clean the valve.
Fig. 12.12: 5 = Plug valve [180].

6 = Butterfly valve.
© Heather Smith/Wikimedia Commons/CC BY-
3.0. https://creativecommons.org/licenses/by/
5 6 3.0/deed.en.
12.3 Valves | 293
7. Check valves (Figure 12.13):

Check valves ensure that flow can only take place in one direction. They pre-
vent backflow from higher lines and vessels or from high-pressure regions to low-
pressure regions. The construction must be carried out in a way that they have
no flow resistance in one direction and complete block for the reverse one. There
are different types [178]. In most guidelines, for safety purposes check valves are
ignored to be on the safe side, even if it is obviously wrong.
Fig. 12.13: Examples of check valves.

© KSB Aktiengesellschaft.
In contrast to the normal function, these valve types can also be used as shut-off
valves, where they are operated automatically by the DCS system to realize the action
of an interlock. In this case, they shall have no additional pressure drop and close
completely.
To prevent manually operated valves from maloperation, they can be secured. One
option is the car sealed valve, where a simple seal made of plastic must be broken on
purpose before the valve can be actuated. The valve is then protected against acciden-
tal maloperation.
A more rigorous measure is the locked valve, which is secured with a padlock or
a chain and can only be actuated after it is unlocked with a key. The key can e.g. only
be obtained after a signing procedure.
The costs for normal isolation valves are usually not a decisive issue. However, for
large pipe diameters it should be carefully checked whether they are really necessary,
as an isolation valve in a 32′′ line can easily exceed the price of a medium-sized car.
12.3.2 Control valves
Control valves are used to control quantities like flow, pressure, temperature or liq-
uid level by fully or partially opening or closing in response to signals received from
controller devices that compare a “setpoint” to a “process variable”.
The opening or closing of control valves is usually done automatically by electri-
cal, hydraulic or pneumatic actuators. In Figure 1.2, a standard arrangement around a
control valve had been shown. The valve is actuated by an electric signal with instru-
ment air. To save one stage in the nominal size, the line to be controlled is restricted
upstream and expanded again downstream the valve. The control valve can be iso-
lated by two gate valves in case maintenance is needed. For this purpose, there are
also two drain valves on both sides of the control valve so that the line can be emptied
completely. There is a bypass line with a ball valve around the control valve, so that
during maintenance the flow can be controlled manually. The valve in Figure 1.2 has
been defined to be failsafe closed (FC). This means that in case the instrument air or
other necessary utilities fail the valve takes the closed position, in contrast to failsafe
open (FO). It must be defined in advance during basic engineering which position is
the safe one.
The usual way for the characterization of valves for liquids is the KV -value, which
is the amount of water in m3 /h that flows through the valve at a pressure drop of 1 bar.
It can be written as
KV V̇ ρ bar
= √ . (12.50)
m3 /h m3 /h g/cm3 ∆p
The KV -value for a fully opened valve is called KVs .
Example
17 m3 /h liquid ethanol (ρ = 790 kg/m3 ) passes a valve with a pressure drop of ∆p = 3 bar. Which KVs
value is necessary?
Solution
The necessary KV value can be determined according to Equation (12.50):
KV 1
= 17√ 0.79 = 8.7 .
m3 /h 3
The KVs value should be 30 % larger, i.e.
KVs = 8.7 m3 /h ⋅ 1.3 = 11.3 m3 /h .
The KV -value can be transferred into a ζ -value according to the following procedure:
The combination of Equations (12.18) and (12.50) gives
ρ kg K 2
105 Pa = ζ w2 = ζ ⋅ 500 3 V2 (12.51)
2 m A
with A as the cross-flow area of the pipe. Solving for ζ yields
105 Pa A2
ζ = 2
(12.52)
500 kg3 KV
m
or
d 4 KV −2
ζ = 1.6 ⋅ 10−3 ( ) ( 3 ) , (12.53)
mm m /h
referring to d as the pipe diameter.
For gases, we distinguish between subcritical and supercritical flow (Chap-

ter 14.2). The KV value is written as
V̇ N ρN T1 (p1 − p2 )p2 −1
KV = √ ( ) (12.54)
514 m3 /h kg/m3 K bar2
for subcritical flow with (p1 − p2 ) < p1 /2 and
V̇ N p1 −1 ρN T1
KV = ( ) √ (12.55)
257 m3 /h bar kg/m3 K
for supercritical flow with (p1 − p2 ) > p1 /2. The indices denote for
N standard state (p = 1.01325 bar, T = 273.15 K);
1 valve inlet;
2 valve outlet.
The relevance of p1 /2 is discussed in Chapter 14.2.
12.4 Measurement devices
To keep all the process quantities controlled, it is certainly obligatory to evaluate them
by measurement with the appropriate accuracy. As a rule of thumb, 10–20 % of the in-
vestment costs of a plant are spent for measurement, control, and process automation.
The implementation of the concepts is usually done by designated specialists, while
some foundations of measurement should be considered by any process engineer.
The most important process quantities to be measured are temperatures, pres-
sures, pressure differences, flows, levels and concentrations. They are briefly dis-
cussed in the following paragraphs.
– Temperature:
One of the most important principles of measurement in chemical plants is that
the signal has to be transformed into an electrical signal which can be sent to the
process control system, where it can be visualized to the operators and possibly be
used for control applications. For the temperature, the most important thermome-
ters are resistance thermometers and thermocouples. Resistance thermometers
use the temperature dependence of the electrical resistance. They are the most
accurate devices and can be used in the temperature range −250–1000 °C. A well-
known thermometer is the Pt-100, measuring the resistance of a platinum wire,
which is 100 Ω at 0 °C.
Alternatively, thermocouples are used in an even wider temperature range of
−200–2000 °C. The principle is that a voltage is built up between a soldering
point of two wires of different materials and the free ends, if the soldering point
is exposed to a different temperature.
In the process, care must be taken that the thermometers are placed in a way that
they take representative temperatures, it has to be avoided to place them in dead
zones, where they are more or less isolated from the process.
– Pressure:
In contrast to the temperature, the pressure is uniform in a certain area unless
there is a defined reason for change, e.g. hydrostatic effects or pressure drops due
to friction losses. In most cases, the pressure is transformed into the elastic de-
formation of a spring. The movement of the spring is transformed in an electrical
signal, often by means of the deformation of a metal membrane, which is turned
into a signal by a piezo element. Using different manometers, the range from a
few mbar up to more than 1000 bar can be covered.
It is often amazing how much confusion is caused when it has to be clearly indi-
cated whether the absolute pressure or the gauge pressure, the difference between
absolute pressure and ambient pressure, is meant. While people from plant op-
eration stick to the gauge pressure, scientists and simulation people can hardly
imagine that anything else than absolute pressure could be meant. The only way
to overcome this is to clearly indicate it writing “g” for gauge (e.g. “barg”) and “a”
for absolute (e.g. “bara”). The latter abbreviation is unknown to most people; at
least it causes a further inquiry, and the possible misunderstanding is overcome.
– Pressure difference:
The measurement of pressure differences is important to get information about
hydrostatic pressures or pressure drops. It is measured in a similar way as the
pressure itself, both pressures are connected to different sides of the spring. Pres-
sure differences cannot be evaluated by measuring both absolute pressures and
taking the difference, as the difference of large numbers can be considerably er-
roneous (p. 261).
– Flow:
Today the dominating measurement principles for the flow are the Coriolis type
and the Vortex type flow meters.
The Coriolis flow meter is more expensive, but its accuracy is remarkable. It mea-
sures the mass flow with an uncertainty of approx. 0.2 %, covering the range from
60 g/h–120 t/h, at pressures up to 900 bar [181]. Figure 12.14 illustrates the princi-
ple, although it must be emphasized that a number of arrangements are possible.
When vibrations are initiated to the tube bends, the two pipe branches stay in
parallel in case there is no flow as on the left-hand side. When there is flow, the
pipe branches are differently affected by the Coriolis force and distorted, as on the
right-hand side of the figure. The extent of the distortion is strongly related to the
mass flow through the device [181]. The amplitudes of the induced vibrations are
too small to be seen (approximately 30 µm), but they can be detected tactually.
The great advantage of Coriolis flow meters is that they measure the mass flow
directly; they are independent from other properties of the flow and from inlet or
profile effects of the flow. Apart from the price, the disadvantages are that Cori-
Fig. 12.14: Illustration of the Coriolis flow meter principle.

© Cleontuni/Wikimedia Commons/CC BY-SA 2.5
olis flow meter need a homogeneous flow and that fouling causes errors in the
measurement.
In contrast to the Coriolis flow meter, the vortex flow meter evaluates a volume
flow, which can be converted into a mass flow by an additional temperature mea-
surement and an appropriate density relationship. It counts the number of vor-
tices formed after an obstacle in the flow path, using a piezoelectric crystal. The
accuracy can be estimated to be 0.75 %. It needs a certain inlet zone. Also, it is not
necessary to know other properties of the stream, such as viscosity. Vortex flow
meters have a huge turndown (approx. 1 : 50 from the lowest to highest value) and
can be used in a wide temperature range (approx. −200–400 °C). They are not
appropriate for low flows, fouling media, and in the case that vibrations occur in
the plant.
A third type which is often used is the magnetic flow meter. The physical principle
is that a magnetic field is applied to the metering tube. Charged particles like ions
will be diverted perpendicular to the flow. This results in a potential difference
proportional to the flow velocity. For application, the fluid must have a minimum
electrical conductivity (> 0.5 µS/cm), and the tube must be electrically isolated.
Magnetic flow meters have no movable parts and no additional pressure drop,
and they are appropriate for aggressive and corrosive fluids. It is an application for
liquids; solids or gas bubbles do not matter. The temperature is limited to 200 °C,
and the minimum flow velocity is 0.5 m/s.
There are many other types of flow meters, which are well and briefly described
in [182].
– Level:
There are a lot of measurement principles for the liquid level. One must distin-
guish between a continuous level indicator and a level switch, which detects when
the level reaches a certain value. The simplest and safest device is the inspec-
tion glass; however, the transformation into an electrical signal does of course
not work.
Level indicators can be based on the buoyancy principle. The more a displacing
piston is dipped into a liquid, the more it is exposed to buoyancy forces which can
be transformed into electrical signal by resistance strain gauges. The drawback is
Sample Sample
Fig. 12.15: A bad (A) and a good (B) example

(a) (b) for sampling.
the mechanical equipment, which might be sensitive to dirt, and that the density
of the liquid, which depends on temperature and composition, must be known.
Air bubblers work in a similar way. Air is bubbled through a dipped pipe into the
liquid. A pressure sensor measure the pressure necessary to overcome the hydro-
statics. This method is not sensitive to dirt, but a disadvantage is that the liquid
is contaminated with the gas, which might not be desirable in all cases. Also, the
hydrostatic pressure can directly be measured and transferred into a liquid level,
taking the liquid density into account.
A number of other electrical signals can be used for liquid level detection or mea-
surement, like electrical conductivity, capacity, radar sensors, microwave or in-
frared sensors or radiometric signals. A useful option is the so-called liquiphant,
which is in principle an oscillating tuning fork. If the liquid level reaches or drops
below the liquiphant, its resonance frequency changes. This is detected and trans-
formed into a signal for a high-level or low-level alarm.
– Analytical measurement:
Analytical measurements are of course done with GC (gas chromatography), HPLC
(high performance liquid chromatography), Karl-Fischer-titration (determination
of the water content), and so on. All these methods are the tasks of designated spe-
cialists and should not be covered in one short paragraph. However, all analytical
methods considerably depend on a good design of the sampling, where it has to
be ensured that the sample is representative. Figure 12.15 shows an inappropriate
and an appropriate example for taking a sample. In example A there is a parallel
branch to the line. It can be initiated by opening the two valves. However, when
the valves are closed again, it is not ensured that a representative sample is ob-
tained. The question is how it is ensured that the previous content in the sample
container (e.g. air, rests from previous sample) has really been replaced. There is
no real motivation for the flow to pass the sample container; the short cut through
the main line has probably a lower pressure drop than the way through the sam-
ple box. Furthermore, the velocity in the sample container is low due to the larger
cross-flow area. In example B the sample container is connected to both the pres-
sure and the suction side of the pump. Therefore, there is a considerable pressure
difference across the sample container, resulting in a well-defined backflow to the
suction side of the pump. The previous content of the sample container is rapidly
replaced. The disadvantage is that it will go through the pump again, and part of
it will again enter the sample container. However, with time the old content will
be more and more diluted, and finally, the sample will be representative.
13 Utilities and waste streams
13.1 Steam and condensate
In process industry, steam is the most widely used heating agent. Most chemical sites
provide a steam net where steam at several pressures is provided. The costs for steam
is a significant criterion for the choice of a site, however, as the costs of a process are
usually determined by the costs of the raw materials, it is rarely decisive. Low pressure
steam is not necessarily cheaper than high pressure steam; usually, the steam genera-
tors produce steam at very high pressure (e.g. 400 °C, 50 bar), which is then throttled
down in a valve to the lower pressure levels (e.g. 210 °C, 17 bar as medium pressure
steam and 170 °C, 6 bar as low pressure steam). For use, steam should not be too far
superheated so that condensation can take place rapidly with an extraordinarily high
heat transfer coefficient in the equipment. For desuperheating, condensate is often
injected to the steam by means of specially designed nozzles.
Example
How much steam condensate (100 °C, 20 bar) must be added to reduce the superheating to 5 K if a
stream of 1000 kg/h high pressure steam (superheated at 400 °C, 50 bar) has been throttled down to
a middle pressure level of 20 bar?
Solution
According to the steam table [183] or using a high-precision equation of state (e.g. [29]), we can set
up the energy balance of this desuperheating process
ṁ HP hHP + ṁ Cond hCond = (ṁ HP + ṁ Cond ) ⋅ h(Ts (20 bar) + 5 K, 20 bar) (13.1)
with
hHP = 3196.7 J/g
hCond = 420.6 J/g
Ts (20 bar) = 212.4 °C
hfinal = h(212.4 °C + 5 K, 20 bar) = h(217.4 °C, 20 bar) = 2813.8 J/g.
Solving Equation (13.1) for ṁ Cond gives
hHP − hfinal 3196.7 − 2813.8

ṁ Cond = ṁ HP = 1000 kg/h = 160 kg/h . (13.2)
hfinal − hCond 2813.8 − 420.6
One should be aware that the heat transfer of superheated steam does not take place
in a two-step sequence consisting of cooling down the steam as a vapor to condensa-
tion temperature and subsequent condensation. For the heat transfer, this would be a
300 | 13 Utilities and waste streams
disaster, as the heat transfer coefficient for cooling of a vapor would be poor in com-
parison with the steam condensation and may determine the size of the condenser.
In fact, for moderate superheating the condensation remains essentially the same as
for a saturated vapor, the only thing which changes is the larger heat to be transferred
due to the superheating [184]. People who wear glasses intuitively know this, as dur-
ing winter time the glasses grow damp immediately after entering a building; it is not
necessary that the whole air in the building is cooled down to dew point temperature.
For moderate superheating, condensation takes place immediately in a technical con-
denser as well. In [79], a criterion is set up to decide in which cases a superheating can
be considered as moderate; however, it is again pointed out that it is definitely disad-
vantageous to regard a part of the condenser as a gas cooler for design.
For each sort of steam, a so-called header is provided, carrying the whole steam
from the tie-in point or from battery limits to the plant and branching off to the various
consumers of the plant. The condensates are collected as well in a condensate header
and usually pumped back as boiler feed water to the steam generator. In some cases,
the steam is used as direct steam, meaning that it is introduced directly into the pro-
cess, e.g. into the bottom vessel of a column where the bottom product is water anyway.
In this case, no condensate can be returned, and often additional costs occur as the
whole amount of steam might end up as waste water. When steam is directly used, it
should also be considered that it contains small amounts of caustic substances, e.g.
ammonia or amines. One should make sure that this has no detrimental influence on
the process. The great advantage of using direct steam is that a reboiler can be omitted.
The use of steam as heating agent has some remarkable advantages. In contrast
to heating agents making use of sensible heat (e.g. hot oil), the temperature stays
constant, as water condenses as a pure substance. It is not necessary to convey the
heating agent to the consumer; it is delivered with a certain pressure which is higher
than the pressure at condensation. The condensation is the conveying mechanism for
the steam. The specific volume of steam is by far larger than the one of the condensate:
At p = 2 bar, v󸀠󸀠 = 0.8857 m3 /kg compared to v󸀠 = 0.00106 m3 /kg, corresponding to
a factor of 835. When steam condenses, the volume decreases drastically, and fresh
steam can follow to maintain the pressure where the heat is obviously consumed. The
heating agent is flowing to the area where it is required without any stimulation. The
only thing which has to be provided are lines with a sufficient cross-flow area for con-
veying the steam without major pressure loss. Of course, it is important that there are
no inerts in the steam.
Moreover, even the appropriate amount of steam will flow to the heat exchange
area. Figure 13.1 shows an arrangement with a heat exchanger where the steam flow
is controlled.
The heat flux through the heat exchanger is given by
Q̇ = kA(Tcond − Tproduct ) = ṁ steam ∆h v . (13.3)

Steam
Fig. 13.1: Control of the steam flow

Condensate as the heating agent.
Following this simple equation, one should be aware that the k value is mainly deter-
mined by the product side, as the heat transfer coefficient on the steam side is very
high and does not affect k very much. Also, within a certain range the enthalpy of
vaporization as a physical property of the steam can be regarded as constant, and A
as the heat transfer area of the heat exchanger does not change anyway. Therefore, the
transferred heat flux is mainly determined by the condensation temperature Tcond of
the steam, which has to be adjusted in an appropriate way for the control of the prod-
uct temperature. As the steam is a pure substance, the condensation temperature is
directly related to a pressure according to the vapor pressure line. The signal for the
control valve varies the discharge opening until it causes a pressure drop which yields
the desired condensation pressure of the steam.
For a comfortable controlling, the pressure drop should not be too low; the rule
of thumb is 10–20 % or 0.5–1 bar. This pressure drop should be taken into account
when the heat exchanger is designed. The steam conditions given in the design basis
indicate the state of the steam upstream the control valve, for the design of the heat
exchanger, the state of the steam downstream the control valve, after adiabatic throt-
tling, is relevant, including the resulting superheating. This steam state must fit to the
desired state in the heat exchanger.
Example
In a thermosiphon reboiler (Chapter 4.5), the bottoms product is to evaporate at t = 100 °C. Low-
pressure steam (LPS) at t = 160 °C, p = 5 bar will be used for heating. For the driving temperature
difference, a value of 30 K is targeted. Calculate the steam state relevant for the heat exchanger design.
Is enough pressure drop for the steam control available?
Solution
The condensation temperature of the steam shall be tcond = 100 °C + 30 K = 130 °C. According to the
steam table, the corresponding condensation pressure is p = 2.7 bar. Adiabatic throttling of the LPS
to this pressure gives a steam state t = 151.8 °C, p = 2.7 bar. The pressure drop across the valve is
sufficiently high (∆p = 2.3 bar or 46 %). The superheating of 21.8 K is acceptable. Otherwise, steam
saturation would have to be applied.
The symbol in the condensate outlet line in Figure 13.1 represents a so-called steam
trap, a device which lets liquid pass and closes if vapor is about to leave the system
without condensing. Thus, it is ensured that any steam entering the heat exchanger is
condensed, as it cannot leave the system as vapor. There are several function princi-
ples [185]. The mechanical one is the simplest. A lever gauge rises if liquid comes and
lowers if it is located in vapor (Figure 13.3). It is connected to a lever which opens and
closes an opening, respectively.
Steam traps are often supposed to be not appropriate for fouling and dirty me-
dia. An alternative scheme without using a steam trap is shown in Figure 13.2 [96].
An extra vessel with level control can take over the function; the pressure-equalizing
line is necessary, otherwise, noncondensed steam might accumulate in the vessel. If
the pressure of the condensate must be increased, a pump can be installed below the
vessel.
Steam Equalizing line
Condensate LC
Fig. 13.2: Steam control without using a steam trap [96].
Air cock
Fig. 13.3: Sketch of a ball float steam trap with air cock for venting [186].
© 2016 Spirax Sarco Limited.
Steam
Condensate Fig. 13.4: Control of the condensate flow.
The alternative to the control scheme in Figure 13.1 is the control of the condensate flow
(Figure 13.4). The advantage is that the control valve can be smaller, as the condensate
has by far a lower volume. The steam condenses at its delivery pressure. To reduce the
heat flux, the control valve throttles the flow. The condensate accumulates in the heat
exchanger and covers part of the heat exchange area. The heat transfer to the liquid
is much lower than for a condensing vapor, and furthermore, the temperature of the
condensate goes down when it is used as heating agent. Therefore, part of the heat
exchange area is not used and the heat flux is reduced as requested.
Increasing the heat flux is unsatisfactory. It is only possible if part of the heat ex-
changer is already flooded. This means that a control in both directions can only be
performed if the heat exchanger is designed in a way that part of the tubes are flooded
at normal operation, meaning that possible heat exchange area is wasted. Further dis-
advantages are [185]:
– The control valve does not necessarily prevent uncondensed steam from passing
the heat exchanger. Therefore, an additional steam trap is necessary, as shown
in Figure 13.4. Otherwise steam could be lost, and the pressure in the condensate
header might increase, making the condensate removal of other heat exchangers
in the process more difficult.
– The controllability of the process is worse than with the steam flow control. For
example, if the heat flux should be reduced from full power to a very low value,
the steam stops entering the heat exchanger when the whole apparatus has been
flooded with liquid. For example, if the volume of the shellside is 1 m3 , approx. 1 t
of steam is consumed after the control valve has been closed. With the steam inlet
control (Figure 13.1), only the steam already being in the shellside condenses. At
p = 2 bar, the density of the saturated vapor is ρ = 1.13 kg/m3 , giving an undesired
1
steam consumption of 1.13 kg, which is approx. 900 of the condensate flow control.
This means that the condensate flow control is much slower.
– At the phase boundary between steam and condensate increased corrosion might
be observed.
– For horizontal reboilers, thermal stress is an important issue. The upper and the
lower tubes are exposed to different temperatures, as hot fresh steam enters the
heat exchanger at the top and the condensate at the bottom might be significantly
subcooled.
For heat-integrated columns, the control of the condensate flow is the preferred op-
tion, as the pressure drop of the inlet valve reduces the driving temperature difference
for the evaporation, which is the critical issue in heat integration.
It might happen that due to the throttling in the valve and the pressure drop of
the steam trap the condensate outlet pressure of the consumer becomes too low to
convey it to the condensate header. To avoid an additional pump, a so-called con-
densate lifter can be used. Figure 13.5 shows a possible arrangement. The function is
as follows. Coming from the steam trap, the condensate is collected in a vessel with
three nozzles, A, B, and C. Inside the vessel there is a lever gauge, which is mechani-
cally connected with the three nozzles (Figure 13.6). With rising liquid levels, the lever
gauge is taken upward. The connections then close the condensate inlet B and open
the condensate outlet C and the steam inlet A. Through A, the condensate in the vessel
is pressurized by the steam and can therefore flow to the condensate header. The lever
gauge sinks again, closing the nozzles A and C and opening nozzle B. The remaining
steam in the condensate lifter can be vented. Condensate lifters work automatically
and without maintenance. The steam as auxiliary energy is available anyway, and the
additional steam consumption is approx. between 0.1–1 %, as the gaseous steam has
to replace the liquid volume of the condensate and the volume ratio between steam
and condensate is in this range, depending on the pressure.
Steam
Condensate
A
To condensate
header
C B
Fig. 13.5: Condensate lifter process.
Fig. 13.6: Sectional view of a condensate lifter [187].

© 2016 Spirax Sarco Limited.
Although essentially only water, condensate is a valuable substance as it has been pu-
rified so that no more salts or other substances are present. There is a certain value just
from its energy content, as the sensible heat of liquid water is approx. 10 % of the heat
transported as steam¹. In European terms, this corresponds to 2–3 €/m3 , without the
additional costs for purification, conditioning and waste water disposal. Therefore, it
is collected from the particular consumers and recycled to the steam generator, where
it is again preconditioned as boiler feed water. A condensate net has in most cases a
number of consumers, often in large distances from each other. One must be aware
that the condensate production is not constant in time, and each consumer delivers
his own condensate outlet pressure. Thus, there are always fluctuations in the con-
densate header, and the pressure in this line should in general be low so that none
of the consumers has difficulties to get rid of the condensate. The condensate line
will in most cases end up in a vessel having a pressure slightly above the ambient
one (e.g. p = 1.2–1.3 bar), and the condensate line is usually operated a bit higher
(e.g. p = 1.5 bar). As the condensate is close to the saturation state when it enters the
condensate line, vapor will be generated in the condensate line due to the expansion.
From the mass fraction point of view, it is not too much; however, it has a considerable
volume, as the following example shows:
Example
A saturated low-pressure steam condensate at p1 = 6 bar is expanded into the condensate line to
p2 = 1.5 bar. How much vapor is generated?
Solution
Calculating an adiabatic throttling, the vapor generation in the condensate line is 9.13 %. The densi-
ties of vapor and liquid in the saturation state are
ρ󸀠󸀠 = 0.863 kg/m3 ,
ρ󸀠 = 949.92 kg/m3 .
Therefore, per kg condensate one must expect a vapor volume of 0.106 m3 and a liquid volume of
0.001 m3 . This means that from the volume point of view only approx. 1 % of the condensate is liquid.
This is a quite typical result. A condensate line is not a water line but a steam line
with a certain amount of liquid. The condensate might remain liquid if some subcool-
ing takes place, e.g. in a long condensate line during winter, but in general it should be
designed as a two-phase line (Chapter 12.1.4). The exact dimensioning of the conden-
sate line is difficult, as all the different operation modes of the particular consumers
1 However, on a low temperature level. Calculated as ((hL (100 °C, 1.1 bar) − hL (30 °C, 1.1 bar))/
(hV (250 °C, 20 bar) − hL (30 °C, 1.1 bar)).
and the significant influences of insulation, ambient temperature and roughness of

the inner surface of the line can hardly be determined [185]. The following items are
recommended [185]:
– The line should be designed as short as possible and have a base slope of at least
1 %, ensuring that the line drains itself in a shutdown.
– For the pressure drop, ∆p = 0.1 bar/100 m is recommended.
– The tie-in of the consumers should be done from the top.
– An injection for rapid mixing makes sense if the condensate temperatures signif-
icantly differ. Otherwise, water hammering might occur if flashed steam bubbles
from hot condensate instantly condense in colder condensate. Their volume col-
lapses immediately, and liquid water fills it with high velocities.
– It must be possible to drain each section of the condensate header completely.
As any long line, the condensate header will have low points where special care
should be taken.
13.2 Heat transfer oil
At very high temperatures (> 250 °C) the use of steam becomes more and more difficult,
as the condensation pressures and therefore the design pressures become large, mak-
ing the heat exchanger expensive. In these cases, it makes sense to use a heat transfer
oil at comparably low pressures. Heat transfer oils can withstand very high tempera-
tures and have low vapor pressures. Also, they remain liquid at very low temperatures.
One must be aware that it is made use of sensible heat, meaning that the temperature
of the heating agent changes and that much larger mass flows are necessary. A com-
pilation of common heat transfer oils is given in [188]. As well, melted salts are used
as heating agents.
13.3 Cooling media
Cooling is much more sophisticated than it seems. It requires the infrastructure of a

site, and each cooling medium has its own restrictions.
Sites are usually located at a natural water reservoir. Its temperature is limiting
the lowest achievable temperature in the process which can only be underrun by ad-
ditional technical equipment. If the site is located at the sea, a sea water cooling cycle
is operated. However, sea water is one of the most aggressive media due to its salt
content. Using it in the process is practically impossible, as valuable materials of con-
struction like hastelloy would have to be chosen. The only economic way to make use
of sea water is the indirect way by operating a secondary cycle of demineralized water
for the whole site. Figure 13.7 shows a number of huge plate heat exchangers, cooling
the water cycle for a site operating with several plants with sea water. Plate exchangers
13.3 Cooling media | 307
Fig. 13.7: Plate heat exchangers for sea water cooling.
for sea water must be made of a material which can withstand the corrosion attacks
by the salt. The incoming sea water must be carefully filtered, and the cycle must be
treated with biocides regularly.
Cooling water is usually taken from a river. Its supply temperature is usually be-
tween 28–32 °C, with a supply pressure of approx. 5–6 bara. Usually, its return temper-
ature can be 10 K higher, i.e. approx. 40 °C. Before returning it back into environment,
it is cooled down again with a cooling tower, as the oxygen content of the water be-
comes unacceptably low at temperatures beyond 26–28 °C, causing suffocation of the
fish in the river. The use of normal cooling water has its pitfalls, as it contains wa-
ter hardness components. When it is heated up, these components (CaSO4 , CaCO3 )
tend to precipitate. This kind of fouling must be avoided. Therefore, cooling water can
only be used up to a certain temperature level of the product side, usually 65–75 °C,
where the wall temperature on the cooling water side is regarded. Cooling water costs
are approx. 0.05 €/m3 . Considering the above mentioned difference between supply
and return temperature, a cooling water flow of approx. 86 m3 /h represents a cooling
power of 1 MW, as
86 m3 /h ⋅ 1000 kg/m3 ⋅ 4200 J/(kg K) ⋅ 10 K = 1.003 MW .

If cooling water cannot be used due to a temperature level on the product side higher
than mentioned above, there are mainly two options.
First, one can install a secondary water cycle (“jacket water”) operated with dem-
ineralized water as cooling agent. This cycle itself is then cooled by original cooling
water at moderate temperatures. A jacket water cycle means additional investment
costs, as one huge heat exchanger for cooling the jacket water is necessary. This heat
exchanger needs a certain temperature difference as driving force, thus, the supply
temperature of the jacket water is about 5 K higher than the cooling water supply tem-
perature.
The second option is the use of air coolers, which usually cause a larger invest-
ment due to the bad heat transfer on the service side (Chapter 4.8).
With cooling water, temperatures down to approximately 35 °C can be achieved
on the process side. To realize lower temperatures in a process, refrigerators must
be used. Refrigerated water can be supplied with cold water aggregates at temper-
atures down to 2 °C. Below this temperature, ice formation must be considered. In
most cases, brine is used, often a mixture of propylene glycol/water or ethylene gly-
col/water, which can be used below 0 °C as well. A compilation of heat transfer fluids
can be found in [189].
13.4 Exhaust air treatment
Exhaust air is defined as the sum of gases, vapors, smokes, dusts, soots, and aerosols
released to atmosphere that have an impact on the composition of natural air [190].
The definition might be a bit weak; it means that the release of components like carbon
dioxide or hydrogen, which are not regarded as air pollutants, does not need to be
restricted². Nevertheless, some of the components occurring in air are considered to
be pollutants, as it is the case e.g. for methane or methyl chloride, which are natural
air components due to the metabolism of animals or plants.
Many of the most interesting fine chemicals, specialty chemicals, and pharmaceu-
ticals are produced batchwise. This implies that it has to be carefully evaluated how
frequently the particular streams occur and what their compositions and amounts are.
In batch processes, exhaust air can be produced in many ways (Figure 13.8). The most
common one is the charging of vessels. If a liquid is stored in a vessel, there will be
a saturated gas phase above the liquid, filling up the rest of the vessel volume. When
additional liquid is filled into the vessel, an equivalent vapor volume is displaced up-
ward, usually directly into the exhaust air line. A similar mechanism is so-called vessel
2 Carbon dioxide is a special issue. The vast amounts of CO2 emissions from combustion of coal and
natural gas are regarded to be responsible for the continuous rise of the CO2 concentration in the air,
currently considered to be resulting in a future climate change. CO2 emissions coming directly from
chemical processes occur in relatively small amounts.
Vessel filling, Flushing

Vessel breathing procedure
PC
Pressure controlling, Gaseous

expansion Reaction by-products Fig. 13.8: Ways for the production of exhaust air
mixture in batch processes.
breathing. If a vessel has an open connection to the environment, air will be sucked in
when the content of the vessel cools down and contracts, for instance at night. When
the vessel is heated up again, its content expands, and the air, now saturated with
the vapor being in equilibrium with the liquid in the vessel, is displaced into the ex-
haust air line. Other well-known mechanisms are the depressurization of vessels, the
flushing of vessels in order to dilute and remove gaseous substances from a vessel, or
simply the removal of gaseous by-products out of a reactor by pressure relief.
Exhaust air has to obey certain restrictions depending on the country where the
plant is located. For example, in Germany, the TA Luft [191] is the decisive regula-
tion. Table 13.1 gives an overview on the particular limitations. The pollutants are
assigned to certain classes according to their hazardous potential. Each class has a
bagatelle limit, i.e. any industrial unit has the right to release this amount without be-
ing prosecuted. Beyond this value, a limiting concentration has to be complied with.
Table 13.1: Limiting concentrations and bagatelle limits according to TA Luft [191].
Lim. conc. Bagatelle lim. Counted Examples

(mg/Nm3 ) (g/h) as
Organic substances 50 500 C methanol, ethyl acetate

Org. subst., cl. I 20 100 acetaldehyde, vinyl acetate
Org. subst., cl. II 100 500 acetic acid, nitromethane
Carcinogenic substances
Carc. subst., cl. I 0.05 0.15 As, Cd
Carc. subst., cl. II 0.5 1.5 acrylonitrile, ethylene oxide
Carc. subst., cl. III 1 2.5 benzene, vinyl chloride
Anorganic substances
Ammonia 30 150
Hydrogen cyanide 3 15
Chlorine compounds 30 150 HCl
Nitrous oxide 350 1800 NO2
Sulphuric oxides 350 1800 SO2
Bromine compounds 3 15 HBr
Carcinogenic substances are treated in an extraordinarily strict way. In some cases,

it is defined how the particular pollutants are counted, for example just the carbon
content for simple organic substances or the typical combustion products for chlorine
and bromine compounds (HCl and HBr). The limiting values are generally based on
the state-of-the-art in removing the particular substances from gas streams. They can
be regarded as very strict, especially the latest amendment from December 2001 has
lowered many limiting values by a factor 2 or more.
There are several options for exhaust air treatment. They can be categorized in
two ways. Combustion processes and biological treatments destroy the pollutants,
whereas condensation, adsorption, absorption, or membrane processes separate
them from the air, and if they are valuable and their purity is sufficient, they can be
recovered. Also, the load of pollutants and the amount of exhaust air are decisive
for the choice of the process. Condensation can only be used for comparably small
exhaust air streams. As it will be shown, only cryo-condensation can normally fulfill
the TA Luft, where the commercial units have a defined size. The load is not important.
Membrane and absorption processes can also handle only comparably small amounts
of exhaust air due to the size of the common commercial units. For absorption, high
loads are advantageous to make it worth to recover the pollutants. For membrane
processes, complete separations can hardly be achieved, therefore, it makes only
sense for low concentrations. Combustion is an effective but expensive exhaust air
treatment. Therefore, a high exhaust air stream is required to make it worth. Thermal
combustion works for all concentrations, but it is especially effective for high pollutant
contents to save fuel for achieving the high temperatures (850–1200 °C). For low pol-
lutant concentrations, catalytic combustion can be used, as only approx. 400 °C are
necessary. For higher concentrations (> 10 g/Nm3 ), the catalyst might be damaged. Ad-
sorption processes can handle large exhaust air streams, but for high concentrations
it is difficult to remove the heat of adsorption out of the bed. Biological treatment
is only possible for substances that can easily be dissolved in water. It is the best
process for large exhaust air streams with low pollutant concentration, but it needs
careful maintenance. Another important issue is the predictability of the processes.
For the combustion, absorption and condensation processes, a mass balance can in
principle be predicted without experiments. Adsorption and membrane processes
usually need experiments even for the design, unless references are available. The
lifetime of membranes for a new task can hardly be estimated, as often many different
components are involved and come into question for spoiling the membrane. This
causes great uncertainties for the final investment costs. Therefore, in most cases
membranes are unlikely to be the best choice. Biological exhaust air treatment pro-
cesses cannot be predicted at all. Long-term trials must be carried out with changing
operation conditions to get a feeling about its performance. In fact, most exhaust air
problems require an immediate solution where someone can give a warranty for the
performance. Therefore, it is often not appropriate to suggest adsorption or biological
treatment, even if the process itself might be interesting.
13.4.1 Condensation
If three or more people are sitting together in a meeting to discuss a condensation project, at
least one of them will have the glorious idea that it is sufficient to cool down to the lowest normal
boiling point.
(Engineering wisdom, possibly a law of nature)
Condensation looks very simple but is not, as we have no feeling for even basic ther-
modynamics. Thus, as cited above many people think that a component completely
condenses when the temperature falls below its normal boiling point. That would
make things pretty easy: for methanol, 65 °C would be a sufficient cooling temper-
ature, for benzene 80 °C, for pentane 36 °C, and so on. Simple cooling water with
t = 30 °C would be an appropriate cooling agent. In fact, things are much more dif-
ficult, as the following example shows.
Example
What is the minimum condensation temperature to guarantee that the methanol concentration is in
line with TA Luft (50 mg/Nm3 C; see Table 13.1)?
Solution
32 = 0.375, therefore, the concentration limit is

The carbon fraction of methanol is approximately 12
50 mg/Nm3 /0.375 = 133 mg/Nm3 . From ideal gas law, we get the corresponding partial pressure:
mRT 133 ⋅ 10−6 kg ⋅ 8.31447 J/(mol K) ⋅ 273.15 K

pMeOH = = = 9.4 Pa . (13.4)
MV 32 ⋅ 10−3 kg/mol ⋅ 1 m3
This partial pressure corresponds to a saturation temperature of t = −70 °C! Similar results are ob-
tained for other substances (toluene: −75 °C, n-hexane: −96 °C, ammonia: −145 °C, vinyl chloride:
−154 °C, n-pentane: −116 °C). Even for the removal of a heavy boiling substance like n-dodecane a
condensation temperature of −13 °C is required³.
Therefore, in most cases TA Luft can only be met by cryo-condensation using a cooling
agent like liquid nitrogen [192]. At ambient pressure, liquid nitrogen has a temperature
of t = −196 °C, where almost all substances have a vapor pressure far below the limit
of TA Luft. On the other hand, most of the substances are solid at that temperature
(water!), which makes their handling complicated. As water is hardly avoidable, it is
important to remove it before the stream enters the cryo-unit, e.g. by an adsorption
bed. Often, two twin plants for cryo-condensation are operated alternately; while one
3 The extrapolation of the vapor pressure curves to these low temperatures is not very reliable. The
values must be considered as estimations.
is used for air cleaning, the other one is being defrosted. Substances like cyclohex-
ane form a kind of snow on the cooling coils. After a short time, the cooling coils are
practically isolated, and the heat exchange breaks down.
The most complicated problem in cryo-condensation is the formation of aerosols.
The mechanism of the formation is as follows. If the temperature difference between
cooling agent and bulk fluid is too large, a temperature profile perpendicular to the
flow direction will develop where the temperature falls below the condensation tem-
perature far away from the wall in the gaseous phase. Spontaneous condensation
takes place, and very small droplets are formed with about 1 µm in size. These droplets
have a rate of descent (Chapter 9) that is too small for being separated within the
equipment. Otherwise, they are too large to take part in molecular diffusion towards
the cooling area. Finally, these droplets can pass the apparatus, although they ac-
tually have been condensed. Behind the condenser, they will be evaporated again
and can be detected as pollutants. Therefore, care must be taken that an appropri-
ate temperature of the cooling agent is used and controlled. In practical applications,
the liquid nitrogen is never used directly as cooling agent. Instead, it is just taken to
control the temperature of a secondary cooling cycle, operating with gaseous nitro-
gen. Figure 13.9 shows a typical liquid nitrogen condenser plant with a cooling cycle
(Cryosolv process).
The liquid nitrogen is evaporated in the cycle gas cooler, where the cycle gas is
cooled down to an appropriate temperature. The pollutants are liquefied on the cool-
ing coil of the cryo-condenser. The condensate can be collected at the bottom of the
condenser. If it is a pure substance, it can be used again. The purified gas can be led
into the environment. For a better utilization of the liquid nitrogen, the cooling of the
Cryo-condenser
Exhaust air
Pure gas
(cool)
Cycle gas
cooler
Liquid Condensate
nitrogen Defroster
–196 °C
Cycle gas fan

Recuperator
Pure gas
Gaseous nitrogen
Fig. 13.9: Principle scheme of a cryo-condensation unit [192].

© Messer Group.
cycle gas is supported by the cold purified exhaust gas and the evaporated nitrogen in
the recuperator. The condensate could be used as well, but normally the amounts are
too small.
Cryo-condensers can be purchased as units. Normally, they are designed for
700–1000 Nm3 /h, which can be considered as relatively small exhaust air streams.
Investment costs for cryo-condensers are low in comparison with other exhaust air
treatment processes. The tank can even be rented. A strong point is the use of the
gaseous nitrogen. In principle, it can be released to the environment as well, but
experience shows that the process can only be operated economically if the gaseous
nitrogen can be utilized, for instance for inertization in the plant itself. In these cases,
cryo-condensation is the favorite way for exhaust air treatment. The operation costs
are then comparably low, as the plant manager has to purchase the nitrogen anyway.
The Cryosolv process has been further improved during the recent years, cur-
rently, the DuoCondex process [193] is considered to be the most effective one.
To summarize the aspects of cryo-condensation, the following statements can be
given:
– Cryo-condensation has low investment and high operation costs due to standard
units.
– To keep operation costs low, the evaporated nitrogen must be used elsewhere in
the plant to compensate for part of the operation costs.
– Only liquid nitrogen and electric current are needed as utilities. The tank neces-
sary for the storage of the liquid nitrogen can be rented.
– The predictability is restricted due to the possibility of aerosol formation and the
performance of the vapor pressure line at extrapolation to low temperatures.
– The cryo-condensation systems in general react very slowly to changes in load.
Therefore, the exhaust air stream should be as steady as possible.
– Only relatively small exhaust air streams can be treated.
– Due to solid or snow formation, twin-plants are usually necessary for operation
and defrosting.
13.4.2 Combustion
If a combustion process is used for exhaust air purification, the pollutants are de-
stroyed by chemical reaction. Combustion is simple if the pollutants only consist of
C, H, and O. In these cases, only carbon dioxide and water are formed as combustion
products, e.g.
C2 H5 OH + 3 O2 󳨀→ 2 CO2 + 3 H2 O .
Both can simply be released to environment, as they are natural air components. More
difficulties come up if other elements occur, as new pollutants can be formed that must
be removed. Chlorine is one of the most widely used elements in chemical industry.
In a combustion process, it will completely be transformed to HCl, e.g.
CH2 Cl2 + O2 󳨀→ CO2 + 2 HCl .
This clear statement might be a bit surprising, as one could easily think of water for-
mation, for example according to the Deacon reaction:
2 HCl + 0.5 O2 󴀕󴀬 H2 O + Cl2 .
The Deacon reaction is an equilibrium reaction. For high temperatures as they oc-
cur in the combustion chamber (≥ 900 °C), the equilibrium is more or less completely
on the HCl side. When the flue gas is cooled down in the steam generator (see be-
low) and in the environment, the equilibrium changes to the Cl2 side, but reaction
kinetics is relatively slow so that only small amounts of Cl2 are formed. Therefore,
a rapid cooling procedure must be performed. Details can be found in [11, 190, 194].
HCl can be removed from the flue gas with a scrubber, usually using water or sodium
hydroxide solution as absorptive agents. If significant amounts of chlorine are formed,
water is no more appropriate for scrubbing. Caustic soda can transform the chlorine
to hypochlorite according to
Cl2 + 2 OH− 󴀕󴀬 OCl− + Cl− + H2 O .
The hypochlorite anion can be transformed to chloride in the presence of the bisulfite
ion:
OH− + OCl− + HSO−3 󴀕󴀬 H2 O + Cl− + SO2−
4 .
The other halogen elements (F, Br, and I) behave analogously.

Sulfur will be transformed into sulfur dioxide SO2 . SO2 can hardly be absorbed
with pure water, but a caustic soda solution is quite efficient. For large sulfur loads in
flue gases, other processes like adsorption on activated carbon or reaction with cal-
cium hydroxide to gypsum are known and well established [195, 196].
Chemical bonded nitrogen (ammonia, amines, etc.) will be transformed to nitrous
oxide to a certain extent that can hardly be determined by theoretical predictions [197].
To be conservative, it is often assumed that chemical bonded nitrogen is converted to
NO completely (fuel NO), e.g. according to
2 C2 H5 NH2 + 8.5 O2 󴀕󴀬 7 H2 O + 4 CO2 + 2 NO .
Fuel NO is formed already at low temperatures as 800 °C. NO can be formed by reac-
tions of elementary nitrogen and oxygen from the air as well (thermal NO), especially
at high temperatures. The amount of thermal NO increases dramatically with temper-
ature; for example, the equilibrium concentration of NO in air is 35 ppm at 1000 K and
1300 ppm at 1500 K [197]. When the flue gas is cooled down, NO forms an equilibrium
with NO2 and other nitrous oxides. At ambient temperatures, NO2 is the dominating
component. Therefore all the nitrous oxides are counted as NO2 in mass balances and
referred to as NOx . For NOx removal, so-called DeNOx processes have to be integrated.
They operate with the reaction
4 NO + 4 NH3 + O2 󴀕󴀬 6 H2 O + 4 N2 .
The “denoxation” can be performed at high temperatures (approximately 900 °C)

without a catalyst (SNCR process, selective noncatalytic reduction) or at low tempera-
tures (approximately 300 °C), using metal oxides like V2 O5 as catalysts (SCR process,
selective catalytic reduction). Most plants operate with ammonia as reductive agent.
As an alternative, urea can be used and decomposed to ammonia according to
NH2 –CO–NH2 + H2 O 󴀕󴀬 2 NH3 + CO2 .
Care must be taken to ensure that there is no excess ammonia in order to to avoid that
the NOx problem is just replaced by the ammonia problem.
N2 O (laughing gas, well-known as an anesthetic gas) is also a very critical com-
ponent that occurs as a by-product in some syntheses. It is not listed in TA Luft, but
it is regarded as one of the most critical greenhouse gases. Therefore, authorities do
usually not accept a defined N2 O emission. Nowadays, N2 O can be decomposed to
nitrogen and oxygen at comparably low temperatures (approx. 425–600 °C) with ap-
propriate catalysts [198, 199].
Chemical bonded Si is converted into solid SiO2 , which is simple quartz or sand.
From the environmental point of view, this is not critical at all, but the small particles
that are formed plug and erode the combustion unit. Therefore, special constructions
and operation conditions have to be chosen when Si occurs.
The thermodynamic calculation of combustion reactions is pretty easy. A neces-
sary combustion temperature is determined by the degradation temperature of the
pollutants. Usually, for a number of pollutants the manufacturers of combustion units
define the necessary residence time in the combustion chamber and the correspond-
ing temperature where they give a warranty to keep the TA Luft. This temperature is
usually not achieved by the combustion of the pollutants theirselves. Instead, natu-
ral gas must be injected and burnt up as well. The combustion temperature can be
calculated with an adiabatic energy balance, using the specific heat capacities and
the enthalpies of formation of the participating substances. The necessary amount of
natural gas is evaluated in an iteration procedure, checking whether the temperature
obtained equals the required one. The natural gas itself is often characterized by its
heating value, however, for process simulation it is advantageous to describe it by its
composition.
In general, we can distinguish between thermal and catalytic combustion. Ther-
mal combustion takes place at temperatures from 850–1200 °C. It is used for high
pollutant loads. The high temperatures make sure that complex molecules degrade
and the simple combustion products can be formed. Due to cost reasons, the heat of
the flue gas has to be utilized, e.g. by steam generation. Figure 13.10 shows a typical
Air
Supplementary
fuel gas Combustion
Combustion
chamber products
Burner
Waste gas Air
Fig. 13.10: Typical incinerator scheme [200].
thermal combustion unit. The supplemental fuel and air are mixed and fired in the
main burner. The combustion chamber is refractory-lined and the exhaust air is intro-
duced into the combustion chamber in the flame zone. Supplemental combustion air
can also be added, if necessary. The dimensioning of such a combustion unit can be
done according to the residence time of the exhaust air stream. Depending on the kind
of pollutants, the residence time should be between τ = 0.6–2 s. It has to be pointed
out that the physical volume flow at combustion conditions is decisive, not the stan-
dardized volume flow in Nm3 . The pollutant concentration are not relevant for the
dimensions of the unit, as for the combustion itself the pollutants are supplemented
by natural gas to keep the necessary combustion temperature anyway.
It is also important to note that combustion units have a limited capacity range.
The ratio between lower and upper bound of the capacity is approx. 1 : 5. At the lower
bound, the danger is that there is not enough turbulence to get an adequate mixing
of the pollutants with air. At the upper bound, the residence time in the combustion
chamber might not be long enough.
The scheme for catalytic combustion is shown in Figure 13.11. The polluted air en-
ters a heat exchanger, where it is preheated by the hot flue gas stream. The gases then
enter the catalyst bed. A noble metal catalyst is used to promote the desired oxida-
tion reactions at relatively low temperatures (250–400 °C) and at faster conversion
rates. Therefore, smaller units can often be specified, and less costly construction
materials can be used. The catalyst bed can be designed in the form of structured
or random packing, made of ceramic. Its volume is determined by the required de-
struction efficiency of the particular pollutants, the flowrate, and the properties of the
vapor stream. As a rule of thumb, 5000–20 000 Nm3 /h per m3 catalyst bed can be pro-
cessed. Catalytic combustion makes sense for low pollutant loads (< 10 g/Nm3 or 25 %
LEL (lower explosion limit, Chapter 14.3) [200]), as in these cases the heating value of
the pollutants is low and the consumption of natural gases to obtain the temperatures
required for thermal combustion would therefore be high. The oxygen concentration
in the waste gas should be < 2 mol % [200]. The advantage of catalytic combustion is
that smaller equipment and less costly materials can be used due to the lower tem-
Heat
exchanger
Clean
Burner air
Preheat Catalyst
section
Polluted
air
Fig. 13.11: Catalytic combustion scheme [200].
peratures. However, catalysts are in general sensitive, and the kind and amount of
pollutants should be clearly defined when a catalytic combustion is chosen. Phos-
phorus, heavy metals, and silicon are catalyst poisons, and occasional high pollutant
loads lead to high temperatures and, subsequently, deactivation. Another problem
called “the classical design mistake” can come up when ammonia and chlorinated
compounds are concurrently present in the exhaust air stream. In the combustion
unit, ammonium chloride will be formed, consisting of small particles that block the
catalyst after a remarkably short time.
Regenerative thermal oxidizers (RTO) are a third kind of combustion units (Fig-
ure 13.12), appropriate for very diluted exhaust air streams. In a typical RTO unit, there
are three ceramic beds for heat recovery. The contaminated gas enters one of the beds
(for simplicity: bed 1), and is effectively preheated by passing the hot ceramic bed
so that the burner itself does only need to cover the last part of the preheating of the
exhaust air. After having been incinerated, the clean exhaust gas stream exits the com-
Burner
Residence chamber
Bed 1 Bed 2 Bed 3
Stack
Contaminated
air
Purge air
Fig. 13.12: Typical RTO unit [200].

bustion chamber through bed 3. Its sensible heat is transferred to the bed, where it can
be used in the next cycle. Part of the clean air gas is led to bed 2, which was the pre-
heating bed during the last cycle, to purge it; in this way, the clean air is contaminated
again and therefore led back to bed 1.
At the completion of each cycle, the task of the beds is changed by switching the
valves at the inlet and outlet lines. Cooling below the dew point in the heat recovery
section should be avoided because of corrosion. The control system which switches
between the beds is comparably complex.
The advantages of combustion processes are the low operation costs due to steam
production (see below). No trials are necessary to predict the outlet streams of the
combustion. The main disadvantages are the high investment costs and the complex
safety concept that is necessary as the flame must be regarded as a permanent ignition
source. The availability is high for the thermal combustion. For the catalytic combus-
tion, it is limited by the lifetime of the catalyst.
After passing a typical combustion unit (Figure 13.13), there are a few other neces-
sary steps before the exhaust air is released into the environment. After “denoxation”
(see above), the heat of the stream has to be utilized due to economic reasons. In the
steam generator, the stream is cooled down to approx. 270 °C. In many cases, the ben-
Waste air
Scrubber
Dilution
air
Natural
gas
Stack Liquid
fuel or
waste
70 °C
Purge 1200 °C
Quench Combustor
NaOH Steam Approx. 270 °C
Makeup water
Convective part
Urea DeNox
chamber
Natural gas Steam generator

1000 °C 1200°C
Radiation part
Feed Water
Fig. 13.13: Thermal combustion scheme.

efit of steam production can fully compensate for the expenses for the natural gas.
Behind the steam generator, the steam is still too hot for the scrubber. It is first cooled
down by direct injection of water in a so-called quench. As part of the water evapo-
rates, the temperature can be lowered to approx. 70 °C. In the scrubber, acid gases like
HCl, HBr, or SO2 are finally chemically absorbed, usually with caustic soda solution
due to chlorine or bromine formation according to the Deacon reaction.
13.4.3 Absorption
Absorption is another exhaust air purification process where the performance and
the design can in most cases be evaluated without experiments. Possible absorptive
agents can be found theoretically as well. The demands for a suitable absorbent are
high capacity, low volatility, viscosity, corrosivity and toxicity, high thermal and chem-
ical stability, and a high flash point. The recycling of valuable solvents from the ex-
haust air is possible, especially for exhaust air streams that contain only one compo-
nent as a pollutant. In these cases, the relatively high operation costs can be signif-
icantly reduced. Nevertheless, there are not many cases known. Absorption is also a
promising alternative if the loaded absorbent can be sold. Aqueous ammonia solu-
tion is one of the few examples. If water is used as an absorbent, it might happen that
it is useful to give the loaded water directly to the biological waste water treatment.
However, water is usually a bad solvent if organic substances have to be absorbed. Of-
ten, organic solvents are taken, for example glycol ethers for chlorinated compounds
[201]. Other options for the removal of organic pollutants are heavy alkanes or even
biodiesel (fatty acid methyl esters) [202], which can be used as fuel afterwards. One
should avoid stupid combinations like taking water as an absorbent for toluene (“au-
thority scrubber”).
Otherwise, a desorption step to remove the load from the absorptive agent can-
not be avoided. Figure 13.14 shows a typical absorption/desorption unit, where the
absorbent is cooled down for the absorption step and heated up for the desorption
step. The desorbed gas can usually be condensed and given to a liquid waste inciner-
ation, which should be available at any chemical site. The desorber column can also
be designed as a conventional distillation with condenser and reflux, if the losses of
the absorbent are too high due to its volatility. It is worth mentioning that at least
the absorption column should be simulated with a rate-based calculation (Chapter 5),
as in most cases the mass transfer resistance in the vapor phase is decisive for the
final design of the column. As absorption equipment, in addition to packed and tray
columns also spray towers, bubble columns, venturi scrubbers, and many other types
come into consideration.
Absorption has severe disadvantages for highly volatile components and when
hydrophobic and hydrophilic substances must be absorbed simultaneously, which is
Cleaned air Pollutants
Absorber Desorber
Exhaust air
Loaded Recycled Fig. 13.14: Scheme of an absorption/

absorptive agent absorptive agent desorption unit.
often hardly possible with one absorbent. The investment costs can be considerable if
high-quality construction materials must be used to avoid corrosion.
13.4.4 Biological exhaust air treatment
Biological processes for exhaust air treatment have become more and more important.
They are appropriate for huge exhaust air streams with low pollutant concentration.
The pollutants must be soluble in water and biodegradable. The exhaust air should
have a temperature in the range 5–60 °C and must not contain toxic substances. If
these requirements are fulfilled, biological processes are in general the best processes
due to their low investment and operation costs. However, their effectivity cannot
be predicted. If a vendor has no references for a defined exhaust air problem, long
term experiments, usually several months, are necessary to prove that targets can be
met. Biological degradation is performed by microorganisms like bacteria or fungi
[195]. All of these microorganisms are surrounded by a water film which they need
for their metabolism. Therefore, it is necessary for the pollutants to be soluble in wa-
ter so that they can get in contact with the microorganisms. Furthermore, nutritients
and trace elements (nitrogen, potassium, phosphorus) must be provided for the mi-
croorganisms. The degradation itself yields carbon dioxide and water as products.
Other elements like chlorine, nitrogen or sulfur will be transformed into inorganic
compounds (HCl, H2 SO4 , nitrates) which may enrich in the water, where they have
a detrimental effect. Increasing temperatures accelerate the degradation process but
decrease the solubility of the gases in water. Thus, an optimum temperature has to
be evaluated experimentally. The mass transfer from the vapor phase into the liquid
and finally to the microorganisms is decisive for the effectiveness of the process. There-
fore, contact areas between the water and the exhaust air which are as large as possible
should be provided. A severe disadvantage of biological processes is that the microor-
ganisms of a specific plant can specialize in degrading the most common substances
in the exhaust air. Components that occur only occasionally can then be ignored, and
they remain dissolved in the water.
There are several process options for biological exhaust air treatment. In biofilters,
the microorganisms are located on a solid filter material, which is sprinkled with wa-
ter. The pollutants are absorbed by the liquid as well as adsorbed by the filter material.
As filters, compost materials, turf, brush-wood, bark, wood, coconut fibers, foams,
and other porous materials can be used. Inorganic nutrients (nitrogen, phosphorus
etc.) can be delivered by the filter material itself or supplied with the sprinkling water.
For the design, it has to be regarded that the exhaust air coming out of the biofilter
is always saturated with water due to the intensive contact with the filter material.
Therefore, biofilters are prone to dry out, which leads to worse conditions for the mi-
croorganisms. Therefore, the humidity has to be controlled carefully. The exhaust air
is often humidified before it is led in the biofilter. As a rule of thumb, for the calcula-
tion of the volume of the filter layer it can be assumed that the exhaust air load should
be in the range 100–250 Nm3 /h per m3 filter material. The degradation capacity for
pollutants can be 10–100 g/h per m3 filter material [195]. This leads to considerable
dimensions for biofilters.
Bioscrubbers are scrubbers where liquid from an activated sludge tank is used as
absorbent. The packing is inert. For the evaluation of the dimensions, bioscrubbers
can be treated like normal scrubbers. They are considerably smaller than biofilters.
Biotrickling filters try to combine the principles of bioscrubbers and biofilters. The mi-
croorganisms settle on the packing so that the absorbed pollutants are degraded right
on the spot. New developments in biological exhaust air treatment have the target
to reduce the dimensions, especially for biofilters. Bioscrubbers could also be imple-
mented as tray columns, where the concentration of microorganisms could be much
higher. It is estimated that a degradation rate of approx. 1300 g/(m2 h) can be real-
ized. A second hydrophobic solvent in addition to water could form a second liquid
phase in a scrubber, which could absorb hydrophobic pollutants from the exhaust air.
It could be recycled in the activated sludge tank. Finally, membranes could be used
where the microorganisms can settle on. A few hundred bioprocesses for exhaust air
treatment are operated in Germany. Most of them are biofilters, used for agriculture,
fish industry, and sewage plants. The application makes sense for pollutant loads of
1000–1500 mg/m3 of organic carbon [195].
13.4.5 Exhaust air treatment with membranes
Membrane applications also have potential for exhaust air cleaning, especially in com-
bination with adsorption. The advantages of membranes are the simple, modular con-
struction and the low space demand. On the other hand, the predictability of their
performance is very low, and even if references are available, there is still doubt about
their mechanical, thermal, and chemical stability as well as about their sensitivity
against fouling. Furthermore, their properties vary during operation, as many mem-
brane materials swell when they are exposed to the pollutants.
There are relatively few references for membrane separations. They refer to simple
cases like removal of toluene or a hydrocarbon from exhaust air. Figure 13.15 shows an
example.
Retentate
purified
Exhaust air exhaust air
High pressure
Low pressure
Membrane
Permeate
Condensate
Fig. 13.15: Process scheme for gas permeation.
As the partial pressure difference is the driving force for the flow through the mem-
brane, a compressor is used on the pressure side and a vacuum pump on the suction
side. The membrane only achieves an enrichment of the pollutants on the suction side.
The pollutants in the permeate are then partially removed by condensation, whereas
the rest is fed back to the pressure side. In comparison to the simplicity of the problem,
this is a quite complex and expensive process. After all, the pollutants can be recycled
if it seems useful.
13.4.6 Adsorption processes
Like membrane processes, adsorption can also be an option for exhaust air treatment.
The foundations and the terms have already been explained in Chapter 7.2.
There are several kinds of adsorbers that differ in the way the adsorbent is treated.
It can be fixed in a packed bed or in a moving bed, or it can be implemented as a
fluidized bed. The most popular way is the fixed bed because of its simplicity and the
low abrasion of the particles. However, the big disadvantage is that the adsorption
process is transient. For a continuous process, a second apparatus is necessary to take
over the task when the first is being regenerated and vice versa (Chapter 7.2).
Adsorption processes are favorable if very low pollutant contents for the exhaust
air are aspired to. Compared to absorption, the investment and operation costs for ad-
sorption are considerably higher, up to a factor up to 3 [195]. However, it is often used
for relatively small exhaust air streams, as no major costs for recycling of liquids oc-
cur. Another advantage of adsorption is the option of recycling the pollutants, which
often compensates the disadvantages easily [195]. A serious disadvantage is that an
extended safety concept is necessary due to the danger of fire. Activated carbon with
huge surface areas, the presence of oxygen, and the release of the heat of adsorption
provide good conditions for fire. In fact, it has been found in many cases that smolder-
ing fires were active inside the adsorber bed which were not detected by the operator
team. Numerous examples are known for the so-called “Monday fires” [142]. On Fri-
day, machines and vessels are often cleaned with large amounts of organic solvents,
which remain in the adsorber for the weekend and cause the fire after the new startup
on Monday.
13.5 Waste water treatment
Water is one of the most often used substances in chemical industry. It is used as
solvent, raw material, medium for chemical reactions and as washing agent for prod-
ucts, gases, and equipment. For this it is loaded with substances and particles. Before
returning it to environment, it has to be cleaned according to the governing rules
(e.g. Germany: Wasserhaushaltsgesetz). Any discharge of waste water needs a per-
mit, which is subject to strict limiting values. A permit is given only if the water is
cleaned according to the current state of the art.
For the treatment of waste water streams, one can distinguish between measures
to remove solids and measures to remove dissolved impurities.
Solids can be removed by
– Sedimentation:
The solids must have a larger density than water. The density difference and the
particle size must be sufficiently high.
– Flotation:
The solids must have a lower density than water, so that it moves up to the surface.
If the density difference is not large enough, auxiliary substances can be used. For
example, gas can be introduced into the water. The bubbles will be attached to the
particles, which lowers their apparent density.
– Filtration:
The waste water can be filtered over flint, sand or industrial filters, where the large
particles are caught as they cannot pass meshes which are smaller than the parti-
cles theirselves. The smaller the particles, the smaller the meshes must be. Beyond
conventional filters, membranes are used in the different applications microfiltra-
tion, ultrafiltration, and nanofiltration (Chapter 7.1).
For dissolved impurities, the typical cleaning of waste water is different from other
separation tasks, as it is in most cases not well defined, i.e. the loads vary and the
polluting components are often not known. Therefore, a waste water treatment pro-
cess cannot be predicted but must be experimentally demonstrated in a piloting unit.
Often, the particular vendors have miniplants where a test amount of genuine waste
water can be processed to check the performance.
The load of a waste water is characterized by the TOC (total organic carbon), the
COD (chemical oxygen demand) and BOD (biological oxygen demand) values. The TOC
value is the concentration of carbon atoms of organic molecules in the waste water.
It can be measured with good accuracy by determining the carbon dioxide after ox-
idation. The COD value is less accurate. It refers to the amount of oxygen necessary
to convert the organic substances into CO2 , H2 O, and NH3 . The biological oxygen
demand is more complicated and can only be determined experimentally. It is mea-
sured how much oxygen is consumed by microorganisms in contact with the waste
water during a five-day-period (often termed as BOD5 ) at t = 20 °C [7]. It is a very use-
ful quantity, but it takes five days for determination. The BOD is always lower than
the COD, as the microorganisms often use only parts of the molecules for combus-
tion, while the rest is used for growth. The ratio between BOD and COD is between
BOD/COD = 0.05–0.8, i.e. it is completely unpredictable. If nothing better is known,
one can use BOD/COD ≈ 0.35.
Example
A waste water stream contains 500 wt. ppm methyl tert-butyl ether (MTBE). Determine TOC, COD, and
BOD.
Solution
The chemical formula of MTBE is C5 H12 O, with the molecular weight M = 88.148 g/mol. With MC =
12.011 g/mol as the molecular weight of carbon, the TOC can be determined to be
5 ⋅ 12.011
TOC = 500 wt. ppm ⋅ = 341 wt. ppm .
88.148
The oxidation reaction of MTBE is given by
C5 H12 O + 7.5 O2 󳨀→ 5 CO2 + 6 H2 O ,
and therefore one gets using MO = 15.9998 g/mol as the molecular weight of oxygen
7.5 ⋅ 2 ⋅ 15.9998
COD = 500 wt. ppm ⋅ = 1361wt. ppm .
88.148
Without experimental information, the BOD can only be estimated to be
BOD = 0.35 COD = 476 wt. ppm .

For dissolved substances, there are a number of different processes for waste water
cleaning that are often used in combination:
– Evaporation:
Often, the pollutants are heavy boiling substances which cannot be vaporized.
In this case, the waste water can be concentrated by evaporation. The remaining
residue can be sent to incineration or to disposal if possible. Of course, waste wa-
ter evaporation is very energy-intensive. It is more or less obligatory to use at least
one of the heat integration options described in Chapter 3.2, i.e. multieffect evap-
oration or vapor recompression. The statement that all the pollutants might be
heavy boiling substances is usually weak. In most cases, the condensate is not
pure water but contains components which are light ends or form low-boiling
azeotropes with water. Then, evaporation must be supplemented by a condensate
polishing measure, either reverse osmosis, chemical destruction or adsorption
(see below). The target is always to return as much non-contaminated water as
possible back to environment.
– Reverse osmosis:
Reverse osmosis can be used for the cleaning of waste water as stand-alone or
as a supplementary measure. The method is restricted to low-concentrated pollu-
tants, as the osmotic pressures should be limited to 40–100 bar. Membranes are
used which do not work as a filter but by means of solubility and diffusion (Chap-
ter 7.1). Reverse osmosis can remove large amounts of pure water which can be
returned to environment. The biggest problem is the durability of the membrane
which is necessary to test before an application takes place. A regular exchange
of the membrane is often necessary but not too expensive.
– Adsorption:
Pollutants can also be removed by adsorption. Due to the wide variety of possible
pollutants, an adsorbent like activated carbon is one of the first choices, as it re-
moves organic components quite reliably. As described in Chapter 7.2, a twin plant
(Figure 7.7) is necessary to ensure a continuous operation. As activated carbon is
quite inexpensive, the loaded adsorbent can be sent to incineration or be regen-
erated by specialized service providers. The activated carbon is not removed as a
powder, instead, a whole filter unit is removed and replaced, which is quite fast
and clean at the site and easy [203].
– Chemical destruction:
An interesting procedure for the reduction of TOC and COD are the advanced oxi-
dation processes (AOP). A photochemical reaction due to absorption of ultraviolet
radiation can activate the pollutant molecules. The activated state can cause a fur-
ther reaction to products which are easier biodegradable. Additionally, oxidants
like ozone or hydrogen peroxide [204] can be split into oxidizing radicals, which
rapidly react with the pollutants and decompose them into CO2 and H2 O, as long
as no other elements than C, H, and O are involved. Ozone is taken for processes
where the pollutant concentration is low, as the solubility of ozone in water is bad.
If low boiling pollutants are present in the waste water, it happens that they are
stripped into the offgas. The activation of the molecules can also be increased by
ultrasound. In case the oxidation has stopped at some intermediate products, the
AOP can be supported by a biological treatment.
Maintenance of these systems is hardly necessary, and another great advantage
is that the waste water stream is not split into two streams, where one of them
contains the pollutants and has to be further processed. Even highly concentrated
solutions up to 250 g/l COD have been successfully treated [205]. The disadvantage
is that an infrastructure for the oxidant must be provided.
– Biological waste water treatment:
The waste water treatment by microorganisms is probably the most often used
final treatment for waste water. We distinguish between aerobic and anaerobic
processes. Aerobic processes need aeration to ensure the continuous availabil-
ity of oxygen. The organic substances are decomposed to CO2 and H2 O, however,
it must be checked how other elements than C, H, and O are converted. Anaer-
obic processes produce even valuable substances like methane (“biogas”). It is
necessary to finally separate the biomass from the cleaned water, usually by sed-
imentation. The biomass can be incinerated or dried and sold as fertilizer.
– Waste water incineration:
Waste water can also be incinerated, which is reliable, but probably the most un-
satisfactory process, as the often large amounts of water do not have a heating
value but must be evaporated in the incinerator. Therefore, the waste water is first
concentrated to reduce the amount of water.
– Pressure hydrolysis:
With this option, water is kept under pressure for some time at high temperatures
(200–250 °C), where hopefully the pollutants decompose to substances which are
easier to handle.
14 Process safety
In the chemical industry there is certainly a hazardous potential, as combustible and
poisonous substances are involved. In the past, a number of serious accidents have
happened, some of which are listed below.
– Ludwigshafen-Oppau (1921):
An explosion in a fertilizer storage of the BASF company killed 561 people. More
than 2000 got hurt. Even in Heidelberg, 30 km away, roofs were untiled. The crater
in Oppau is 125 m long, 90 m wide and 19 m deep (Figure 14.1). Because of the ex-
tent of the catastrophe its reason could not be reconstructed exactly. The main
components involved were ammonium nitrate, a well-known explosive agent, and
ammonium sulfate. In the scheduled mixing ratio, the fertilizer was not explo-
sive. Probably, a demixing had taken place. The attempt to loosen up the densely
packed bed by disrupting it with dynamite led to a booster detonation of the am-
monium nitrate, which caused a knock-on effect on the entire storage with 4500 t
of fertilizer.
Fig. 14.1: The Oppau crater after the explosion in 1921.

© BASF Corporate History, Ludwigshafen/Rhein.
328 | 14 Process safety
– Texas City (1947):

Like in Oppau, ammonium nitrate was also involved here. A cargo ship containing
2500 t of ammonium nitrate caught fire and blew up. As a consequence, the Mon-
santo site nearby and several oil refineries also caught fire, and a large number
of explosions took place. It took several days to get the situation under control.
There were more than 600 casualties and 3000 injured.
– Ludwigshafen (1948):
On a hot summer day a tank wagon with 30 t of dimethylether detonated on the
BASF site in Ludwigshafen with great violence. It was the worst explosion in Ger-
many since the Second World War. There were 207 casualties, almost 4000 in-
jured, and more than 7000 damaged houses in Ludwigshafen and Mannheim.
– Bitterfeld (1968):
In order to exchange a sealing, 4 t of vinyl chloride were relieved from an auto-
clave. A violent detonation took place. 42 people lost their lives, more than 200
were injured. A large part of the site was destroyed.
This accident appears especially absurd from today’s point of view. Vinyl chloride
is highly carcinogenic and poisonous, today, according to TA Luft [191] only an
extremely low emission of 2.5 g/h or a concentration of 1 mg/Nm3 is allowed in
Germany. Even the release of a noncritical substance into environment would be
unthinkable, and only a controlled discharge into an exhaust air system (usually
an incineration) would be possible. A plant without such an equipment would not
be licensed.
– Flixborough (1974):
One reactor in a row of five had to be bridged because of a failure. The connecting
line was not strong enough for the process conditions and broke. In 50 s, approx.
40 t of cyclohexane vapor effused into environment. An ignition followed, and
the whole site was destroyed. There were 28 casualties and 88 injured. The adja-
cent storage tank containing 1600 t of combustible substances caught fire as well.
Even after three days explosions were still taking place. The number of casual-
ties would have been considerably higher if the explosion had happened during
normal working hours and not on a Saturday [206].
For the bridging of the reactor, which is at least a manipulation of the process
operating at t = 150 °C and p = 10 bar, no design work was performed. The con-
struction drawing had been made with chalk on the floor. There was no static
calculation, and valves, which would have made it possible to isolate the reac-
tors against each other, were not provided.
– Seveso (1976):
In an autoclave producing 2,4,5-trichlorophenol the agitator was switched off by
mistake, after the reaction and the shift were finished. The heat removal from the
reactor became much worse, and the product was involved in further reactions.
Because of the missing heat removal, the reactions were accelerated. One of the
follow-up products was dioxine (2,3,7,8-tetrachlorodibenzodioxine), on a kg-scale.
14 Process safety | 329
Finally, the safety valve of the reactor actuated. There was neither a collecting
vessel nor a defined line to a flare. The safety valve opened for 30 min, and the
vapor went straight into the environment. It took hours before an experienced
crew arrived with the next shift. The reactor could be shut down. 18 km2 were
contaminated. Plants wilted, and more than 3000 carcasses were found. Approx-
imately 200 people suffered from chlorine acne. The number of casualties as a
consequence of the accident is not known. The cancer rate rose significantly. It
took eight years before all decontamination measures were finished.
– Bhopal (1984):
In Bhopal (Central India) the Union Carbide company manufactured carbaryl, an
insecticide, via the intermediate methyl isocyanate, which is an extremely poi-
sonous substance. Accidentally, water intruded into a tank filled with 40 t methyl
isocyanate and caused a chemical reaction. Carbon dioxide was formed and built
up pressure. The methyl isocyanate, which is quite volatile anyway (normal boil-
ing point: 39 °C) was evaporated by the heat of reaction. Within two hours, the
whole content of the tank was released through the safety valve. More than 2000
people were killed. Probably about 200 000 people were hurt. A decontamination
of the area around the plant has to date not been performed.
The safety devices provided had not worked at all. The cooling system of the stor-
age tank and the gas flare had been switched off months previously, an emergency
scrubber was not ready to operate, and the tanks were overfilled. The personnel
staff had been reduced and was not sufficiently trained. The alarm system was
switched off in order not to disturb anyone, and no emergency plan existed; many
people died when trying to escape directly through the poisonous cloud.
This large list of failures and unlucky circumstances is supplemented by the bad
process concept [206]. While methyl isocyanate as a poisonous substance was
produced continuously, its further processing took place batchwise so that large
amounts of this substance had to be stored. A completely continuous process
would have drastically reduced the amount of the methyl isocyanate inventory
in the plant.
– Toulouse (2001):
Exactly 80 years after the Oppau accident a violent explosion happened at the site
of the TotalFinaElf in Toulouse, also caused by ammonium nitrate. There were
31 casualties and more than 1000 injured. The cause was never clarified.
Accidents with much less damage can also receive a lot of publicity. One example is the
so-called Carnival Monday accident in 1993 at the Griesheim site of the former Hoechst
AG. Again, an autoclave was involved. The reactants, methanolic caustic soda and
o-nitrochlorobenzene, have a broad miscibility gap. Therefore, the methanolic caus-
tic soda was slowly added to the organic phase, while sufficient mixing and thereby
the reaction should be achieved by an agitator. The staff did not notice that the agitator
was switched off, and a large amount of the reaction mixture got into the reactor, form-
ing two liquid phases. Because the reaction did not take place, the usual temperature
increase was missing. Therefore, the reactor was heated up, contrary to the manual.
Then, the error was noticed, and the stirrer was switched on, with the large inven-
tory of reactants at high temperature. The reaction started immediately, the cooling
system could not remove the large heat of reaction, and the safety valves opened.
10 t of the product o-nitroanisol were relieved to the environment. Due to the cold
weather, the product condensed as an aerosol. A greasy yellow layer covered large
parts of the neighboring district [206]. It is clear that the reactor was not sufficiently
protected against maloperation. The dosing of the caustic soda while the agitator is
not running and the external heating of the reactor should have been prevented by
interlocks. Although no one was killed or injured, the accident was often cited in an
unjustified way in connection with Bhopal or Seveso. The loss of image due to the bad
information management of the Hoechst AG was very serious.
In the next section, we shall outline that many improvements concerning the
safety have been introduced. Nevertheless, severe accidents are definitely not just a
thing of the past. While this book was being written, two major accidents took place.
On April 7th, 2013 a serious explosion took place in West, close to Waco/Texas.
There has still been no official statement about the reason. It is clear that a fire broke
out without sufficient fire protection, where 240 t of ammonium nitrate were stored.
An explosion took place. 14 people were killed, among them 12 firemen. More than 250
people were injured [207].
On August 12th, 2015 a fire broke out in the bulk storage of the harbor of Tianjin/
China. After 45 min, two detonations took place within one minute. There were at least
165 casualties, and approx. 800 people were injured. In the harbor, about 3000 t of
hazardous substances were stored, among them sodium cyanide, calcium carbide,
and, again, ammonium nitrate. The current theory for the reason is that acetylene
was formed due to contact of the calcium carbide with the fire fighting water [208].
14.1 HAZOP procedure
The safety of its process facilities is the most important target of a chemical company.
Although it is not possible to avoid accidents completely, the chemical industry has
learned a lot. Most hazards happen because of flaws in design, material, or due to hu-
man error, whereby the latter reason is considered to be the most important one. In a
chemical plant, there is a large potential for human error during design, procurement,
construction, and operation. It is desirable that these errors are anticipated before the
plant is commissioned.
There are a number of procedures for the safety analysis [209]. The most estab-
lished one in the chemical industry is the so-called HAZOP analysis (HAZard and
OPerability). It was developed in the 1970s after the Seveso accident. It is important to
note that HAZOP looks for incidents with the potential for severe impacts. The minor
ones (“slips, trips, and falls”) are the subject of the company’s general safety require-
ments. Not only for new plants but also for changes in the process supposed to be
minor ones a HAZOP session is recommended, as the implications are often underes-
timated.
Basically, HAZOP is a communication technique. Information is presented, dis-
cussed, analyzed, and recorded [209]. The safety aspects are systematically identified,
to check the measures taken to prevent major accidents. The HAZOP procedure is quite
time-consuming and requires a number of skills of the participants. It is recommended
that the team consists of approx. 5 people. More participants dilute the effectiveness,
as there are too many communication routes between the people. All of them should
be able to communicate fluently in the language applied (usually English), otherwise
too much effort is necessary to keep everyone up to date. The participating team mem-
bers should be familiar with the chemical process under examination. The partici-
pants can be members of different organizations, e.g. from the engineering company,
from the future plant owner, or from a consultant company. They should cover the
following items or, respectively, areas:
– HAZOP leader:
The HAZOP is guided by a HAZOP leader, who takes care that the meeting keeps
its target focus. The HAZOP leader is the person who is essential for the success
of the team. He leads the team through the procedure and brings out the concerns
of the process. He announces the key words to be discussed and, in case there is
no minute taker, notes down the deviations, causes, and countermeasures. It is
not intended that he takes part in the discussion, but it is also not forbidden. One
of his most important tasks is to keep the discussion under control, which some-
times turns out to be an engineering review or a personal dispute between two
participants. For HAZOP leaders, a certain authoritative personality is necessary.
Preferably, the HAZOP leader had not been involved in the design of the process,
so that he is not biased. He also takes care that the necessary documents used in
the HAZOP are available. The team leader must be aware that the attendance of
the team members in the sessions is the most costly part of the HAZOP process.
Therefore, it is up to him to avoid unnecessary discussions and meet the schedule.
– Process Engineer, responsible for the unity in the project:
The people who have designed the process know it very well and can explain how
the particular measures interact. They should be able to give first answers to the
various key word items.
– Operations representative:
The operations representative is more experienced in operating the plant and fo-
cuses on items which are caused by operational errors rather than by the design.
– Safety expert:
The safety expert knows the impact of the items and is familiar with the rules and
standards.
– Instrumentation & control representative:

The representative of instrumentation & control knows the cause & effect matrix
and how the control cycles work. His special expertise ensures that a number of
rules are maintained which other participants are not familiar with; e.g. an indi-
cator involved in a control loop must not be used as a safeguard as long as the
control loop itself can be the cause of the deviation.
– Consultant:
The consultant is often someone who is familiar with similar processes but not
with the one to be examined. This is useful, as consultants have the advantage
of being unbiased and independent. They did not participate in the project, and
hopefully, they will ask questions that the others who have become blind to the
shortcomings in the project will not do.
Other additional people able to give important input are maintenance representatives
and material specialists. The HAZOP team should have a certain experience. If the
majority of the team has never participated in a HAZOP, the HAZOP leader will be
completely preoccupied with instructing the team members rather than having them
contribute to the review. The team members should be encouraged to ask “stupid”
questions [209].
For the whole plant, line by line and for each piece of equipment it is examined
what the consequences of deviations might be and which additional countermeasures
should be applied. All necessary documents should be available, i.e. (amongst oth-
ers) PIDs (including the PIDs for vendor packages) and PFDs, material balances, plot
plans, cause & effect charts, interlock descriptions, fire protection measures, list of
safety valves/rupture discs with relief case descriptions, data sheets of equipment, in-
strument and control valves, ambient data, utility list, and the properties of the main
components.
First, the plant should be divided into parts, the so-called nodes, which have a
well-defined objective (e.g. separation, reaction, heating/cooling, pressure increase).
There is a list of guidewords (temperature, pressure, flow, etc.) which covers at least a
large number of possible deviations systematically.
A number of computer programs is available which can support the procedure.
For each node, every guideword is considered with the following work flow [209]:
– Definition of the node.
– Short process description, usually given by the process engineer.
– Selection of the process parameter and assignment of the deviations, one by one.
Go through all the streams of the node with the process parameter selected before
changing it. The process parameters and the deviations are:
– Flow (no flow, less flow, more flow, reverse flow)
– Temperature (low temperature, high temperature). Double-check the design
temperatures of the particular pieces of equipment with respect to the scenar-
ios.
– Pressure (low pressure, high pressure). Double-check the design pressures

and the pressure relief cases. Ruptures or leakages can be the reason for de-
viations.
– Level (high level, low level)
– Identification of the causes and hazards of the deviation.
– Identification of the consequences of the hazard without regarding the safe-
guards.
– Specification of the appropriate safeguards and recommendations to control the
hazards.
– List of recommendations in the order of their priority.
– Ensure that the actions proposed are implemented and documented and that for
each action a responsible person is assigned.
Certainly, the HAZOP is no guarantee that nothing can happen in a process. The tar-
get which is achievable is to lower the probability of accidents, especially of serious
ones, as many scenarios are reflected and people become sensitive to possible con-
sequences. Risks can be categorized into equipment failures (ruptures, control valve
failures, etc.), operational errors (wrong valve state), external events (corrosion, fire),
and product deviations (e.g. off-spec). The assessment of the risk takes place based
on common sense, personal experience, knowledge, and intuition of the participat-
ing people. Non-reasonable risks should be ruled out. A rupture of a line has a real
possibility of occurring, whereas a meteorite striking the facility is not really proba-
ble, and the consequences can neither be avoided nor mitigated by better engineering.
Similarly, the “double jeopardy” case should be in mind, meaning that two indepen-
dent events at a time do not need to be considered, e.g. a tube rupture as mentioned
above and a simultaneous malfunction of a control valve. The probability of the si-
multaneous occurrence of two independent events is negligibly low. However, two
simultaneous events are not necessarily independent; e.g. a failure of the reflux pump
and a loss of cooling in the condenser of a distillation column. In this case, the fail-
ure of the reflux pump might cause accumulation of liquid in the condenser, with the
consequence of the cooling loss. In principle, there is one first error, and the second
one is the consequence of the first.
The notes should clearly indicate the exact naming of the equipment and the in-
strumentation devices. Otherwise, the review of the report will take time and might be
erroneous. Minor spelling errors are not important and can easily be corrected latter.
The costs of the countermeasures should not be taken into account or discussed
during the HAZOP session. Of course, for the final decision the practicability and the
costs must be examined. A ranking of countermeasures is useful for the accelera-
tion of the decision process afterwards. The criteria are life safety, protection of the
environment, protection of the equipment, and, finally, continuation of production.
Countermeasures for high risks tend to be more costly and complex than those for
low risks and should be considered first. The proposed recommendations must be
forwarded to the acknowledged experts for an evaluation.
Currently, there is a trend to encounter safety issues by means of process con-
trol systems, meaning that instruments are used to prevent or mitigate a hazardous
situation. In general, electronic components behave differently in comparison with
mechanical components. Regarding the probability of error, the error rate of electronic
components is high at the beginning of the lifetime in operation. With a bit of dark hu-
mor, this is called “infant mortality”. After a short period of time it drops down and
remains constant on a low level. At the end of the lifetime it rises again strongly. In
contrast, mechanical devices also have a high error rate at the beginning, but after a
period with a low error rate it rises linearly with time [210]. The measuring and con-
trol devices which have a safety-relevant function are classified by the so-called SIL
(“safety integrity level”) analysis, where it is rated which certificates for the proba-
bility of failure for the various devices are necessary, according to the risk associated
when the device function fails. There are four SIL numbers, the higher the SIL num-
ber, the lower is the risk if the device fails. Table 14.1 gives the average probabilities
for failure on demand in the low demand mode, which is defined that the frequency
of operations in a safety relevant function is not greater than once per year.
Table 14.1: Probabilities for failure on demand for the various SIL classes.
Safety Integrity Level (SIL) Probability of failure on demand
4 10−5 – 10−4
3 10−4 – 10−3
2 10−3 – 10−2
1 10−2 – 10−1
The greater the consequences and the higher the probability of occurrence, the higher
must be the SIL number. Figure 14.2 is based on the international standard IEC/EN
61508 and connects these “risk parameters” and the SIL class.
Example
A shell-and-tube heat exchanger is operated continuously with two fluids which can react with each
other in case of a tube rupture. The reaction would be exothermic. If the reaction takes place quantita-
tively, it might happen that the gaskets of the apparatus fail. If people are working in the vicinity, they
might be seriously injured. With a surveillance of all pressures and temperatures involved, even small
deviations from the set point are indicated. If two signals are out of tolerance, an interlock closes a
shut-off valve to make sure that the feed line is closed to limit the extent of the reaction. Which SIL
classification is necessary?
Risk parameters
Consequence/severity SIL calculation

C=1 minor injury or damage
C=2 serious injury or one death,
temporary serious damage
C=3 several deaths, long term damage C=1
C=4 many dead, catastrophic effects
no special safety requirements

Frequency/exposure time
F=1 rare to quite often
F=2 frequent to continuous
C>1
Possibility of avoidance
P=1 avoidance possible
P=2 unavoidable, scarcely possible SIL = C + F + P + W – 6
Probability of occurrence SIL ≤ 0 → no special safety requirements

W=1 very low, rarely SIL > 4 → process concept not reasonable
W=2 low
W=3 high, frequent
Fig. 14.2: Relationship between the risk parameters and the SIL number required.
Solution
According to Figure 14.2, the case can be categorized as follows:

C =2 serious injury and temporary serious damage;
F =2 continuous operation and risk exposure;
P =2 avoidance not possible;
W=1 low probability.
The chosen SIL classification is
SIL = C + F + P + W − 6 = 2 + 2 + 2 + 1 − 6 = 1 ,
with a probability of failure on demand of 0.01–0.1 (Table 14.1).

14.2 Pressure relief
Safety valve design is like Christmas:

You know that it will happen soon, but when it is time you are not prepared.
(Anke Schneider)
14.2.1 Introduction
Chemical plants often work at pressures far from the ambient one, and corrosive, toxic,
flammable, and even explosive substances are handled. These pressures must always
remain under control in closed systems which are constructed in a way that they can
certainly withstand the design pressure and the design temperature. However, the
design values are not infinite, and undesired scenarios can happen where they are
exceeded. The plant design must make sure that these scenarios do not have an im-
pact on the safety of the plant [211]. As long as the control system works, the plant
is usually operated in a safe way. If the control system also fails, a common method
to protect the process equipment and keep it in a safe state is the emergency pres-
sure relief or blowdown, which is a topic where an understanding of the process and
knowledge of thermodynamics can contribute to avoid accidents and damage to the
plant. The pressure relief device is the last line of defense and must be capable to
actuate at all times and under all circumstances. It removes the potentially dangerous
contents of the process equipment and transfers them to a safe and lower-pressure
location; i.e. the environment for nonhazardous contents or, in the usual case, a flare
system. Furthermore, it decreases the pressures exerted on the walls of the equipment
and possibly prevents an escalation due to an explosion or a major relief of toxic sub-
stances [212]. Nevertheless, the pressure relief itself is often a hazardous operation.
During the depressurization, the fluid expands and low temperatures can rapidly be
generated, possibly causing brittle-fracture of the vessel walls (Chapter 11). In distil-
lation columns, the flow through the packing or the trays can be much higher than
designed for normal operation, and the equipment can be seriously damaged during
a pressure relief (p. 355).
There is a large to-do-list for the engineer when a pressure relief device must be
designed [212]:
– fix the rate of relief from the piece of equipment to ensure that the design pressure¹
is not exceeded;
– determine the restriction orifice to make sure that the corresponding equipment
is protected;
– make sure that the relief flow can be transferred to the low-pressure destination;
1 more exact: relief pressure; see below

– evaluate the minimum temperatures the particular materials should be designed

for;
– evaluate the total flare capacity;
– evaluate the repulsive forces generated by the relief flow for the design of the var-
ious fixtures;
– organize the inquiry².
Two kinds of equipment are used for pressure relief: safety valves and rupture discs.
Figures 14.3 and 14.4 show a picture and a sketch of a safety valve, respectively.
A safety valve opens gradually at a certain pressure and closes again after relief.
The spring characteristics is adjusted in a way that the safety valve opens at actua-
tion pressure, i.e. at the design pressure of the adjacent piece of equipment. It opens
gradually; it is fully open when the pressure reaches the maximum allowable over-
pressure, corresponding to 110 % of the design pressure³. If there are several safety
valves available to protect the equipment, the maximum allowable overpressure is
116 % of the design pressure, and in the case of fire it is even 121 %. It should be em-
phasized that these pressures including the design pressure refer to overpressures.
Fig. 14.3: Safety valve protecting a heat exchanger.

© Markus Schweiss/Wikimedia Commons/CC BY-SA 3.0
https://creativecommons.org/licenses/by-sa/3.0/
deed.en.
2 In a large plant, 150–200 safety valves and rupture discs are not unusual, so it takes a lot of effort
to gather the information in an appropriate form
3 It is a bit disturbing that the design pressure can be exceeded on purpose. In fact, this case and
the following ones are covered by the definition of the design pressure; exceeding the design pressure
does not mean destruction of the equipment.
Fig. 14.4: Sketch of a safety valve.

© Rasi57/Wikimedia Commons/CC BY-SA 3.0
Safety valves have a hysteresis; if the pressure decreases again to the value of the de-
sign pressure, it will not be fully closed again. For this purpose, the pressure must
decrease to 90 % of the design pressure. The only intention of a safety valve is the
protection of the adjacent apparatus or device. It must not be misused as a pressure
regulation valve.
Working with safety valves, it is most important to distinguish between the par-
ticular pressure terms, which are illustrated in Figure 14.5 with exemplary pressure
values. First, the vessel to be protected shall be explained. As indicated, it is nor-
mally working at a pressure of pnormOp = 3 barg, the maximum operating pressure
is expected to be pmaxOp = 4 barg. The design pressure of the vessel is pDes = 6 barg,
which is relatively far above the maximum operating value. A possible reason is that
6 barg is a standard value for the design pressure of low-pressure vessels. The vessel
is protected by a safety valve, which actuates when the design pressure of the equip-
ment is reached. It is fully open at a pressure which is 10 % higher, i.e. at 6.6 barg,
corresponding to 7.6 bara. This pressure is the maximum allowable overpressure, it
must be maintained. The relief amount is transferred to the safety valve via the inlet
line. The pressure drop in the inlet line must be below 3 % of the actuation pressure.
Just downstream the safety valve in the outlet line, there is the back pressure. It is the
sum of the superimposed back pressure and the built-up back pressure. The super-
p = 1.6 bara = 0.6 barg

back pressure
Outlet line; Δpout = 0.6 bar

built-up
pAct = 6 barg back pressure Header
pfully open = 6.6 barg
p = 1 bara
Inlet line; Δpin = 0.05 bar superimposed
back pressure
Vessel
pDes = 6 barg
pmaxOp = 4 barg
pnormOp = 3 barg
Fig. 14.5: Example case for the illustration of the safety valves pressure terms.
imposed back pressure is the pressure at the end of the outlet line, in Figure 14.5 the
pressure in the header where the relief stream is disposed to. The built-up back pres-
sure is caused by the pressure drop of the relief stream in the outlet line. It should be
less than 10–15 % of the actuation pressure. The values can vary slightly, depending
on the guideline used in the project (e.g. DIN, ASME) and the vendor specifications.
In contrast to a safety valve, a rupture disc (Figure 14.6) opens completely⁴ at a
certain pressure and does not close again, since it is destroyed on being actuated. The
advantages of rupture discs are their rapid actuation at fast pressure buildups, their
low costs and their small space demand. As they have no moving parts, they are very
robust, and almost all materials can be used. Rupture discs are used for large relief
streams with huge cross-flow areas and for fouling or viscous media. Also, they can
be placed upstream a safety valve with a slightly lower actuation pressure than the
safety valve itself. In this way, the safety valve is protected against corrosion, fouling,
and dirt until it actuates. Furthermore, the tightness of the pressure relief arrange-
ment is improved. It should be mentioned that the actuation pressure of a rupture
disc decreases with increasing temperature, depending on the chosen material. For
the design, this significant effect must be taken into account. The pressure terms are
just the same as for safety valves, whereas the pressure drop limitations for inlet and
outlet line are not relevant.
4 depending on the type of the rupture disc. Some types do not open the whole cross-flow area.
Holder outlet
Locating pin Name plate
Locating
pin
J-hook
Pre-assembly
clip Flow Fig. 14.6: Rupture disc.
Holder direction
Rupture disc inlet © Jens Huckauf/Wikimedia Commons/CC BY-SA
tag
(burst) 3.0. https://creativecommons.org/licenses/by-
3-D Rupture disc tag sa/3.0/deed.en.
14.2.2 Mass flow to be discharged
For the design of a pressure relief device, the first step is to derive from the process
knowledge which mass flow has to be discharged. One or more scenarios have to be
fixed where the pressure relief device actuates. In most cases, the standard scenarios
(Chapter 14.2.4) can be taken. Applying these scenarios, the mass flow to be discharged
can be determined. It is useful to do this always in the same way. The “volume balance”
method can help a lot to understand what is happening. It requires the use of a process
simulation program. The following three steps (Figure 14.7) are performed:
1. Calculate the state just before actuation:
Starting from normal operation, the pressure buildup is traced until the actuation
pressure is reached. If the vessel can be assumed to be closed, i.e. no inlet and
outlet flows occur, the volume of the content and therefore the overall density
ΔV must be removed from vessel via safety valve
p = p0 p = p⁎ p = p⁎
Q⁎ ΔQ
τ=0 τ = τ⁎ τ = τ + Δτ
Starting State Moment of Moment just after
actuation actuation
Fig. 14.7: Explanation of the volume balance.

remains constant. The safety valve characteristics, where the maximum pressure
relief is obtained when the safety valve is fully opened, is neglected; instead, for
simplicity it is assumed that the safety valve opens abruptly at the maximum al-
lowable pressure (p. 337). At this pressure, the physical properties are evaluated
as a co-product of the calculation in the process simulator.
2. Calculate the flow just after actuation:
A differential consideration is made to evaluate what will happen just in the mo-
ment of actuation. For example, in the case of fire (see below) a small amount
of heat ∆Q is added. Due to the safety valve, the pressure stays constant, and the
new state of the content of the vessel, especially its new volume, can be evaluated.
The difference between the new volume and the volume of the vessel is the relief
volume ∆V, and the relief amount is simply ∆m = ρ∆V. Assigning a time step ∆τ
(e.g. ∆τ = ∆Q/Q), ̇ the relief stream just at actuation is (Figure 14.7):
ρ ∆V Q̇
ṁ = ρ∆V/∆τ = . (14.1)
∆Q
The fire case example (p. 345) will illustrate this procedure. Note that for a system
in vapor-liquid equilibrium the density of the vapor must be inserted instead of
the overall density, as only vapor is considered to be relieved.
3. Track the blowdown process:
Often, the first relief stream is the largest, and normally, a safety valve should
actuate only for a short time. However, an increase of the relief stream is at least
possible. Therefore, one should regard different points of time during relief and
check with a differential step what happens. According to the tendency, it should
be possible to estimate where the maximum occurs. In fact, for this purpose dy-
namic simulation would be the appropriate tool [213], especially if additional inlet
and outlet streams occur. On the other hand, an averaging of the relief stream over
a long period of time or even over the whole pressure relief process is not advis-
able. Per definition, “averaging” means that the value obtained is smaller than
the maximum value. However, the pressure relief device must be able to govern
any state during pressure relief including the maximum one. An averaging would
lead to a systematic underestimation of the cross-flow area needed.
The advantages of the volume balance procedure are:

– The temperature-elevation of the liquid (and of the vapor as well) is correctly
taken into account, which is important for mixtures with a wide boiling range.
– The volume increase of both vapor and liquid due to the temperature-increase is
correctly represented.
– The change of the equilibrium during the relief is taken into account.
– The “vanishing” of liquid volume due to the evaporation (Equation (14.3), see
below), which is important for actuation near the critical point⁵, is correctly rep-
resented.
– The procedure is as well suitable for the safety valve design of packed columns,
where the holdup of the packing is small. Tray columns are more difficult and
require dynamic simulation [213].
– If necessary, the time until actuation takes place can be evaluated⁶. Often, this
time is considerably high, and many actuation cases turn out to be non-realistic.
14.2.3 Fire case
The fire case is the simplest and most frequently occurring case. Its calculation can
be taken as the basis how to deal with phase equilibria in relief cases. It is subject to
discussion whether the fire case is actually relevant or not. In fact, one should consider
that the heat input with steam is usually higher than the heat input by an external fire,
so that the fire case is often not the governing case. At the sites, in most cases the fire
brigade needs less than ten minutes to be at a plant. This time is normally shorter than
it takes to reach the actuation case with an external fire. And, moreover, most plants
have a sprinkler system, which further reduces the probability of an actuation case.
Nevertheless, the design of safety valves against external fire has more or less become
standard.
Figure 14.8 illustrates what is assumed when fire causes a pressure relief. A vessel
is partly filled with liquid. It is exposed to the heat generated by a pool fire. The vessel
is completely closed, i.e. the valves shown in the adjacent lines are shut. The vessel
is protected by a safety valve. Without liquid in the vessel, there would be hardly any
heat removal to the vessel content. The walls of the vessel would be heated uncontrol-
lably, and the design temperature would probably be soon exceeded. The vessel might
even be destroyed. With liquid in the vessel, there is a better heat transfer from the ves-
sel walls to the liquid. The temperature of the walls is limited due to a working heat
removal. However, the removed heat causes a partly evaporation and a temperature
increase of the liquid. Thus, the pressure in the vessel increases, and after reaching
the design pressure of the vessel, pressure relief is necessary. The safety valve must
be designed in a way that the vapor generated by the heat transferred to the liquid
can be removed by the safety valve without further pressure increase. According to
the established standards, the pool fire reaches a height of 25 ft, i.e. 7.62 m.
5 In fact, such actuations are not as exotic as one would guess. For a vessel with a pure substance, an
actuation pressure close to the critical pressure is sufficient to run into this problem. See also p. 344.
6 Be careful: The heat balance in a closed system has to be performed with the internal energy, not
with the enthalpy as in a process simulation program (Q12 = U2 − U1 ).
Fig. 14.8: Sketch for the fire case.

The valves are assumed to be closed.
With A as the wetted surface up to a height of 25 ft, the commonly used fire formula is
given in the API-521 [215]:
Q̇ H A 0.82
= 43.2F ( 2 ) , (14.2)
kW m
where F is a factor considering the influence of the insulation of the vessel. In the
design basis of many projects, it is instructed that it should be neglected, i.e. F = 1.
Often, there is also a recommendation to which extent the volume of the adjacent pip-
ing shall be considered (normally additional 10 %). The use of Equation (14.2) is often
mandatory and hardly ever questioned. Nobody expects it to be an exact equation, but
in fact, one should know some details about its background. Equation (14.2) refers to
large vessels, like the ones which occur in petrochemical industry. For smaller vessels,
the heat transfer by the fire is different, as the flames have the opportunity to surround
the vessel. In the API-2000 [214], a number of different fire formulas depending on the
size of the vessels is given. Figure 14.9 clearly indicates that they fit much better to the
data, which in turn are given in API-521 [215]. Note that the diagram is logarithmic, so
deviations which seem to be small might indicate a relatively large error. Moreover,
10000
Heat Flux/kW
1000
100
API-521
10 API-2000
Data API-521
1
0.1 1 10 100 1000 Fig. 14.9: Fire formulas in API-2000 [214]
Area/m2 and API-521 [215].
ṁout
mV
mL
.
Q
Fig. 14.10: Sketch of a closed vessel.
Equation (14.2) is not conservative; it calculates values which are systematically too
low. Nevertheless, it is nearly always used.
To evaluate the pressure relief stream, consider a closed vessel with a liquid and a
vapor phase which is heated and protected by a safety valve (Figure 14.10). Assuming
that it is filled with a pure substance, one can derive that the ratio between the given
heat flow Q̇ and the relief stream ṁ out to maintain the relief pressure is
Q̇ ρL
= ∆h v , (14.3)
ṁ out ρL − ρV
where the temperature is the boiling temperature at relief pressure. This relation is
only applicable far away from the critical point. Both ∆h v and ρL − ρV would become
zero at the critical point, and in the vicinity they are inaccurate. Thus, one can assume
that ρL ≫ ρV and
Q̇
= ∆h v . (14.4)
ṁ out
The difference between Equations (14.3) and (14.4) is that the generated vapor was liq-
uid before and leaves some empty volume after evaporation, which will be filled by the
vapor again. Far away from the critical point, Equation (14.4) is accurate; however, this
is the reason for many users to apply it in any possible case. As said, this is acceptable
for pure components, but it is not at all justified to apply it to mixtures.
An enthalpy of vaporization can be assigned if a liquid evaporates at constant tem-
perature and pressure. This is only the case for a pure substance or for an azeotrope,
where, however, the composition changes with the pressure. If a mixture is evapo-
rated, more low-boilers than high-boilers will be evaporated; thus, more high-boilers
remain in the liquid, and the boiling temperature rises. The heating of the liquid re-
quires part of the energy, so less liquid will be evaporated than expected by Equa-
tion (14.4). But it is even more important to find out what is really evaporated, as
vapor and liquid concentration can differ significantly in a mixture. In [11], an exam-
ple is given which shows that averaging the enthalpies of vaporization with respect to
their liquid concentrations can lead to large errors. The only way to account for both
the vapor concentrations and the liquid temperature increase is a flash calculation
using a process simulation program. The coordinates to be specified are the pressure
̇ ṁ out can easily be determined.
and the differential heat flow, and the ratio Q/
Example
In a cylindrical vessel (D = 2.5 m, H = 5 m, height of lower tangent line 5 m) there are 5000 kg of 1,2-
dichloroethane and 5000 kg vinyl chloride at t = 20 °C. The head of the vessel and the adjacent piping
will be neglected. The design pressure of the vessel is pDes = 8 barg, and the ambient pressure is
pU = 1 bar. The vessel is exposed to a pool fire. Calculate the relief amount using a process simulation
program. Determine the relief amount and its state.
Solution
The following steps are performed to obtain the solution:

1. Determination of the wetted area up to 25 ft and the state at the beginning.
First, the volume of the vessel is determined. It is
V = H π D2 /4 = 24.54 m3 .
Using a process simulation program, the state in the vessel is calculated in an iterative procedure.
The pressure in the vessel is repeatedly estimated. The volumes of the vapor and the liquid phase
are determined. If the sum is equal to the volume of the vessel, the pressure estimate was correct.
In this case, the result at t = 20 °C is p = 2.15 bar, with VL = 9.38 m3 and VV = 15.16 m3 . Thus,
the liquid height inside the vessel is evaluated to be
VL 9.38 m3
HL = = = 1.91 m .
π D2 /4 4.909 m2
Adding the height of the tangent line, we get HL,wetted = 6.91 m < 7.62 m. This means that all the
wetted part of the wall is influenced by the fire. The wetted area is
A = π D2 /4 + π D ⋅ HL = 19.91 m2 ,
where the contribution of the adjacent piping is neglected.

2. Heat stream transferred by the fire:
According to Equation (14.2), one gets with F = 1
Q̇ H = 43.2 ⋅ 19.910.82 kW = 502 kW .
3. State of safety valve actuation:

With the process simulation program, the state just before actuation is calculated. For calcula-
tion, it is assumed that actuation takes place at 121 % of the design pressure (fire case), i.e. at
p = 1.21 ⋅ 8 barg = 9.68 barg = 10.68 bara. The vessel temperature must be found where the
volume of the content is equal to the volume of the vessel at pressure p.
It turns out that the actuation temperature is t = 87.99 °C. At this temperature, the liquid volume
is 10.3166 m3 and the vapor volume is 14.2234 m3 , which adds up to the vessel volume⁷. The
vapor composition is already 90.5 wt.% vinyl chloride and 9.5 wt.% 1,2-dichloroethane⁸, being
far away from the liquid concentration at the beginning. The heat required, calculated with a
process simulator as the difference of the internal energies, is
Q = (H2 − p2 V) − (H1 − p1 V) = (H2 − H1 ) − (p2 − p1 )V

= 1.142 ⋅ 106 kJ − (10.68 − 2.15) ⋅ 105 Pa ⋅ 24.54 m3 = 1.121 ⋅ 106 kJ ,
meaning that the time until actuation is
1.121 ⋅ 106 kJ
τ = Q/Q̇ H = = 2233 s ≈ 37 min .
502 kW
4. Determination of relief amount:
A differentially small amount of heat (500 kJ) is added at constant pressure, corresponding to the
heat input by the fire during approx. 1 s. The volumes of vapor and liquid are now VL = 10.315 m3
and VV = 14.286 m3 , giving altogether 24.60 m3 . This exceeds the vessel volume by 0.061 m3 .
This volume, coming certainly from the vapor phase, has to be relieved through the safety valve.
The properties of this stream are part of the stream report of the process simulation program.
For the density of the relief stream, 26.72 kg/m3 are obtained. The relieve flow can then be de-
termined using
∆τ = ∆Q/Q̇ H = 500 kJ/502 kW = 0.996 s
to be (Equation (14.1)):
0.061 m3
ṁ out = ρ∆V/∆τ = 26.72 kg/m3 = 1.636 kg/s = 5891 kg/h .
0.996 s
For comparison with a fictive enthalpy of vaporization according to Equation (14.4), the value for
̇ ṁ out is determined to be
Q/
̇ ṁ out = 502 kW/1.636 kg/s = 307 kJ/kg .

Q/
Averaging of the enthalpies of vaporization would have yielded a value of 281 J/g. An example
with a much more drastic difference is given in [11].
5. Strictly, the whole calculation would have to be repeated for different times, as it is possible that
the relief stream increases with time. For the current example with more than half an hour time
until actuation, this is not really relevant. If it is, the application of dynamic simulation is strongly
recommended.
A strong difficulty for this procedure is the occurrence of inert components like N2 ,
O2 , H2 , etc. If inerts are dissolved in the liquid, the relief pressure is often already
reached at comparably low temperatures. Then the inert will be transferred into the
vapor phase, and the heat added will be mainly used for heating up the liquid. Despite
specifying a differential flash, the temperature then rises significantly by several K. For
7 The accuracy of the numbers seems to be unreasonable; however, one must have in mind that the
target is to evaluate the difference between large numbers, which is always sensitive. See example on
p. 261.
8 using the γ-φ-approach with NRTL and PR for the vapor phase.
the quantity Q/ ̇ ṁ out very large values (e.g. 50 000 J/g) can be obtained, leading to low
design loads for the pressure relief device. To get a reasonable procedure, one should
take into account what would really happen in such a case. It will take some time until
the liquid is really heated up. During this time, the safety valve will really actuate at
the calculated Q/ ̇ ṁ out , giving a very low relief amount. Once the inerts are removed,
normal values will be obtained again. In principle, the standard procedure (p. 340)
works, but item 3, the tracking of the blowdown process with time, has an even larger
meaning.
The following procedure is pragmatic for the design of the safety valves: one could
leave out the inert components and pretend that they were already blown off. It can be
justified by the fact that gas solubility equilibria are reached very slowly. In a process,
one cannot assure that the equilibria calculated in the downstream steps have really
been reached and that the light gases are really dissolved in the mixture, not to men-
tion the uncertainty of the mixing rules for the Henry coefficient (Chapter 2.3.1). The
safety valve must not be designed in a way that the assumed inert gas concentration
is decisive for the evaluation of the relief flow.
14.2.4 Actuation cases
There are a number of reasons why the pressure in a piece of equipment can exceed
the corresponding design pressure [216]. The following itemization can be used as a
checklist for each safety valve, where it is decided whether the particular item is rele-
vant or can be ruled out.
1. Fire case (Chapter 14.2.3)
2. Blocked discharge line:
If the outlet of a vessel is blocked while the feed is still in operation, substance
in the vessel accumulates, which finally leads to a pressure buildup. The most
frequent case is the filling of liquid into a blocked vessel with a pump. If the
pump can build up a pressure higher than the design pressure of the vessel ac-
cording to its characteristics and if other safety measures are already exploited
(e.g. minimum bypass, p. 227), a pressure buildup will be the consequence. The
relief device must then be designed in a way that the feed according to the pump
characteristics can be safely removed. It is always useful to calculate how much
time is necessary before an actuation case is really created. Often, this time is un-
realistically long. Furthermore, one should take into account that the operator
probably gets an alarm after the liquid level range is exceeded.
3. Thermal expansion of the vessel content:
Thermal expansion can become a safety issue when liquids are blocked up in a
closed vessel or a pipe. When there is no gas blanket, meaning that the closed vol-
ume is completely filled with liquid, large pressures are built up when the temper-
ature of the liquid even slightly increases. For example, consider a liquid volume
of water at t = 20 °C, p = 1 bar. During the heating, the density remains constant
at ρ = 998.21 kg/m3 , assuming that the thermal expansion of the vessel is negli-
gible. A temperature increase of just 5 K to t1 = 25 °C at constant density gives a
pressure of p2 = 26.8 bar, which may exceed the design pressure by far. Therefore,
all volumes or pieces of pipe that can be blocked must be protected with a safety
valve which can release part of the liquid. It is called a thermal expansion valve.
Usually, a design calculation is not necessary; the smallest safety valve should be
sufficient.
For gases, thermal expansion is more gradual, so that extreme pressure buildups
do not happen. Often, the relief stream is led to the flare so that it is interesting to
quantify it. For the design, it is likely that the smallest safety valve is sufficient.
A frequently occurring case is the thermal expansion of a gas in a vessel which
is exposed to the sun radiation. Usually, the maximum solar radiation is given in
the design basis. If not, the solar constant (S = 1370 W/m2 ) can be taken, and the
maximum area which the sunlight can hit can be considered⁹. Usually, even this
by far conservative assumption will certainly lead to the selection of the smallest
safety valve. If not, drop the conservative assumptions one by one:
– introduce an absorption coefficient “a” for solar radiation (normally a < 0.6);
– consider heat removal to ambient air by free convection (α = 4–5 W/(m2 K));
– consider radiation exchange with environment.
Example
A gas storage vessel in form of a sphere (D = 10 m) is filled with natural gas (for simplicity: methane).
During normal operation, it is operated at t = 50 °C and p = 100 bar. For the design of the safety valve, it
is assumed that the vessel is blocked and exposed to sun radiation. In the design basis, the maximum
sun radiation is specified to be Smax = 800 W/m2 . To be by far conservative, the heat removal to the
environment by convection and radiation shall be neglected¹⁰. The design pressure of the vessel is
pDes = 110 bar. Calculate how long it will take until the safety valve actuates and the mass flow to be
discharged.
Solution
First, the volume of the sphere is calculated to be

πD3
V= = 523.599 m3 .
6
The projection of the surface exposed to the sun radiation is
πD2
A= = 78.5 m2 ,
4
9 Only the projection area perpendicular to the sun radiation is relevant.

10 which is not justified if a realistic value is targeted.
giving the heat flux

Q̇ = Smax A = 800 W/m2 ⋅ 78.5 m2 = 62.8 kW .
With a high-precision equation of state [29], the mass of the content in the vessel and the heat nec-
essary for reaching the actuation state can be calculated to be
m = ρ(50 °C, 100 bar) ⋅ V = 66.596 kg/m3 ⋅ 523.599 m3 = 34870 kg
and
Q = m ⋅ [u(110 bar, 66.596 kg/m3 ) − u(100 bar, 66.596 kg/m3 )]
= 34870 kg ⋅ (−135.846 J/g − (−178.289 J/g)) = 1480 MJ ,
as the density during the heating phase remains constant. It takes at least
τ = Q/Q̇ = 1480 MJ/62.8 kW = 23567 s = 6.55 h
until the safety valve actuates. The temperature at actuation is t1 = 72.7 °C. The safety valve is fully
open at p2 = 121 bar. In the following, the calculation assumes that the valve opens fully and imme-
diately after reaching p2 .
The heat flux during the first 100 seconds after actuation is regarded, as the safety valve is open,
the procedure is isobaric, and the specific enthalpy is decisive for the energy balance. The relatively
long time is chosen to make sure that the differences between the two states are significant. The gas
expands to the new density ρ100s due to the heat flux:
Q100 s = m ⋅ [h(121 bar, ρ100 s ) − h(121 bar, 66.596 kg/m3 )]

62.8 kW ⋅ 100 s = 34870 kg ⋅ (h(121 bar, ρ100 s ) − 94.367 J/g)) ,
giving h(121 bar, ρ100 s ) = 94.547 J/g. The corresponding density can be evaluated to be ρ100 s =
66.581 kg/m3 , giving a volume after the 100 s of
V100 s = m/ρ100 s = 34870 kg/66.581 kg/m3 = 523.720 m3 .
To maintain the pressure, the difference of the new volume and the volume of the vessel must be
relieved through the safety valve:
∆V = V100 s − V = 523.720 m3 − 523.599 m3 = 0.121 m3 ,
giving the relief amount
∆m = ∆V ⋅ ρ100 s − V = 0.121 m3 ⋅ 66.581 kg/m3 = 8.07 kg ,
which corresponds to a very low mass flow to be discharged of
ṁ discharge = ∆m/100 s = 8.07 kg/100 s = 290.5 kg/h .
4. Chemical reaction:
Runaway chemical reactions are undoubtedly the most serious actuation cases
that can occur. Usually, safety valves do not react fast enough, and the pressure
relief is performed with rupture discs. One of the best known actuation cases was
the so-called Carnival Monday accident (p. 330). The consideration of chemical re-
actions for pressure relief requires at least a rough knowledge about the reaction
kinetics and the heat removal from the vessel where the reaction takes place. An
increase of the conversion of an exothermic reaction develops more heat, which

in turn increases the temperature, and in a vicious circle the reaction kinetics are
further accelerated. As discussed in Chapter 10.2, for runaway reactions the para-
metric sensitivity has to be examined.
5. Tube rupture in a heat exchanger:
In a shell-and-tube heat exchanger, shell side and tube side might have different
design pressures. In case of a tube rupture, the side with the lower design pressure
will be exposed to the higher pressure from the other side. The applied design code
determines whether such an event must be regarded as a pressure relief case or
not. The latest ASME code requires that equipment and piping are tested at 130 %
of the design pressure. Thus, if the lower design pressure is less than 10 13 of the
higher one, tube rupture must be considered as a relief case. Previous revisions of
the ASME code required a test pressure of 150 % of the design pressure; for these
cases, the 23 rule was the criterion.
In Chapter 11, it was pointed out that a tube rupture will happen in longitudinal
direction (Figure 14.11) or at locations which are weakened anyway, like welding
seams. Through the hole formed, substance passes from the high pressure to the
low pressure side and might cause a relief case there. However, the size of the
tube rupture hole is undefined and cannot be predicted at all. The following con-
sideration leads to a scenario which is conservative. The substance passing to the
low-pressure side must pass the two circular cross-flow areas at the ends of the
tube as well. Thus, these cross-flow areas can be regarded as the critical ones. If
the hole created by the tube rupture is smaller, it is conservative, and if it is even
larger, the cross-flow areas are the critical ones anyway.
Fig. 14.11: Tube rupture in a steam reformer tube [217].

Courtesy of IfW Essen GmbH.
The calculation of the relief streams is described in detail in [218], using the ω-
method [219, 220]. The most dramatic pressure relief case is generated if the tube
rupture takes place between a gas at high pressure and a liquid at low pressure.
The critical flow through the cross-flow areas is a gas flow which corresponds to a
large volume flow due to the low density of gases. After having passed the cross-
flow areas, the gas expands to the low pressure and further increases its volume.
To maintain the pressure on the low-pressure side, an equivalent liquid volume
must be removed through the pressure relief device, which in turn corresponds to
a large mass flow due to the high densities of liquids. For this case, the calculation
is comparably easy.
Example
In a shell-and-tube heat exchanger there is nitrogen (tmax = 100 °C) in the tubes (d = 1′′) and cooling
water (tmin = 30 °C) on the shell side. The design pressures are pDes,1 = 50 bar on the tube side and
pDes,2 = 5 bar on the shell side. Determine the necessary relief amount for the safety valve on the shell
side.
Solution
At tube rupture, two circular cross-flow areas at the ends of the tube form the narrowest cross-flow area
at tube rupture, where the nominal diameter of the tubes is taken as the approximate inner diameter
of the tubes:
π π
Atube rupture = 2 ⋅ d2 = 2 ⋅ ⋅ 25.42 mm2 = 1013 mm2 .
4 4
Assuming design pressure and tmax on the nitrogen side, the maximum mass flow can be determined
using the algorithm described in Chapter 14.2.6 (Figure 14.17) to ṁ = 37750 kg/h.
This stream is expanded to p = 5 bar on the shell side. The temperature of the expansion is cal-
culated by adiabatic expansion via
hN2 ,tube = hN2 ,shell .
The heat transfer due to the direct contact to the liquid is neglected, as the relief with the maximum
load is considered to be very fast. The result is tN2 ,shell = 94.9 °C, with the corresponding density of
ρN2 ,shell = 4.57 kg/m3 . The volume flow is then determined to
ṁ 37750 kg/h
V̇ = = = 8260.4 m3 /h .
ρ 4.57 kg/m3
This volume flow will flow onto the shell side and try to displace the cooling water. Neglecting the
small compressibility of the water, this volume flow must be removed from the shell side, otherwise
the nitrogen would build up a pressure which rapidly exceeds the design pressure of the shell side.
As the cooling water is probably next to the safety valve on the shell side, this volume will in the first
moment be displaced as cooling water, with a density of ρ(30 °C, 5 bar) = 995.83 kg/m3 [29]. The
resulting water mass flow is then
ṁ = V̇ ⋅ ρ = 8260.4 m3 /h ⋅ 995.83 kg/m3 = 8225948 kg/h .

6. Abnormal heat input:

If the heating agent in a heat exchanger is not or no longer flow limited, large
temperatures and, subsequently, large pressures on the product side can be the
consequence. A well-known case is the full opening of the steam control valve
so that the largest possible steam flow enters the reboiler of a distillation column
(see “control valve failure” below). A simplified assumption like “steam flow is
proportional to heat input” is not adequate. At relief pressure, one must take into
account that the temperatures on the product side both in the reboiler and in the
condenser are higher than in the design case. The scenario must be evaluated in
an iterative procedure. The column has to be recalculated in the process simu-
lator at relief pressure with estimated reboiler and condenser duties. Then, the
duties have to be verified in the heat exchanger design program, taking into ac-
count that the driving temperature difference is lower in the reboiler and larger in
the condenser. To be conservative, it must be additionally regarded that there is
no fouling in the reboiler for maximum heat input and a fully developed fouling
layer in the condenser to minimize the heat removal, as the relief case can occur
just after the reboiler has been cleaned, while the condenser has been left as it
was. The procedure has to be repeated until the estimated duties in the process
simulation are in line with the ones obtained in the heat exchanger design pro-
gram. The procedure can be simplified if the heat transfer coefficient is supposed
to be constant¹¹.
7. Cooling system failure:
A deficit or a full breakdown of the cooling system can cause pressure buildup,
similar to the scenario “abnormal heat input” described above. For columns ¹²,
an analogous calculation must be performed, taking into account that the reflux
of the column is also affected, as soon as the storage amount in the reflux drum
has been consumed. For this purpose, the application of dynamic simulation is
desirable [213].
8. Power failure:
The power failure is the third scenario which is interesting especially for distilla-
tion columns. The scenario is often weaker than the “cooling system failure”, as
not only the cooling but also the feed pumps and, in some cases, the supply for
the heating agent fail. It has to be clearly pointed out in the scenario which electric
driven pieces of equipment are affected.
9. Control valve failure:
A failure of a control valve is presumed, where the control valve does not reach a
fail safe open position (p. 293) but remains fully open. According to the pressures
11 In this case, the clear separation of process simulation and equipment design (p. 102) is in fact not
advantageous.
12 For distillation columns, “cooling system failure” can also mean that the overhead line is blocked.
involved, more flow than usual will enter the piece of equipment which is to be
protected. During basic engineering, the control valves have not been specified so
far so that a preliminary solution must give an idea about the pressure relief case.
As long as no better knowledge is available, the assumption that the maximum
flow through the valve is twice as large as the maximum flow specified might be
useful. During detail engineering, this assumption should be replaced by actual
information about the valve.
10. Pump failure:
Similar to the control valve failure, it is assumed that the pump flow control is out
of order. According to the pump characteristics, it has to be found out which max-
imum flow can be conveyed by the pump at relief conditions when the pressure on
the discharge side is higher. If the pump has not been specified, some simplifying
assumptions have to be used. Again, a revision during detail engineering should
take place.
All these scenarios are not necessarily independent from each other. If two of them
occur at the same time, it must be distinguished whether one is the consequence
of the other or whether both are assumed to occur together by coincidence (“double
jeopardy”). For the latter case, the probability is generally too low. Unrelated failures
should not occur simultaneously. If they are nevertheless considered, it might lead
to an unreasonably high capacity of the flare system. Control devices or interlocks as
substitutes for pressure relief devices are in general not allowed, unless they have a
SIL classification.
If design cases with significantly different relief amounts occur, it might be an
option to place different safety valves for the different cases in parallel. The actuation
pressures must be chosen in a way that the smaller the relief case, the lower the actu-
ation pressure is chosen. For example, for two different actuation cases and a design
pressure of 6 barg the first safety valve could actuate at 5.9 barg without bothering the
one for the larger relief case, which actuates at 6 barg.
14.2.5 Safety valve peculiarities
In general, things become even more confusing if the pressure relief does not take
place in a vessel with a well-defined content but in a column with a concentration
profile across the apparatus. To start with this topic, the column is closed during the
pressure relief at first, meaning that all inlet and outlet streams are blocked. This might
happen during the fire case, when the entire column is blocked in by quick-acting shut-
off valves.
The formalism for closed vessels as described above can then be applied for the
column; however, in the column there are various liquid reservoirs as holdups on the
particular separation stages, each of them having a different temperature and com-
position and being more or less in equilibrium with its own vapor phase. The simple
method to neglect these holdups and consider only the bottom reservoir might be use-
ful for packed columns with a small holdup and a large bottom area, but in case of
larger holdups it might lead to absurd results, for instance that no light ends are re-
lieved due to their removal in the stripping area, which is usually the purpose of the
column. This is often coupled with a wrong specification of the safety valves. Getting
rid of this problem and take the concentration of the reflux, which has the highest con-
centration of light ends, leads consequently to an overdesign. Many of these shortcut
approaches are in use and give results, but they are neither correct nor even useful. If
some holdup remains on the stages, the column will perform as a distillation column
at pressure relief. When heavy ends evaporate and go upwards, they will tend to get
into equilibrium on the upper stages, where they condense and evaporate light ends.
As light ends tend to have a lower molecular weight and a lower molar heat of vapor-
ization, it might happen that a larger volume flow has to be relieved through the safety
valve than it was generated at the bottom. However, there are some types of columns
where the assumption that the rectification effect can be neglected is justified.
– Sieve trays:
When a sieve tray column is blocked, the liquid flow from above caused by the
reflux is stopped. The pressure profile will break down, as there is no directed
flow any more. As the vapor generated can no more leave at the top, the pressure
rises in the whole column. The trays will be emptied one by one, starting from
the top. This will take some time. If all trays are emptied before actuation, the
column can actually be treated like a vessel. However, to make sure that this pre-
sumption is fulfilled, a separate consideration has to be made. If it is reasonable,
the final composition of the liquid in the bottom can be determined by adding
up all the holdups on the trays with their particular concentration. The amount
of the holdup can be estimated using the weir height¹³, while their compositions
are available from the thermodynamic modeling of the column. Also, the vapor
holdups can be added up. The vapor and the liquid obtained in this way are cer-
tainly not in equilibrium, but they can be used as a starting point for simulating
the further pressure buildup of the column (p. 342).
– Random and structured packings:
Packed columns are emptied after the reflux and the feed are blocked; the entire
holdup of the column, which is comparably small (< 5 %), goes down to the bot-
tom. The considerations discussed for the sieve tray column above hold for packed
columns as well. The holdups can be evaluated by the hydrodynamic models (En-
gel [102], Billet/Schultes [103]). Furthermore, it is important that the holdups in
the collectors and distributors are taken into account.
13 Reasonable assumption: clear liquid on the tray up to 120 % of the weir height.
– Bubble cap and valve tray columns:

Bubble cap trays are more difficult to assess. After reflux and feeds are blocked,
the trays do not completely drain off. Depending on the construction details of the
bubble caps, the vapor will at least partly pass the liquid on the trays and performs
a heat and mass transfer, so that a rectification effect will occur. On valve trays,
the valves close the holes completely after the pressure profile has been equal-
ized. The tray will be filled with liquid up to the weir height. To reach pressure
equilibrium when the column content is heated up, the valves will partly open,
and through the holes part of the liquid will drop down onto the next tray. It can-
not be said whether the trays are completely emptied; this depends on the whole
valve construction. When the safety valve actuates, the vapor will probably pass
at least a small liquid layer, and a certain mass transfer giving a rectification effect
will result.
To summarize: even for the comparably easy case of a column which is fully blocked
there are a number of assumptions involved which cannot be clarified just by setting
up a plausible scenario. The favorite tool is the dynamic process simulation, which
can quantify the assumptions made above for the particular cases. There is hardly
any chance to calculate scenarios where feed, reflux or product stream are still active
without dynamic simulation. Some further considerations about dynamic simulation
for the pressure relief are explained in [213, 221].
Especially for columns, but also in other cases there is another important aspect.
For an oversized safety valve, the pressure relief will achieve its target to reduce the
pressure in the piece of equipment to be protected. On the other hand, the resulting
stream will exert a larger load on the equipment and the piping. A larger relief stream
in a column might destroy the packing or the tray fixings. For example, tray fixings
can withstand a tray pressure drop of approx. 20 mbar; therefore, it is quite probable
that they will be damaged during pressure relief¹⁴. Another point is that the flare loads
can be higher than expected. The most serious item is the repulsive force on the safety
valve and the adjacent piping, which is illustrated in Figure 14.12. During the pressure
relief, the relief flow enters the safety valve from below and leaves it to the right hand
side. The momentum balances for the x- and the y-direction indicate that the reaction
forces R x and R y are induced, giving the resulting reaction force R by vector addition.
In Figure 14.12, some exemplary numbers are set. The result for the repulsive force is
21 kN, equivalent to a weight of approx. 2.1 t. It is clear that pipe engineers must be
aware of the forces exerted on the pipes and their fixings. An underestimation of the
flows involved could lead to possible mechanical damage during pressure relief.
14 It is possible to provide stronger tray fixings; however, in case of an explosion the trays should be
destroyed to protect the shell.
Example:
ρ1 = 331.8 kg/m3 ρ2 = 53.6 kg/m3
p1 = 300 bar p2 = 30 bar
w1 = 65.6 m/s w2 = 45 m/s
A1 = 506 mm A2 = 4558 mm2
p0 = 1 bar
Conservation of momentum:
Rx
– ρ1w12A1 = Ry + (p1 – p0) A1
ρ2w22 A2 = Rx + (p2 – p0) A2
Rx = 13.71 kN
Ry = –15.85 kN
R = R2x + Ry2
R = 20.96 kN Ry
Fig. 14.12: Illustration of repulsive forces.

© Rasi57/Wikimedia Commons/CC BY-SA 3.0.
There are many misunderstandings concerning the expression “conservative assump-

tion”. Frequently, this term is used as a phrase for a “simplifying assumption” [222],
which has often hardly anything to do with being “conservative”. Simplifying assump-
tions are necessary to save time, but they should be applied in a consistent way. For
example, just taking the overhead flow of a column during normal operation for the
pressure relief stream is simplifying. It is neither ensured that it is larger than the
correct relief stream nor that it is at least in the correct order of magnitude. This as-
sumption cannot replace an adequate simulation of the pressure relief. In contrast,
taking the normal flowrate of a pump in an overfilling relief case is conservative, as
the pump flow will in fact be reduced according to the pump characteristics due to the
elevated back pressure. On the other hand, this assumption can be by far too large, and
it might make sense to apply a reasonable estimate of the real relief flowrate. A remark
should be documented with regards to making sure that this estimate is checked as
soon as the characteristics of the pump are known.
It must be distinguished between the mass flow to be discharged and the dis-
chargeable mass flow. From the process calculation of the particular scenarios, one
gets the mass flow to be discharged, i.e. the stream to be relieved to maintain the
pressure in the apparatus at an acceptable level. However, pressure relief devices like
safety valves are produced with defined sizes. One cannot choose a safety valve which
fits exactly the calculated relief stream but the next one in the list where the certified
capacity is sufficient. This means that the relief stream obtained with this safety valve
could be higher than requested. This is the dischargeable mass flow. It must be noted
that according to most of the standards all pressure drop calculations (inlet line, out-
let line) must be based on this dischargeable mass flow, not regarding whether this
is possible from the process point of view [223]. According to a rule of thumb, it is
reasonable if the dischargeable mass flow is about 10–20 % above the mass flow to be
discharged. More overdesign leads to chattering in the actuation case.
Things become even more complicated when there is a two-phase flow through
the safety valve. Many safety valve designs neglect the fact that the pressure relief
is not a smooth equilibrium process as assumed if only vapor is relieved. If vapor is
relieved from a vapor-liquid equilibrium, a bubbling-up of the liquid will take place,
as it might happen that more vapor is formed than can escape through the surface of
the liquid. The reason is that the rising velocity is limited [224], especially when the
viscosity of the liquid is high (> 100 mPa s) or if there is a foam layer on the liquid. The
liquid rises, and in case the liquid level at pressure relief conditions was high enough,
it will be partially relieved through the safety valve as well. For the mentioned system
with high viscosity or foam, this happens even at low liquid levels [225]. In everyday
life, this effect is known as the champagne effect¹⁵.
For the design of the safety valve itself, one should be aware that the entrained liq-
uid covers part of the opening area. Therefore, less gas can be relieved and less energy
can be removed by evaporation of the liquid. Larger opening areas are necessary.
In Figure 14.13, a criterion is given to decide whether only vapor or a two-phase
flow goes through the safety valve [224]. The decisive quantity is the ratio between
the superficial vapor velocity uG0 , calculated with the necessary relief stream (Equa-
tion (14.5)), and the rising velocity of the bubbles u∞ (Equation (14.6)):
ṁ out
uG0 = , (14.5)
ρV A
0.25
[σL g(ρL − ρV )]
u∞ = K∞ . (14.6)
ρ0.5
L
A is the cross-flow area for the gas in the vessel, and the factor K∞ can be taken from
Figure 14.13 according to the model chosen for the description of the bubbling-up. The
diagram is valid for vertical vessels with H/D = 1–3. If it is applied to horizontal ves-
sels, it can be estimated that the result for the limiting liquid level is approximately
5 % too low. Typical values for u∞ are 0.2 m/s for low viscosities and 0.05 m/s for high
viscosities. Reasonable values for the limiting liquid level φlim are approx. 70 % for
systems with low viscosity, approx. 20 % for systems with high viscosity and approx.
10 % for systems with foam¹⁶.
15 After having wasted large amounts of this valuable beverage due to underestimation of the cham-
pagne effect, the author claims that from the pressure relief point of view, the opening of a champagne
bottle is by far too large.
16 φ = VL /Vvessel at the beginning of the pressure relief.
1
Two-phase flow
φ0 > φlim
Limiting level φlim 0.8 Non-foaming,
low viscosity
0.6 (ηL > 100 mPas) Churn turbulent
flow (K∞ = 1.53)
Non-foaming,
0.4 high viscosity Bubbly flow
Vapor flow (ηL > 100 mPas) (K∞ = 1.18)
0.2
φ0 < φlim
Homogeneous
Foaming
0 flow (K∞ = 1.18)
0.1 1 10
Superficial vapor velocity uG0/u∞
Fig. 14.13: Limiting level to avoid two-phase flow through the safety valve [224].
If the liquid level is below the limiting one at the moment just before the actuation, it
can be continued to assume a vapor flow through the safety valve. Unfortunately, this
is only a preliminary result. As mentioned above, the chosen pressure relief device will
be larger than necessary, and the calculation should be repeated with the discharge-
able mass flow, giving a higher uV , and it can happen that the limiting liquid level is
then exceeded.
If the liquid level is above the limiting one, things are more sophisticated. The
size of the pressure relief device can be determined with the ω-method mentioned
above [219, 220]. Again, the solution can only be obtained iteratively. After choosing a
certain size, the calculation must be repeated with the new dischargeable mass flow.
The calculation is only finished if the phase ratio does no more change from step to
step. The procedure is thoroughly described in [224, 226].
14.2.6 Maximum relief amount
Having determined the mass flow to be discharged, the design of the pressure relief
device can take place, i.e. mainly the choice of the opening area at actuation. This area
must be large enough to let the mass flow to be discharged pass in order to maintain
the pressure. The maximum relief amount through an opening area is determined by
the critical flow phenomenon, which must be thoroughly understood.
The opened cross-flow area can be generally treated as an orifice. Figure 14.14
shows the relationship between the mass flux density through an orifice and the outlet
pressure. Consider a pressure vessel at p1 , which has to be relieved through an orifice
to an environment at a pressure p2 . If p2 = p1 , there is no driving force for a flow and
the mass flux density will be zero. The mass flux density increases when the pressure
p2 is lowered. However, at a certain level of p2 (rule of thumb: ≈ 0.5 p1 ) the mass
60000
Critical flow Subcritical flow
50000
Mass flux density kg/m2s
40000
30000
20000
10000
0
0 50 100 150 200 250
Outlet pressure p2/bar
Fig. 14.14: Mass flux density through an orifice as a function of the outlet pressure.
Calculated for nitrogen, p1 = 200 bar, t1 = 20 °C.
1 c 2
p1 p2 Fig. 14.15: Throttle valve.
flux density starts decreasing in the calculation (dashed line in Figure 14.14). It can
be shown with the Second Law that a further acceleration is not possible. In fact, the
mass flux density stays constant. This value is called the critical mass flux density. Its
existence does not depend on the state: it can be liquid, vapor, or two-phase. How-
ever, it is most important and significant for vapor flow, where it can be shown that
the speed of sound will occur in the narrowest cross-flow area [11]. It is a common
misunderstanding that the calculation of the pressure relief is equal to the one for an
adiabatic throttling (Figure 14.15). The well-known condition for adiabatic throttling
h1 = h2 (14.7)
is correct, but it refers to states 1 and 2 far away from the narrowest cross-flow area,
where the velocities are similar or at least in the same order of magnitude so that the
kinetic energy term can be neglected. h1 and h2 are not equal to the enthalpy directly
in the orifice hc , as the velocity is much larger there and should be taken into account
in the energy balance:
w21 w2
+ h1 = c + hc . (14.8)
2 2
Equation (14.8), the continuity equation, the isentropic change of state, the speed of
sound and the equation of state are used to derive a procedure for the determination
of the maximum mass flux density. For an ideal gas, we get [11]
2 κ+1
ṁ c 2 p1 κ pc κ pc κ
=√ [( ) − ( ) ] (14.9)
A v1 κ − 1 p1 p1
with κ
pc 2 κ−1
=( ) . (14.10)
p1 κ+1
1 c 2
Fig. 14.16: Flow pattern through an orifice.

p1 p2 Courtesy of Dana Pfenning.
Equation (14.9) should be supplemented by a factor KD , which considers the fact that
the cross-flow area directly in the orifice is not necessarily the narrowest cross-flow
area. Instead, the typical flow pattern is that the flow is more constricted downstream
the orifice (Figure 14.16), and the effective cross-flow area is smaller than the orifice
itself. The lack of knowledge is summarized in the KD value, which is usually in the
range¹⁷ 0.6 < KD < 0.85:
2 κ+1
ṁ c 2 p1 κ pc κ pc κ
= KD √ [( ) − ( ) ]. (14.11)
A v1 κ − 1 p1 p1
Having calculated the mass flux density, one can easily get the relief amount by simple
multiplication with the chosen cross-flow area.
The vendor of a safety valve or the piping element should be able to deliver more
details about KD , otherwise some advice is given in [227]. Equation (14.11) is sometimes
given in a way that it can barely be recognized, and furthermore, different results are
obtained [228]. So far, it could always be shown that the equations and procedures
are equivalent to Equation (14.11). The differences in the results occur due to different
default values for KD .
17 If nothing else is known, one can use KD = 0.65 for liquids and KD = 0.8 for gases.
For a real gas, it is often considered to use the real specific volume for v1 in Equa-
tion (14.9) instead of the ideal gas density v1 = RT1 /p1 , and furthermore, the real
values for c p and c v are used to determine κ as κ = c p /c v , which often leads to un-
reasonable values. This is contradictory to the thermodynamic derivation and leads
to an inconsistent result. For example, at t = 50 °C, p = 100 bar the ratio for ethylene
is c p /c v = 3.1, whereas the largest theoretically possible value is κ = 1.67 for a one-
atomic ideal gas (He, Ne, Ar, …). It is not possible for a real gas to assign a κ which
fulfills both the ideal gas condition for an isentropic expansion
κ−1
T2 p2 κ
=( ) (14.12)
T1 p1
and the equation for the speed of sound of an ideal gas
w∗ = √κRT . (14.13)
For the above-mentioned condition, the κ for the isentropic expansion which yields
the correct value would be κ ≈ 1.2, whereas κ = 0.998 would be required to get the
correct speed of sound. Both values are obviously different and far away from the ratio
c p /c v . κ < 1 is even impossible, as c p is always larger than c v .
In fact, one is often lucky in using Equation (14.9), as long as the pressures are
not too high. In [229] a number of examples have been compared. The result was
that Equation (14.9) is in fact a good approximation of the exact solution. The max-
imum mass flux density and, correspondingly, the necessary free cross-flow area are
often met quite well. However, it is also pointed out that temperatures and pressures
along the line cannot be reproduced, leading to possibly inadequate design condi-
tions. When applying Equation (14.9), one should take care that κ is in a reasonable
range (1 < κ < 1.67), otherwise, something might go wrong. At least, one should check
how sensitive the results are to variations of κ.
Often the dependence is not strong. If one is in doubt, one should prefer the real
gas calculation [11]. For example, in the LDPE process¹⁸ at p = 3000 bar, where the
density looks more like a liquid than like a gas density, the ideal gas calculation (Equa-
tion (14.9)) does not make sense anymore. For a correct consideration of real gas behav-
ior, an analogous procedure has been set up in [11]. It is not a formula, as the equation
of state is not standardized, but an iterative procedure. The calculation scheme is
listed in Figure 14.17.
For the design of a safety valve or a rupture disc, the inlet line, the pressure relief
device, and the outlet line are a unit which should be designed together in one step.
The obvious advantage is that the properties of the relief stream must be entered only
once. Figure 14.18 illustrates the requirements which the calculation has to fulfill, with
the simplified assumption of a gas flow.
18 LDPE–low density polyethylene.

Starting condition p1, T1

Set s1 = s1
Estimate pc pc1 = pc ⁎1.0001
Calculate Tc(pc,sc), hc(pc,sc), pc(pc,sc) Calculate Tc1(pc1,sc1),hc1(pc1,sc), pc1(pc1,sc)
wc from First Law wc1 from First Law
ṁc from continuity equation ṁc1 from continuity equation
ṁc – ṁc1 < ε
verify wc = w⁎(Tc,pc)
Result
Fig. 14.17: Calculation procedure for the mass flux density of a real fluid.
Flow state
unknown
ρ 2 L
Δpinlet line = λ w
2 d pc
p0 pE ρ 2 L
w
Δpoutlet line = λ
Demand: 2 d
3% of (p0–pU) Max. mass pA Segmental ! pU
flux density
in narrowest Demand:
flow area 10% of (p0–pU)
Fig. 14.18: Inlet line, safety valve and outlet line as a unit.
First, a safety valve is considered as pressure relief device, simplified as an orifice in

the center of the drawing. From the vessel with the relief pressure p0 the relief stream
enters the inlet line on the left hand side. The pressure drop of this line, which is
usually short (1–2 m), can easily be calculated with the conventional formula Equa-
tion (12.1). The requirement is that this pressure drop shall be below 3 % of the driving
pressure difference¹⁹ (p0 − pU ), where pU is the pressure of the destination of the re-
lief stream (usually flare or environment if possible). The safety valve itself cannot be
subject of a conventional pressure drop calculation, e.g. according to Equation (12.18).
19 The requirement can vary slightly, according to the guideline used.

Instead, what we know is that the maximum mass flux density and the critical pressure
are reached in the narrowest cross-flow area, and for a vapor flow we get the speed of
sound. According to the design guidelines, the pressure drop in the outlet line shall
not exceed a certain value, e.g. 10 % of the total pressure difference between the ves-
sel and the lower-pressure location²⁰. The pressure drops of both outlet and inlet line
must be calculated with the dischargeable mass flow through the safety valve, not with
the mass flow to be discharged.
The whole procedure for the dimensioning of the outlet pipe has been described in
Chapter 12.1.3 (p. 281). As the thermodynamic state of the fluid may vary considerably
along the line, the outlet pipe is divided into increments so that the pressure drop
of an increment can be determined with the updated state variables. In an iterative
procedure the pressure after expansion downstream the safety valve is estimated, and
the outlet state of the pipe is calculated, until the estimated pressure after the safety
valve yields the speed of sound at the outlet or expands to environmental pressure. If
the pressure drop exceeds the 10 %, the outlet line diameter must be increased.
For rupture discs, there is no limitation like this, and in most cases the limitation is
the speed of sound at the pipe outlet. The pressure drop can again not be evaluated by
a single pressure drop calculation, as the state of the relieved fluid varies significantly
along the pipe. Instead, an increment-wise calculation must take place. This is espe-
cially important for compressible gas flows and flashing fluids, where the pressure
drop causes further evaporation. The procedure has been described in Chapter 12.1.3
and can be transferred to the two-phase flow pressure drop as well. In contrast to the
inlet line, the course of the outlet line can only be guessed during basic engineering,
as the locations of the particular pieces of equipment and the tie-in points to the flare
system are not known, not to mention the number and kinds of the bends. There must
always be a note that the outlet lines have to be updated in detail engineering.
Some remarks should be given about the pressure drop of the inlet line. Accord-
ing to the established guidelines, the pressure drop in a safety valve inlet line should
not exceed 3 % of the actuation pressure, referring to the dischargeable mass flow
[230]. Although there is no physical background for this rule, Figure 14.19 shows that
it makes some sense.
The blue lines show a case where the 3 %-criterion has been kept²¹. After the ac-
tuation pressure has been reached, the safety valve starts to open, and within a few
milliseconds the full valve lift is obtained. The pressure in the vessel remains almost
constant, and the pressure at the safety valve inlet slightly drops.
20 10–15 %; in the following, 10 % are used for simplicity

21 At first glance, it seems that the pressure drop might be larger than 3 % (10.5 bar at safety valve inlet,
12 bar in the vessel). The reason is that the dynamic pressure due to the velocity is not considered in
the diagram.
100
80
relative valve lift /%
60 3% inlet line pressure drop
20
0
0 50 100 150 200
time/ms
15
14
13
p/bar

10
9
8
0 50 100 150 200
time/ms
Fig. 14.19: Valve lift and pressure at safety valve inlet flange as a function of time for 3 % (blue)
and 6 % pressure drop (red) in the inlet line.
Qualitative remake from [230].
The red lines show a case where the same safety valve has been used with an increased
inlet pipe length so that the pressure drop in the line amounts to approximately 6 % of
the actuation pressure. A completely different situation arises in this case. After actua-
tion, the pressure at the inlet flange of the safety valve drops significantly, causing the
valve lift to drop as well. Then both valve lift and pressure start high-frequency oscil-
lation with approximately 50 Hz. Due to this oscillation, the valve is hardly ever fully
opened, and there are even periods when it is almost closed. The relief stream will be
much lower than the specified one, and the protection of the vessel will probably fail.
For the compliance with the 3 % rule, it often causes trouble that the dischargeable
mass flow has to be considered, and not the mass flow to be discharged. In these cases,
it might be useful to limit the valve lift in a way that the opening of the safety valve
limits the relief stream to the design value. However, this requires an exact knowl-
edge of the possible relief scenarios, and furthermore, one should have a look into the
guidelines and see whether this lift stop is applicable or not. Often, a joint effort of
process and layout engineers can achieve a better location of the safety valve so that
the inlet line pressure drop is reduced. Some examples are described in [230].
The 10 % pressure drop criterion for the outlet line is an arbitrary choice. Certainly,
when the back pressure increases, there will be a point where it is no longer possible
for the safety valve to operate properly. The safety valve will start chattering [231]. It
depends on the safety valve construction at which point this behavior occurs. The 10 %
criterion can be taken as a conservative value given in the particular guidelines. Cases
can occur where it is more or less not possible to keep the 10 % pressure drop in the
outlet line, e.g. at relatively low actuation pressures. Using bellows, it is possible to
extend the pressure drop limit in the outlet line to 30 %. Figure 14.20 illustrates how
the bellows work. On the left hand side, a conventional safety valve is shown. The
back pressure directly acts on the sealing face, which counteracts the simple opening
of the safety valve due to the pressure load of the protected equipment. On the right
hand side, the safety valve has a bellow, which is in principle a gasket which prevents
the back pressure to act on the sealing face. Instead, the pressure to be overcome is the
ambient pressure (see the open hole in the upper part), which is in most cases much
lower and therefore improves the situation.
pamb
FSV FSV
Bellow
pback pback
pback pback
p1 p1
Fig. 14.20: Safety valve sketch with and without bellows.
Bellows are also recommended if a significant part of a liquid relief (≈ 5 %) flow will
flash. Another option is again the above mentioned lift stop. It should be mentioned
that in case of liquid relief the outlet line should be designed with a slope and, if pos-
sible, without pockets to avoid that a liquid column is formed downstream the safety
valve.
It is not always the case that only vapor is relieved without any phase change.
In these cases, there is a maximum mass flux, but in the narrowest cross-flow area
speed of sound does not occur. The algorithm described in Figure 14.17 can be further
applied, but the sensitivities of the pressure ratio are much larger. The following cases
can be distinguished:
– Two-phase flow in the safety valve; already at the inlet:

If there are both vapor and liquid in the relief stream upstream the safety valve,
the necessary cross-flow area of the safety valve could in the past be determined
by simple addition of the cross-flow areas necessary for the vapor and the liquid
phase alone. At present most of the guidelines recommend the ω-method [219,
220].
– The two-phase flow occurs in the outlet line due to flashing:
If the evaporated mass flow downstream the safety valve is less than 50 % of the
entire mass flow, the safety valve should be designed for liquid relief. Often, the
pressure in the narrowest cross-flow area is the boiling pressure of the liquid, as
the first bubble would occupy much more space than the same mass as a liquid.
Therefore, it is clear that the maximum mass flow occurs at conditions where
the whole flow is liquid. The flashing in the outlet line makes it necessary that
the pressure drop correlations for two-phase flow are applied, maybe even as an
EXCEL file where the line is divided into increments, if the pressure drop is so
large that changes due to the phase equilibrium take place. The use of bellows is
strongly recommended in this case.
– Condensation in the safety valve:
In Chapter 8.2 (p. 239) it is explained that there are substances, especially “large”
ones with more than three C-atoms, which form liquid droplets in the compressor,
as the compression yields a lower temperature than the boiling temperature of the
relief stream due to the compression. The other way round, this means that the
substances which do not show liquid formation in a compressor are prone to form
droplets at pressure relief. Astonishingly, water belongs to these substances when
a saturated vapor is expanded. Starting at the dew point line, the temperature
drop during expansion is larger than the drop of the dew point temperature. Also,
speed of sound is not reached in these cases.
14.3 Explosions
If an exothermic reaction is very fast, in might happen that the heat of reaction cannot
be removed. The temperature elevation causes the reaction rate to further increase,
which in turn gives a higher temperature again. Finally, an explosion takes place,
which is in principle a reaction of a substance in a very short period of time (usu-
ally within milliseconds). Starting from a certain point, the reaction finally covers
any amounts of the substance which is in close contact. In contrast to a compara-
bly slow fire event, there is no chance to act and mitigate the consequences; the only
thing an engineer can do is to take all precautions to prevent an explosion under all cir-
cumstances. Concerning the initiation of the ignition, we distinguish between induced
ignition, where energy is supplied from an external source (e.g. sparks, hot surfaces),
and self-ignition, where the substance is heated due to chemical reactions without
sufficient heat removal.
An explosion is coupled with a rapid expansion of gases, which has a large de-
structive potential. One can distinguish between flash fires, deflagrations, and deto-
nations. Flash fires are defined as combustion reactions with rapidly moving flame
fronts. The flame velocity is below the speed of sound. Glass panes break, and people
are injured. There is a continuous transition to deflagrations, where the flame front is
also slower than the speed of sound but can be heard. It can destroy buildings, and
there are often people who are seriously injured or even die. In detonations, the flame
front velocity is faster than the speed of sound [232]. There is extensive damage in a
wide area, as well as casualties.
For the discussion of possible explosions in a plant, we distinguish between the
following zones:
– Zone 0:
an explosive atmosphere occurs more than 50 % of the operation time;
– Zone 1:
an explosive atmosphere occurs frequently, at least 30 min per year;
– Zone 2:
during normal operation, an explosive atmosphere does not occur. By accident it
is possible, but less than 30 min per year.
There are some characteristic numbers which indicate and classify the hazardousness
of a substance.
The flash point is the lowest temperature where the vapor pressure is large enough
to generate ignitable mixtures in ambient air. When the ignition source is removed,
the substance stops burning. In contrast, the fire point is essentially the same, but the
substance continues to burn after removal of the ignition source.
At the autoignition temperature, the substance burns even without an external
source. According to this value, the particular substances are classified in groups, as
indicated in Table 14.2.
Table 14.2: Temperature classes.
Temperature class Range of autoignition temperature Example
T1 > 450 °C hydrogen 536 °C

T2 300–450 °C ethanol 363 °C
T3 200–300 °C diesel 205 °C
T4 135–200 °C diethyl ether 160 °C
acetaldehyde 140 °C
T5 100–135 °C no examples
T6 85–100 °C carbon disulfide 90 °C
(only one)
Explosions take place under certain conditions; there is a lower and an upper ex-
plosion limit. Between these limiting concentrations, a gas can ignite and explode.
The explosion limits refer to air as oxygen containing gas. They are temperature- and
pressure-dependent. Moreover, there are substances where the upper explosion limit
is missing, meaning that they are explosive even without the presence of oxygen. Ex-
amples are ethylene and ethylene oxide. Special care must be taken for dust.
The lower explosion limit (LEL) is the lowest concentration of a gas or a vapor in air
where an ignition source (e.g. flame, heat) causes a flash of fire. Below the LEL there is
not enough fuel to develop an explosion, meaning that concentrations lower than the
LEL are too lean to burn. The LEL generally decreases with increasing temperature and
increases with increasing pressure. For example, methane has an LEL of 4.4 vol.%²² at
t = 138 °C. At t = 20 °C, the LEL is 5.1 vol.%. Below these concentrations, an explosion
cannot take place.
The upper explosion limit (UEL) is the highest concentration of a gas or a vapor
where an ignition or explosion is possible. Mixtures with concentrations higher than
UEL are too rich to burn. The UEL increases with increasing temperature and increas-
ing pressure.
There is a mixing rule for the LEL according to Le Chatelier:
−1
xi
LELmix = (∑ ) , (14.14)
i
LEL i
where x is the mole concentration. For the UEL, it is sometimes recommended to use
Equation (14.14) as well, but in principle, there is no assured mixing rule.
In practical applications like exhaust air lines, a safety margin (usually 50 % LEL)
must be considered, and there must be an analytical surveillance that this limit is
kept. The detector most often used is the flame ionization detector (FID). However, it
only detects C-atoms in the mixture; it does not distinguish between the components.
Therefore, one component is identified to set a conservative standard, as the following
example shows.
Example
A pollutant flow (t = 20 °C, p = 1 bar) of 2.4 kg/h acetaldehyde (1, C2 H4 O, M1 = 44.053 g/mol, LEL1 =
4 mol % = 73.25 g/m3 ) and 3.8 kg/h cyclohexanol (2, C6 H12 O, M2 = 100.161 g/mol, LEL2 = 1 mol % =
41.64 g/m3 ) is transported with an exhaust air stream to the exhaust air treatment. It must be taken
22 vol.% as concentration unit is simply annoying for engineering purposes. It is not exactly defined
how it can be interpreted, and in principle it is temperature-dependent. For gases, it can be assumed
that the gases are ideal, so the volume concentration is equal to the mole concentration. For liquids,
it can be assumed that amounts corresponding to the volume concentration of the particular compo-
nents are mixed, and the excess volume is neglected.
into account that a flame ionization detector cannot distinguish between components which contain
only C, H, and O, as the combustion products are the same. How much exhaust air flow is necessary
to make sure that the resulting stream has less than 50 % LEL according to a conservative standard?
Solution
Considering the C-atoms, the LELs of the two components can be interpreted as follows:
2 ⋅ 12.01 g C/mol
LEL1 = ⋅ 73.25 g/m3 = 39.94 g C/m3
44.053 g/mol
6 ⋅ 12.01 g C/mol
LEL2 = ⋅ 41.64 g/m3 = 29.96 g C/m3 .
100.161 g/mol
Thus, component 2 (cyclohexanol) should be the reference component with the lowest LEL per C-atom.
In the mixture, the detector will identify
2 ⋅ 12.01 g C/mol 6 ⋅ 12.01 g C/mol

ṁ C = ⋅ 2.4 kg/h + ⋅ 3.8 kg/h = 4.04 kg C/h .
44.053 g/mol 100.161 g/mol
Therefore, the stream must be diluted with air to a volume flow of
ṁ C 4.04 kg C/h
V̇ air = = = 270 m3 /h .
0.5 LEL2 0.5 ⋅ 29.96 g C/m3
Most chemical processes do not operate with air but rather with any mixtures which
can contain arbitrary amounts of oxidizing substances. Figure 14.21 shows an exam-
ple for the flammability diagram of the system methane/nitrogen/oxygen. For a given
temperature and pressure, the region where explosions can happen is dark colored.
Additionally, some useful straight lines are shown.
First, there is the stochiometric line, where there is as much oxygen in the mixture
for the complete oxidization of the methane. According to the reaction
CH4 + 2 O2 󳨀→ CO2 + 2 H2 O ,
2 mol oxygen are needed for the combustion of 1 mol methane, meaning that the mole
concentration of methane is approximately 33 % in case there is no nitrogen. The line
shows all mixtures where this ratio is kept.
Second, the air line is shown, where the ratio between nitrogen and oxygen is
approx. 7921 as it is in air, independent of the methane concentration. The intersections
of this line with the limits of the explosion region indicate the LEL and, respectively,
the UEL. For process control, the LOC (limiting oxygen concentration) line is the de-
cisive one. It indicates the lowest oxygen concentration where the explosion region
is touched. Independent of the concentrations of the other substances, staying below
the LOC ensures that an explosion does not take place. The oxygen concentration is
relatively easy to supervise.
100 0
90 10
80 20
70 30
Stoichiometric line
60 40 Air line
Fuel (methane) 50 50 Nitrogen LOC
40 60
30 70
20 80
UEL
10 Flammable mixtures 90
LEL
0 100
100 90 80 70 60 50 40 30 20 10 0
Oxygen
Fig. 14.21: Explosion regions of the system methane/nitrogen/oxygen.

© Power.corrupts/Wikimedia Commons/CC BY-SA-3.0
260
250
240
230
T/K
T_Joule–Thomson
220 T_gas
210 T_metal
T_valve
200
190
3 5 7 9
p/MPa
Fig. 14.22: Temperature course of a nitrogen gas cylinder at emergency inertization.

Courtesy of Wystrach GmbH.
The addition of an inert gas (usually nitrogen) increases the LEL and lowers the UEL.
The knowledge of the explosion limits enables the engineer to determine the necessary
flow of inert gas. It has to be assured that this flow can be delivered in any case; e.g.
the failure of a compressor or the breakdown of the electrical energy supply must not
cause a lack of inert gas delivery. Often, gas cylinders filled with nitrogen under pres-
sure are provided as an independent emergency supply for a limited time. However, it
must be carefully determined how much gas is really in the gas cylinders, so the filling
conditions must be well defined. When the temperature decreases, for instance on a
cold winter day, the pressure inside the cylinders will drop. The coldest winter day is
the basis for the dimensioning of the inert gas supply. At actuation, the temperature
in the cylinders further decreases due to the expansion, which in turn causes the pres-
sure in the cylinder to decrease even more rapidly. On the other hand, the temperature
decreases more slowly than thermodynamics suggests, as the steel of the vessel itself
with its large mass behaves as a heat storage and transfers heat to the gas by natural
convection. With time, the wall temperature decreases as well, and this is a decisive
issue for choosing the material for the cylinders. It must be evaluated in advance how
much inert gas can be delivered at emergency case. Furthermore, it must be consid-
ered that the temperature downstream the outlet valve further decreases due to the
Joule–Thomson-effect. But the lowest temperature occurs inside the valve because of
the enthalpy loss due to the acceleration to speed of sound. Figure 14.22 shows the
course of the various temperatures of interest [233].
Glossary
Active area: The area on a distillation tray where the mass transfer takes place.
Activity coefficient (γ): A factor describing the deviations from Raoult’s law. It can be interpreted as a
correction factor for the concentration. γ is a function of temperature and concentration. It is not
particularly dependent on the pressure.
Adsorbate: Phase on the surface of the adsorbent.
Adsorbent: Adsorptive agent; a solid which can develop bonds to one or more fluid substances to
remove them from a liquid or a gas.
Adsorptive: One or more components in a gas or a liquid which can be adsorbed by the adsorbent.
Advanced process control: Number of measures for improvement of process economics with sophis-
ticated control strategies, e.g. feedforward control or simulation-based predictive control.
Aerosol: An aerosol consists of liquid droplets in the vapor phase which are so small that they do not
precipitate (Equation (9.9)) but large enough that they do not take part in molecular diffusion.
Their occurrence is caused by oversaturation in the vapor phase. Well-known examples occur
when sulfuric acid or hydrogen chloride are absorbed in aqueous phases or if the cooling in cryo-
condensation is too strong (Chapter 13.4.1).
Azeotrope: Phase equilibrium where vapor and liquid concentration of all components are identical,
while the equilibrium pressure at constant temperature or, respectively, the equilibrium temper-
ature at constant pressure shows a maximum or minimum. The separation of the components
of an azeotrope is not possible with simple distillation. The closer the vapor pressures of two
components are, the more probable is the occurrence of an azeotrope. We distingush between
homogeneous and heterogeneous azeotropes, where the latter shows a miscibility gap in the
liquid phase. Azeotropes are also possible for ternary mixtures. Also, there are few examples for
quaternary azeotropes. There is no evidence for azeotropes consisting of more than four compo-
nents.
Battery limits: Battery Limits are the physical boundaries of a plant. Usually, flow meters are installed
at this location in order to determine the economic performance of a plant.
BOD: Biological oxygen demand, p. 323.
Boiler feed water: Demineralized and pretreated water suitable for generating steam. Metal ions,
salts, organics, oxygen, carbon dioxide, and hydrogen sulfide have been removed to an accept-
able level. Often, nitrogen-based weak caustics are added (ammonia, amines).
Brainstorming: A technique for solving problems in a group. It is based on spontaneous contributions
of the particular members of the group. During the brainstorming phase, the proposals must not
be subject to critics.
By-product: A product formed due to undesired side reactions.
Cause & effect matrix: A matrix which gives an overview on the actions caused by interlocks in a
process. For each deviation, the actions caused by the interlocks are marked so that the often
complex interlock description can be interpreted more easily.
Check valve: A valve which is fully open in one flow direction. For reverse flow, it closes due to its
mechanical construction.
Coalescer: An apparatus where droplets unify to a single phase.
COD: Chemical oxygen demand, p. 323.
Compressibility factor: Deviation from the ideal gas behavior, defined as Z = pv/RT . For an ideal gas,
Z = 1. At the critical point, Z is in the range Z = 0.23–0.29. At very high pressures, Z can show
large values, e.g. Z = 4.57 for ethylene at t = 100 °C, p = 3000 bar.
Contingency: Cost estimation item to account for uncertainties in the process or in project execution.
374 | Glossary
Cooling water: Water used for cooling purposes, usually taken from natural sources like rivers, wells,
or sea water. In open cooling water cycles, it is used with no further treatment so that it might
contain salts which can lead to fouling. The supply temperature ranges from 25–35 °C. Usually,
the return temperature is 10 K higher. In most applications, the returned cooling water is cooled
down again in a cooling tower.
Coproduct: A product generated because it occurs in the reaction equation of the desired main reac-
tion.
Depreciation: Depreciation is the value loss of an asset with time. At the end of its lifetime, the value
of the asset becomes zero. During this time, its value continuously decreases from the purchase
price to zero. In the easiest case, this happens linearly with time, other courses are possible. De-
preciations are costs which can be assigned each year and therefore, they have an impact on the
amount of taxes the company has to pay due to the income statement. The more the depreciation
is, the less taxes have to be paid by the company. It is legally required that the depreciation is
spread over the estimated lifetime. The course of the depreciation with time is defined by the
government. The lifetime itself is also fixed by law; often 10 years for ISBL items and 20 years for
OSBL items.
Design basis: A document which collects all the facts and assumptions known in advance before the
project starts. It defines the boundary conditions of the project (environmental conditions, phys-
ical state and composition of raw materials, utilities, products, etc.).
Design pressure: Chapter 11.
Design temperature: Chapter 11.
Deviation: Departure from design and operating intention [209].
Dip-pipe: A feed line to a vessel which does not end at the nozzle but is elongated to the bottom inside
the vessel so that it dips into the liquid during operation. The purpose is to prevent backflow of
the vapor phase of the vessel.
Double jeopardy: Double jeopardy scenarios are two unrelated failures occurring simultaneously. As
the probability of simultaneous independent errors is small, they should not be considered.
Enthalpy: An item explained so well in many thermodynamic textbooks [153]. To make an own short
try: Enthalpy is defined as internal energy plus potential energy in the pressure field: h = u + pv.
It is the usual quantity for the thermal energy of a flowing substance, whereas the internal energy
is relevant for static systems.
Entropy: No reasonable explanation in just a few sentences possible. Again a short try: The entropy
represents the experience associated with the behavior of a system. For a closed system where
neither mass nor heat can pass the system border, the entropy reaches a maximum according
to the Second Law of thermodynamics. Such a system will end up in a state which is the most
probable one. An example: Two gases separated by a wall will mix when the wall is removed, until
the concentration is the same in every volume element. Better and more extensive explanations
can be found in [153] and [234].
Equation of state: A mathematical relationship between pressure (p), specific volume (v), and tem-
perature (T ).
Excess enthalpy: Enthalpy change when two or more liquid or gaseous components are mixed at
constant temperature and pressure. The mixture is supposed to remain liquid or gaseous, re-
spectively.
Excess volume: Change of the specific volume which occurs when two or more liquid or gaseous com-
ponents are mixed at constant temperature and pressure. The mixture is supposed to remain
liquid or gaseous, respectively.
Fixed costs: Operation costs which occur independently from the production, e.g. personnel costs.
Froude number (Fr): Ratio between gravity force and inertia force. The general calculation formula is
Fr = w 2 /(g l), where l is a characteristic length.
Glossary | 375
Fuel gas: Natural gas. Normally, methane is the dominating component, the rest of the composition
depends on the case. Ethane, propane, butanes, higher hydrocarbons, nitrogen, hydrogen, oxy-
gen, carbon dioxide, carbon monoxide, helium, argon, hydrogen sulfide, and water can occur as
components.
Fugacity coefficient (φ): A correction representing the deviation of the chemical potential from ideal
gas behavior. It is thoroughly explained in [11].
Grashof number (Gr): Ratio between buoyancy force and friction force.
Guideword: A simple phrase used to identify possible deviations during a HAZOP procedure.
HAZOP: Hazard and Operability review. A formal and systematic approach for the identification of the
potential of hazards and operating problems caused by deviations from the intended design and
operation [209].
HETP: Height equivalent of one theoretical plate. The height of layer of random or structured pack-
ing which corresponds to a theoretical stage. It is a measure of the separation efficiency of the
packing. The reciprocal value is the number of theoretical stages per m.
Holdup: Relative liquid content of the packing during operation.
Internal energy: Quantity for the description of the thermal energy of a substance which is not flowing
but encased in a vessel. A better explanation can be found in [153].
Interlock: A defined automatic intervention of the process control system. Reaction of the control sys-
tem to encounter unacceptable deviations from normal process conditions.
ISBL: All items which are in the scope of the engineering company and which are directly related to
the process.
Lever rule: The lever rule says that in phase equilibrium diagrams showing the concentration on the
x-axis the ratio of the amounts of the phases corresponds to the ratio of the opposite lever arms
of the tie-line.
Liquidus line: The liquidus line is a limiting line in a solid-liquid phase equilibrium diagram. Above
the liquidus line, there is no solid present.
Mach number (Ma): The Mach number is the ratio between the actual velocity and the speed of sound.
Node: A part of the process which covers a dedicated task, e.g. a distillation step.
NPSH value: net positive suction head; see Chapter 8.1.
Nußelt number (Nu): Ratio between heat transfer by convection and heat transfer by conduction. The
general formula is Nu = α l/λ, with l as a characteristic length.
OSBL: Outside battery limits. Auxiliary units which are required for the functioning of the production
unit, but are not directly involved in the process. In contrast to the main units, they can be shared
among different production units. Examples: steam generator, cooling water system, inert gas
supply, refrigeration unit, instrument air unit.
Package unit: A package unit is a compilation of pieces of equipment that fulfill a certain task in the
project. It is delivered as a unit with defined inlets and outlets by one vendor. Examples are com-
pressors, refrigeration units, crystallizers, or adsorber units.
PFD (process flow diagram): An engineering drawing illustrating the process without showing details
which are not necessary for the understanding. It is often supplemented by process and equip-
ment data.
pH value: Quantity describing acid/alkaline behavior and the strength of an electrolyte solution. It is
defined as the negative of the logarithm of the H3 O+ ion concentration in mol/l to the basis of 10.
In fact, it would be more correct to replace concentration by activity, which is actually performed
in process simulation. The pH of neutral water is 7. Acids have lower pHs, caustic solutions have
higher ones.
PID (piping and instrumentation diagram): An engineering drawing showing the arrangement of pip-
ing and equipment like vessels, heat exchangers, columns, pumps, compressors, and the as-
sociated measurement and control devices. The pipes are depicted together with information
376 | Glossary
about nominal diameter, design pressure, medium, piping class, and an identification number.
The function of the control loops should be clarified on the PID, together with other documents
(cause and effect matrix, data sheets for measurement devices), as well as the installation height
of the apparatuses.
Prandtl number (Pr): Ratio between kinematic viscosity and thermal diffusivity. Pr = η/(c p λ)
Process control system: Computer system to enable the operators to keep the overview on the state
of the whole process and to control it from a central measuring station. A “PFD” of the plant is
displayed and controllers are represented as software, enabling the operator to change control
parameters. Any measured data can be monitored and stored so that arbitrary trend lines can be
visualized; furthermore, any actions can be carried out centrally from this station (e.g. vary the
set point of a pressure or switch off a pump).
Process water: Water which has to be pretreated in a way that it can be used in the process.
Rectifying section: Column region between feed and top, where the light ends are enriched.
Reynolds number (Re): Ratio between inertia force and friction force. The general calculation formula
is Re = w l ρ/η, where l is a characteristic length, for instance the inner diameter in a pipe.
Safety valve: A valve which opens automatically when the pressure in an apparatus exceeds an ac-
ceptable value. In case of actuation, substance is released from the equipment so that the pres-
sure is lowered.
Safeguard: Countermeasure to prevent or mitigate the risk of a deviation [209].
Separation factor: The ratio (y1 /y2 )/(x1 /x2 ). If it is far away from 1, the separation is easy. The closer
it is to 1, the more difficult is distillation.
Speed of sound: The speed at which a pressure disturbance can be transported through a substance.
It can be measured with a remarkable accuracy and is therefore a key quantity for the develop-
ment of equations of state. Flows in pipes cannot exceed the speed of sound, which is a key
fundamental for all pressure relief calculations.
Split block or separation block: A block in process simulation which just defines how a stream is split
by a certain unit operation. The split can be different for the particular components. It is just a
book-keeper function; the physical background is not questioned, and there is no physical check
whether the suggested separation is possible or not.
Steamout: Steamout means that the vessel is cleaned by exposing the surfaces to steam, where high
temperatures are applied. Polymer deposits and other solids might be melted due to the high
temperature or dissolved and therefore removed from the wall.
Stripping section: Column region between bottom and feed, where the heavy ends are enriched.
Tangent line: Level in a vessel which indicates the position of the cylindrical part; the bottom is left
out.
Tear stream: In a chemical process there are usually recycle streams. They are a challenge for process
simulation, as they cannot be known in advance. To come to a solution, they are first estimated,
and after recalculation it is checked whether the estimation was accurate enough. If not, the
estimation of this stream is revised in a certain manner depending on the convergence algorithm.
These streams are called tear streams.
Tie-in points: Defined points where a new part of a plant is connected to an existing one.
TOC: Total organic carbon, p. 323.
Typical: Depiction of an example for the arrangement of the standard equipment, i.e. valves, pumps,
etc.
Value engineering: An engineering procedure which generates suggestions for the economic and
technical improvement of a process and gives an assessment whether these suggestions should
be realized or not. Usually, it starts with a brainstorming session, where new ideas are de-
veloped. In a second phase, people are assigned to evaluate the economic improvement of
Glossary | 377
the particular measures. These people compile standardized reports so that their assessment
becomes comprehensible for both current and future colleagues.
Vapor pressure: The pressure of a pure substance exerted by the vapor which is in equilibrium with
its condensate in a closed system. The vapor pressure is a key quantity for the estimation of pure
component properties and for evaporation, condensation and distillation processes.
Variable costs: Costs which are directly related to the production amount (raw materials, auxiliary
chemicals, utilities)
Weber number (We): Measure of the relative importance of inertia forces compared to the surface ten-
sion.
Working capital: Working capital comprises inventories of raw and auxiliary materials, catalysts,
stores of products and intermediate products, debt claims, and liquid assets.
ω-method: The ω-method is a simplified model to consider two-phase flow through a pressure relief
device. It assumes equilibrium between vapor and liquid phase. There shall be no friction be-
tween the phases. The liquid phase is assumed to be incompressible, the vapor phase obeys the
ideal gas equation of state. The method defines a simplified equation of state for a two-phase
flow, where the whole input information can be used from the state at the inlet. There should
be sufficient distance to the critical point. The critical pressure ratio and the maximum mass flux
density can be calculated iteratively with a simple EXCEL file. The ω-method is considered to be
conservative. Thorough information can be found in [219, 220].
List of Symbols
Symbol Unit Explanation
a N m4 /mol2 attractive parameter in cubic equations of state

ai activity
A m2 area
A ij , B ij , C ij , D ij interaction parameters for Wilson, NRTL, UNIQUAC
b m3 /mol repulsive parameter in cubic equations of state
B, C, D virial coefficients
c m3 /mol parameter for volume translation
C kg/h capacity
cp J/(mol K), J/(g K) spec. isobaric heat capacity
cv J/(mol K), J/(g K) spec. isochoric heat capacity
D m diameter
d m diameter
dh m hydraulic diameter
f Pa fugacity
g J/mol spec. Gibbs energy
g 9.81 m/s2 gravity acceleration
G J Gibbs energy
gE J/mol spec. excess Gibbs energy
h J/mol, J/g specific enthalpy
hW m weir height
H m delivery height
H J enthalpy
H ij Pa Henry coefficient of component i in solvent j
hE J/mol excess enthalpy
I A electric current
k W/(m2 K) heat transition coefficient
k ij interaction parameter in cubic equations of state
k m roughness
K chemical equilibrium constant
KV m3 /h valve characterization value
M g/mol molecular weight
Ma Mach number
ṁ kg/h mass flow
n mol number of moles
p Pa pressure
P € price
P W power
Poy j Poynting correction
ps Pa vapor pressure
Q̇ W heat flow
R 8.31447 J/(mol K) universal gas constant
R Ω electrical resistance
R K/W thermal resistance
Re Reynolds number
s J/(mol K), J/(g K) specific entropy
380 | List of Symbols
s m wall thickness
t °C Celsius temperature
T K absolute temperature
Tb K normal boiling point
Tr reduced temperature, Tr = T/Tc
U J internal energy
U V voltage
u m/s velocity
u J/mol, J/g specific internal energy
v m3 /mol, m3 /kg specific volume
V m3 volume
V̇ m3 /h volume flow
wt J/g technical work
w m/s velocity
w∗ m/s speed of sound
x mol/mol, g/g liquid concentration
x mol/mol vapor quality
x ij mol/mol local concentration of molecule i around molecule j
y mol/mol, g/g vapor concentration
z mol/mol vapor or liquid concentration
Z compressibility factor
Greek symbols
α W/m2 K heat transfer coefficient
α ij separation factor
γj activity coefficient
∆hm J/mol, J/g enthalpy of fusion
∆hv J/mol, J/g enthalpy of vaporization
η Pa s dynamic viscosity
η efficiency
ηM Murphree efficiency
κ isentropic exponent
λ W/Km thermal conductivity
λ friction factor
ν m2 /s kinematic viscosity
Π Pa osmotic pressure
ρ mol/m3 , kg/m3 density
σ N/m2 mechanical tension
σ N/m surface tension
τ s time
φj fugacity coefficient of component j
ω acentric factor
Subscripts
a axial
DC downcomer
i, j, k components, molecules
jet jet pump
L liquid
List of Symbols | 381
m melting
M mixture
r reduced (divided by the critical property)
res residence
rev reversible case
s saturation
SV safety valve
t tangential
t technical
U ambient
V vapor
w weight
Supercsripts
id ideal gas
L liquid
S solid
V vapor
∞ infinite dilution
󸀠 saturated liquid
󸀠󸀠 saturated vapor
I 1st liquid phase
II 2nd liquid phase
Bibliography
[1] Franke A, Kussi J, Richert H, Rittmeister M, von Wedel L, Zeck S. Offene Standards verbinden.
CITplus 2013;16(7/8):23–27.
[2] Mosberger E. Chemical Plant Design and Construction. Weinheim: Wiley-VCH; 2012.
(Ullmann’s Encyclopedia of Industrial Chemistry).
[3] Available at: www.mbaofficial.com/mba-courses/operations-management/what-are-the-
objectives-principles-and-types-of-plant-layout/.
[4] Bowers P, Khangura B, Noakes K. Process plant engineering models; Available at:
http://spedweb.com/index.php/component/content/article/392.html.
[5] Rovaglio M, Scheele T. Immersive virtual plant reality; Available at: http://software.
schneider-electric.com/pdf/white-paper/immersive-virtual-reality-plant-a-comprehensive-
plant-crew-training-solution/.
[6] Borissova H. Produktlebensphasenorientierte Informationsvisualisierung mit graphischen
Metaphern. Diploma task, University of Karlsruhe; 2005.
[7] Smith R. Chemical Process. Design and Integration. West Sussex: John Wiley & Sons; 2005.
[8] Baerns M, Behr A, Brehm A, Gmehling J, Hinrichsen KO, Hofmann H, et al. Technische Chemie,
2nd ed. Wiley-VCH; 2013.
[9] Buskies U. Economic process optimization strategies. Chem Eng Technol 1997;20:63–70.
[10] Cie A, Lantz S, Schlarp R, Tzakas M. Renewable acrylic acid. Tech. rep., University of Pennsyl-
vania; 2012.
[11] Gmehling J, Kolbe B, Kleiber M, Rarey J. Chemical Thermodynamics for Process Simulation.
Weinheim: Wiley-VCH; 2012.
[12] Salerno D. Data on demand. Benefits of NIST TDE in Aspen Plus; 2014. Presentation
ASPEN V8.4.
[13] Nannoolal Y, Rarey J, Ramjugernath D. Estimation of pure component properties, Part 3:
Estimation of the vapor pressure of non-electrolyte organic compounds via group contribu-
tions and group interactions. Fluid Phase Equilibria 2008;269(1/2):117–133.
[14] Kleiber M, Axmann JK. Evolutionary algorithms for the optimization of Modified UNIFAC
parameters. Computers and Chemical Engineering 1998;23:63–82.
[15] Krooshof G. Can molecular modeling meet the industrial need for robust and quick predic-
tions?; 2014. Presentation ESAT, Eindhoven.
[16] van Ness HC. Thermodynamics in the treatment of vapor/liquid equilibrium (VLE) data.
Pure Appl Chem 1995;67(6):859–872.
[17] Loehe JR, van Ness HC, Abbott MM. Vapor/liquid/liquid equilibrium. Total-pressure data and
GE for water/methyl acetate at 50 degree C. J Chem Eng Data 1983;28(4):405–407.
[18] Gaw WJ, Swinton FL. Thermodynamic properties of binary systems containing hexafluoro-
benzene, Part 4: Excess Gibbs free energies of the three systems hexafluorobenzene +
benzene, toluene, and p-xylene. Trans Faraday Soc 1968;64:2023–2034.
[19] van der Waals JD. Over de Continuiteit van den Gas- en Vloeistoftoestand. Thesis, Leiden;
1873.
[20] Bronstein IN, Semendjajew KA. Taschenbuch der Mathematik, 21st ed. Thun/Frankfurt a. M.:
Verlag Harri Deutsch; 1984.
[21] Peng DY, Robinson DB. A new two-constant equation of state. Ind Eng Chem Fundam 1976;
15(1):59–64.
[22] Diedrichs A, Rarey J, Gmehling J. Prediction of liquid heat capacities by the group contribution
equation of state VTPR. Fluid Phase Equilibria 2006;248:56–69.
384 | Bibliography
[23] Benedict M, Webb GB, Rubin LC. An empirical equation for thermodynamic properties of light
hydrocarbons and their mixtures. I: Methane, ethane, propane and n-butane. J Chem Phys
1940;8:334–345.
[24] Benedict M, Webb GB, Rubin LC. An empirical equation for thermodynamic properties of light
hydrocarbons and their mixtures. II: Mixtures of methane, ethane, propane, and n-butane.
J Chem Phys 1942;10:747–758.
[25] Bender E. Equations of state for ethylene and propylene. Cryogenics 1975;667–673.
[26] Span R, Wagner W. Equations of state for technical applications. I: Simultaneously optimized
functional forms for nonpolar and polar fluids. Int J Thermophys 2003;24(1):1–39.
[27] Span R, Wagner W. Equations of state for technical applications. II: Results for nonpolar
fluids. Int J Thermophys 2003;24(1):41–109.
[28] Span R, Wagner W. Equations of state for technical applications. III: Results for polar fluids.
Int J Thermophys 2003;24(1):111–162.
[29] Wagner W. FLUIDCAL. Software for the calculation of thermodynamic and transport properties
of several fluids. Tech. rep., Ruhr-Universität Bochum; 2005.
[30] Kunz O, Wagner W. The GERG-2008 wide-range equation of state for natural gases and other
mixtures: An expansion of GERG-2004. J Chem Eng Data 2012;57:3032–3091.
[31] Wilson GM. Vapor-liquid equilibrium. XI: A new expression for the excess free energy of mix-
ing. J Am Chem Soc 1964;20:127–130.
[32] Renon H, Prausnitz JM. Local compositions in thermodynamic excess functions for liquid
mixtures. AIChE Journal 1968;14(1):135–145.
[33] Abrams DS, Prausnitz JM. Statistical thermodynamics of liquid mixtures: A new expression
for the excess Gibbs energy of partly or completely miscible systems. AIChE Journal 1975;21:
116–128.
[34] Gmehling J, Brehm A. Lehrbuch der Technischen Chemie. Band 2: Grundoperationen.
Stuttgart/New York: Georg Thieme Verlag; 1996.
[35] Prausnitz JM, Lichtenthaler RN, de Azevedo EG. Molecular thermodynamics of fluid-phase
equilibria. Prentice-Hall; 1986.
[36] Anisimov VM, Zelesnyi VP, Semenjuk JV, Cerniak JA. Thermodynamic properties of the mixture
FC218-HFC134a (Russ.). Inzenernyi fiziceskij zyrnal 1996;69(5):756–760.
[37] Gmehling J, Menke J, Krafczyk J, Fischer K. Azeotropic Data, 2nd ed. Weinheim: Wiley-VCH;
2004. 3 volumes.
[38] Hankinson RW, Thomson GH. A new correlation for saturated densities of liquids and their
mixtures. AIChE Journal 1979;25(4):653–663.
[39] Wagner W. New vapour pressure measurements for argon and nitrogen and a new method of
establishing rational vapour pressure equations. Cryogenics 1973;13:470–482.
[40] Moller B, Rarey J, Ramjugernath D. Estimation of the vapor pressure of non-electrolyte organic
compounds via group contributions and group interactions. Journal of Molecular Liquids
2008;143:52–63.
[41] Poling BE, Prausnitz JM, O’Connell JP. The Properties of Gases and Liquids. McGraw-Hill;
2001.
[42] Kleiber M, Joh R. Liquids and gases. VDI Heat Atlas, 2nd ed., chap. D3.1. Berlin/Heidelberg:
Springer-Verlag; 2010.
[43] McGarry J. Correlation and prediction of the vapor pressures of pure liquids over large pres-
sure ranges. Ind Eng Chem Proc Des Dev 1983;22:313–322.
[44] Hoffmann W, Florin F. Zweckmäßige Darstellung von Dampfdruckkurven. Verfahrenstechnik,
Z VDI-Beiheft 1943;2:47–51.
[45] Cordes W, Rarey J. A new method for the estimation of the normal boiling point of non-
electrolyte organic compounds. Fluid Phase Equilibria 2002;201:409–433.
Bibliography | 385
[46] Franck EU, Meyer F. Fluorwasserstoff III, Spezifische Wärme und Assoziation im Gas bei
niedrigem Druck. Z Elektrochem, Ber Bunsenges Phys Chem 1959;63(5):571–582.
[47] Chen CC, Britt HI, Boston JF, Evans LB. Local composition model for excess Gibbs energy of
electrolyte systems, Part I: Single solvent, single completely dissociated electrolyte systems.
AIChE Journal 1982;28(4):588–596.
[48] Chen CC, Evans LB. A local composition model for excess Gibbs energy of aqueous electrolyte
systems. AIChE Journal 1986;32(3):444–454.
[49] de Hemptinne JC, Ledanois JM, Mougin P, Barreau A. Select Thermodynamic Models for Pro-
cess Simulation. Paris: Edition Technip; 2012.
[50] Löffler HJ. Thermodynamik, Band 2: Gemische und chemische Reaktionen. Berlin/Heidelberg:
liq
[51] Kleiber M. The trouble with cp . Ind Eng Chem Res 2003;42:2007–2014.
[52] Soave G. Equilibrium constants from a modified Redlich–Kwong equation of state. Chem Eng
Sc 1972;27:1197–1203.
[53] Plöcker U, Knapp H, Prausnitz JM. Calculation of high-pressure vapor-liquid equilibria from a
corresponding-states correlation with emphasis on asymmetric mixtures. Ind Eng Chem Proc
Des Dev 1978;17(3):324–332.
[54] Enders S. Polymer thermodynamics. In: Gmehling J, Kolbe B, Kleiber M, Rarey J. Chemical
Thermodynamics for Process Simulation. Weinheim: Wiley-VCH; 2012.
[55] Available at: www.ddbst.de.
[56] Available at: www.nist.gov.
[57] Fredenslund Å, Jones RL, Prausnitz JM. Group-contribution estimation of activity coefficients
in nonideal liquid mixtures. AIChE Journal 1975;21(6):1086–1099.
[58] Gmehling J, Li J, Schiller M. A modified UNIFAC model. 2. Present parameter matrix and results
for different thermodynamic properties. Ind Eng Chem Res 1993;32:178–193.
[59] Weidlich U, Gmehling J. A modified UNIFAC model. 1. Prediction of VLE, h E , and γ ∞ . Ind Eng
Chem Res 1987;26:1372–1381.
[60] Available at: www.unifac.org.
[61] Schmid B, Schedemann A, Gmehling J. Extension of the VTPR group contribution equation of
state: Group interaction parameters for 192 group contributions and typical results. Ind Eng
Chem Res 2014;53(8):3393–3405.
[62] Klamt A, Eckert F. COSMO-RS: A novel and efficient method for the a priori prediction of
thermophysical data of liquids. Fluid Phase Equilibria 2000;172:43–72.
[63] Kleiber M, Joh R. Calculation methods for thermophysical properties. VDI Heat Atlas, 2nd ed.,
chap. D1. Berlin/Heidelberg: Springer-Verlag; 2010.
[64] Nannoolal Y, Rarey J, Ramjugernath D. Estimation of pure component properties, Part 2:
Estimation of the saturated liquid viscosity of non-electrolyte organic compounds via group
contributions and group interactions. Fluid Phase Equilibria 2009;281(2):97–119.
[65] Technical University of Denmark DoEE. Heat transfer fluid calculator; Version 2.01, Copy-
right 2000.
[66] Matejovski D. Die Modernität der Industrie und die Ästhetisierung des Ökonomischen.
Priddat BP, West KW, editors, Die Modernität der Industrie. Marburg: Metropolis-Verlag;
2012.
[67] Minton PE. Handbook of Evaporation Technology. Westwood, NJ: Noyes Publications; 1986.
[68] Linnhoff B. Pinch technology training course; 1995. Linnhoff March Ltd., Frankfurt.
[69] Dhole VR, Linnhoff B. Distillation column targets. Comp Chem Eng 1993;27(5/6):549–560.
[70] Toghraei M. Wide design margins do not improve engineering. Hydrocarb Process 2014;93(1):
69–71.
386 | Bibliography
[71] Baehr HD, Stephan K. Wärme- und Stoffübertragung. Berlin/Heidelberg: Springer-Verlag;

1994.
[72] Verein Deutscher Ingenieure. VDI Heat Atlas. Berlin/Heidelberg: Springer-Verlag; 2010.
[73] Frankel M. Facility Piping Systems Handbook, 2nd ed. New York: McGraw-Hill; 2002.
[74] Perea E. Mitigate heat exchanger corrosion with better construction materials. Hydrocarb
Process 2013;92(12):49–51.
[75] Bouhairie S. Selecting baffles in shell-and-tube heat exchangers. Chemical Engineering
Progress 2012;27–33.
[76] HTRI manual; HTRI Xchanger Suite 6.0.
[77] Drögemüller P. The use of hiTRAN wire matrix elements to improve the thermal efficiency of
tubular heat exchangers in single and two-phase flow. Chem-Ing-Tech 2015;87(3):188–202.
[78] Ackermann G. Wärmeübergang und molekulare Stoffübertragung im gleichen Feld bei großen
Temperatur- und Partialdruckdifferenzen. VDI-Forschungsheft 1937;8(382).
[79] Schlünder EU. Film condensation of binary mixtures with and without inert gas. VDI Heat
Atlas, 2nd ed., chap. J2. Berlin/Heidelberg: Springer-Verlag; 2010.
[80] Colburn AP, Drew TB. The condensation of mixed vapors. Trans Am Inst Chem Engrs 1937;33:
197–215.
[81] Arneth S. Dimensionierung und Betriebsverhalten von Naturumlaufverdampfern. Thesis,
TU München, München; 1999.
[82] Arneth S, Stichlmair J. Characteristics of thermosiphon reboilers. Int J Therm Sci 2001;40:
385–391.
[83] Baars A, Delgado A. Non-linear effects in a natural circulation evaporator: Geysering coupled
with manometer oscillations. Heat Mass Transfer 2007;43:427–438.
[84] Dialer K. Die Wärmeübertragung beim Naturumlaufverdampfer. Thesis, ETH Zürich; 1983.
[85] Scholl S, Rinner M. Verdampfung und Kondensation. Goedecke R, editor, Fluid-Verfahrens-
technik. Weinheim: Wiley-VCH; 2011.
[86] Das T. Achieve optimal heat recovery in a kettle exchanger. Hydrocarb Process 2012;91(3):
87–88.
[87] Bethge D. Kurzweg- und Molekulardestillation. Jorisch W, editor, Vakuumtechnik in der
Chemischen Industrie. Weinheim: Wiley-VCH; 1999.
[88] Martin H. Pressure drop and heat transfer in plate heat exchangers. VDI Heat Atlas, 2nd ed.,
chap. N6. Berlin/Heidelberg: Springer-Verlag; 2010.
[89] Gmehling J, Kleiber M, Steinigeweg S. Thermische Verfahrenstechnik. Chemische Technik:
Prozesse und Produkte, 5th ed. Weinheim: Wiley-VCH; 2006.
[90] Schmidt KG. Heat transfer to finned tubes. VDI Heat Atlas, 2nd ed., chap. M1.
Berlin/Heidelberg: Springer-Verlag; 2010.
[91] Müller-Steinhagen H. Fouling of heat exchanger surfaces. VDI Heat Atlas, 2nd ed., chap. C4.
[92] Zhenlu F, Dengfeng L, Xiangling Z, Xing Z. Determine fouling margins in tubular heat ex-
changer design. Hydrocarb Process 2015;94(9):79–82.
[93] Dole RH, Vivekanand S, Sridhar S. Mitigate vibration issues in shell-and-tube heat exchang-
ers. Hydrocarb Process 2015;94(12):57–60.
[94] Gelbe H, Ziada S. Vibration of tube bundles in heat exchangers. VDI Heat Atlas, 2nd ed.,
chap. O2. Berlin/Heidelberg: Springer-Verlag; 2010.
[95] Kister HZ. Distillation -Design-. McGraw-Hill; 1992.
[96] Kister HZ. Distillation -Operation-. McGraw-Hill; 1990.
[97] Sattler K. Thermische Trennverfahren, 2nd ed. Weinheim: VCH Verlagsgesellschaft; 1995.
[98] Stichlmair J, Fair JR. Distillation: Principles and Practice. New York: Wiley-VCH; 1998.
Bibliography | 387
[99] Schultes M. The impact of tower internals on packing performance. Chem-Ing-Tech 2014;
86(5):658–665.
[100] Kister HZ, Mathias P, Steinmeyer DE, Penney WR, Crocker BB, Fair JR. Equipment for distilla-
tion, gas absorption, phase dispersion, and phase separation. Green DW, Perry RH, editors,
Perry’s Chemical Engineers’ Handbook, 8th ed., chap. 14. McGraw-Hill.
[101] Stichlmair J, Bravo JL, Fair JR. General model for prediction of pressure drop and capacity of
countercurrent gas/liquid packed columns. Gas Separation & Purification 1989;3:19–28.
[102] Engel V. Fluiddynamik in Packungskolonnen für Gas-Flüssig-Systeme. Fortschritt-Berichte,
Reihe 3: Verfahrenstechnik. VDI-Verlag; 1999.
[103] Billet R, Schultes M. Prediction of mass transfer columns with dumped and arranged pack-
ings. TransIChemE 1999;77(Part A):498–504.
[104] Spiegel L, Meier W. Structured packings. Chem Plants + Processing 1995;28(1):36–38.
[105] Kister HZ. Distillation -Troubleshooting-. Hoboken, NJ: Wiley-Interscience; 2006.
[106] Kister HZ. Practical distillation technology; 2013. Course Notes.
[107] Bolles WL. Optimum bubble-cap tray design. Part I: Tray dynamics. Petroleum Processing
1956;65–80.
[108] Bolles WL. Optimum bubble-cap tray design. Part II: Design standards. Petroleum Processing
1956;82–95.
[109] Bolles WL. Optimum bubble-cap tray design. Part III: Design technique. Petroleum Processing
1956;72–95.
[110] Bolles WL. Optimum bubble-cap tray design. Part IV: Design example. Petroleum Processing
1956;109–120.
[111] Kister HZ. Effects of design on tray efficiency in commercial towers. Chem Eng Prog 2008;
104(6):39–47.
[112] Stichlmair J. Grundlagen der Dimensionierung des Gas/Flüssigkeit-Kontaktapparates Boden-
kolonne. Weinheim/New York: verlag chemie; 1978.
[113] Rennie J, Evans F. The formation of froths and foams above sieve plates. British Chemical
Engineering 1962;7(7):498–502.
[114] Senger G, Wozny G. Experimentelle Untersuchung von Schaum in Packungskolonnen. Chem-
Ing-Tech 2011;83(4):503–510.
[115] Pahl MH, Franke D. Schaum und Schaumzerstörung – ein Überblick. Chem-Ing-Tech 1995;
67(3):300–312.
[116] Brierly RJP, Whyman PJM, Erskine JB. Flow induced vibration of distillation and absorption
column trays. I. Chem. E. Symp. Ser. 56; 1979. p. 2.4/45–2.4/63.
[117] Priestman GH, Brown DJ. The mechanism of pressure pulsations in sieve-tray columns. Trans
IChemE (London) 1981;59:279–282.
[118] Priestman GH, Brown DJ. Pressure pulsations and weeping at elevated pressures in a small
sieve-tray column. I. Chem. E. Symp. Ser. 104; 1987. p. B407–B422.
[119] Wijn EF. Pulsation of the two-phase layer on trays. I. Chem. E. Symp. Ser. 73; 1982. p. D79–
D101.
[120] Fractionation Research Inc. Causes and prevention of packing fires. Chemical Engineering;
2007, Jul. p. 34–42.
[121] Schuler H. Was behindert den praktischen Einsatz moderner regelungstechnischer Methoden
in der Prozeßindustrie? Automatisierungstechnische Praxis 1992;34(3):116–123.
[122] Friedman YZ, Kane L. Two DCS control configurations: Mass balance and heat balance; 2010.
Webinar, Hydrocarbon Processing.
[123] Sorensen E. Design and operation of batch distillation. Gorak A, Sorensen E, editors, Distilla-
tion: Fundamentals and Principles. Elsevier; 2014.
388 | Bibliography
[124] Brinkmann T, Ebert K, Pingel H, Wenzlaff A, Ohlrogge K. Prozessalternativen durch den Ein-
satz organisch-anorganischer Kompositmembranen für die Dampfpermeation. Chem-Ing-Tech
2004;76(10):1529–1533.
[125] Berascola N, Eisele P. Aspen rate-based distillation; 2011. Seminar, March 21st.
[126] Taylor R, Kooijman HA. Mass transfer in distillation. Gorak A, Sorensen E, editors, Distillation:
Fundamentals and Principles. Elsevier; 2014.
[127] Duncan JB, Toor HL. An experimental study of three component gas diffusion. AIChE J 1962;
8(1):38–41.
[128] Krishna R. Uphill diffusion in multicomponent mixtures. Chem Soc Rev 2015;44:2812.
[129] Schaber K. Aerosolbildung durch spontane Phasenübergänge bei Absorptions- und Konden-
sationsvorgängen. Chem-Ing-Tech 1995;67(11):1443–1452.
[130] Schaber K. Aerosolbildung bei der Absorption und Partialkondensation. Chem-Ing-Tech 1990;
62(10):793–804.
[131] Kaibel G. Distillation columns with vertical partitions. Chem Eng Technol 1987;10:92–98.
[132] Galindez H, Fredenslund Å. Simulation of multicomponent batch distillation processes. Com-
puters and Chem Eng 1988;12(4):281–288.
[133] Hildebrand JH, Scott RL. The solubility of nonelectrolytes. J Phys Coll Chem 1949;53:944–947.
[134] Arlt W. Thermische Grundoperationen der Verfahrenstechnik. Lecture notes, Technical Univer-
sity of Berlin; 1999.
[135] Bocangel J. Design of liquid-liquid gravity separators. Chemical Engineering; 1986, Feb.
p. 133–135.
[136] Henschke M. Dimensionierung liegender Flüssig-Flüssig-Abscheider anhand diskontinuier-
licher Absetzversuche. VDI Fortschritt-Berichte. (Reihe 3; no. 379), Düsseldorf: VDI-Verlag.
[137] Henschke M, Schlieper LH, Pfennig A. Determination of a coalescence parameter from batch-
settling experiments. Chem Eng J 2002;85:369–378.
[138] Pfennig A, Pilhofer T, Schröter J. Flüssig-Flüssig-Extraktion. Goedecke R, editor, Fluid-
Verfahrenstechnik. Weinheim: Wiley-VCH; 2011.
[139] Huang C, Xu T, Zhang Y, Xue Y, Chen G. Application of electrodialysis to the production of
organic acids: State-of-the-art and recent developments. J of Membrane Science 2007;288:
1–12.
[140] Kucera J. Reverse Osmosis – Industrial Applications and Processes. Salem, MA: Scrivener
Publishing; 2010.
[141] Nunes SP, Peinemann KV, editors. Membrane Technology in the Chemical Industry. Wein-
heim/New York: Wiley-VCH; 2001.
[142] Bathen D, Breitbach M. Adsorptionstechnik. Berlin/Heidelberg: Springer-Verlag; 2001.
[143] Bethge D. Energy-saving concepts for the dehydration of alcohol. Zuckerindustrie 2005;
130(3):213–214.
[144] Samant KD, O’Young L. Understanding crystallization and crystallizers. CEP; 2006, Oct.
p. 28–37.
[145] Beckmann W, editor. Crystallization – Basic Concepts and Industrial Application. Weinheim:
Wiley-VCH; 2013.
[146] Mersmann A, Kind M, Stichlmair J. Thermische Verfahrenstechnik, 2nd ed. Springer-Verlag;
2005.
[147] Available at: www.bungartz.de.
[148] Kernan D, Choung E. Run your pumps like a pro: Tips for boosting production and reducing
risk at the refinery. Hydrocarbon Processing Webcast; 2010.
[149] Beitz W, Grote KH, editors. Dubbel – Taschenbuch für den Maschinenbau, 19th ed. Berlin/
Heidelberg: Springer-Verlag; 1997. Ch. B.
Bibliography | 389
[150] Sterling Fluid Systems Group. Liquid vacuum pumps and liquid ring compressors;
Available at: www.sterlingfluidsystems.com.
[151] Deiters UK, Imre AR, Quiñones-Cisneros SE. Isentropen von Fluiden im Zweiphasengebiet.
ProcessNet, Thermodynamic Colloquium. Stuttgart; 2014.
[152] Grave H. Dampfstrahl-Vakuumpumpen. Jorisch W, editor, Vakuumtechnik in der Chemischen
Industrie. Weinheim: Wiley-VCH; 1999.
[153] Baehr HD, Kabelac S. Thermodynamik. Berlin/Heidelberg: Springer-Verlag; 2006.
[154] Jorisch W, editor. Vakuumtechnik in der Chemischen Industrie. Weinheim: Wiley-VCH; 1999.
[155] Vogel HH. Die Finite-Elemente-Methode am Beispiel des Strahlapparates. Chem-Ing-Tech
2006;78(1/2):124–133.
[156] GEA Wiegand GmbH. Überlegungen bei der Projektierung einer Dampfstrahl-Vakuumpumpe.
[157] Toghraei M. Overflow systems are the last line of defense. Hydrocarb Process 2013;92(5):
T92–T94.
[158] Marr R, Moser F, Husung G. Schwerkraft- und Strickabscheider – Berechnung liegender Gas-
Flüssig-Abscheider. verfahrenstechnik 1976;10(1):34–37.
[159] Marr R, Moser F. Die Auslegung von stehenden Gas-Flüssig-Abscheidern – Schwerkraft- und
Gestrickabscheider. verfahrenstechnik 1975;9(8):379–382.
[160] Brauer H. Grundlagen der Einphasen- und Mehrphasenströmungen. Verlag Sauerländer;
1971.
[161] Bürkholz A. Droplet Separation. Weinheim: Wiley-VCH; 1989.
[162] Jess A, Wasserscheid P. Chemical Technology. Weinheim: Wiley-VCH; 2013.
[163] Montebelli A, Tronconi E, Orsenigo C, Ballarini N. Kinetic and modeling study of the ethy-
lene oxychlorination to 1,2-dichloroethane in fluidized-bed reactors. Ind Eng Chem Res 2015;
54(39):9513–9524.
[164] Riedel E. Allgemeine und Anorganische Chemie, 10th ed. Berlin/New York: Walter de Gruyter;
2010.
[165] Macejko B. Is your plant vulnerable to brittle fracture? Hydrocarb Process 2014;93(11):67–78.
[166] Sims JR. Improve evaluation of brittle-fracture resistance for vessels. Hydrocarb Process 2013;
92(1):59–62.
[167] Rähse W. Praktische Hinweise zur Wahl des Werkstoffs von Maschinen und Apparaten.
Chem-Ing-Tech 2014;86(8):1163–1179.
[168] Wagner MH. Heat transfer to non-newtonian fluids. VDI Heat Atlas, 2nd ed., chap. M4.
[169] Truckenbrodt E. Lehrbuch der angewandten Fluidmechanik. Berlin/Heidelberg/New York/
Tokyo: Springer-Verlag; 1983.
[170] Lockhart RW, Martinelli RC. Proposed correlation of data for isothermal two-phase, two-
component flow in pipes. Chem Eng Prog 1949;45(1):39–48.
[171] Friedel L. Druckabfall bei der Strömung von Gas/Dampf-Flüssigkeits-Gemischen in Rohren.
Chem-Ing-Tech 1978;50(3):167–180.
[172] Friedel L. Eine dimensionslose Beziehung für den Reibungsdruckabfall bei Zweiphasen-
rohrströmung zwischen Wasser und R12. vt Verfahrenstechnik 1979;13(4):241–246.
[173] Friedel L. Improved friction pressure drop correlations for horizontal and vertical two phase
pipe flow. 3R international 1979;18(7):485–491.
[174] Beggs HD, Brill JP. A study of two-phase flow in inclined pipes. J Petrol Technol 1973;607–617.
[175] Muschelknautz S. Druckverlust in Rohren und Rohrkrümmern bei Gas-Flüssigkeit-Strömung.
VDI-Wärmeatlas, 8th ed., chap. Lgb. Berlin/Heidelberg: Springer-Verlag; 1997.
[176] Schmidt H. Two-phase gas-liquid flow. VDI Heat Atlas, 2nd ed., chap. L2. Berlin/Heidelberg:
390 | Bibliography
[177] Lee S, Seok W. Major accident and failure of stationary equipment in the RFCCU. Hydrocarb
Process 2016;95(1):65–70.
[178] Sahoo T. Pick the right valve. Chemical Engineering; 2004, Aug. p. 34–39.
[179] Nitsche M. Industriearmaturen. Chemie-Technik 1985;14(3):99–102.
[180] Johnson RD, Lee B. Valve design reduces costs and increases safety for US refineries. Hydro-
carb Process 2010;89(8):37–40.
[181] Stepanek D. Was den Betreiber von Massedurchflussmessern nach dem CORIOLIS-Prinzip
interessiert. Tech. rep., Schwing Verfahrenstechnik GmbH; 2004. Corporate Publication.
[182] Ignatowitz E. Chemietechnik, 7th ed. Haan-Gruiten: Verlag Europa-Lehrmittel; 2003.
[183] Wagner W, Kruse A. Properties of Water and Steam. Berlin/Heidelberg/New York: Springer-
Verlag; 1998.
[184] Numrich R, Müller J. Filmwise condensation of pure vapors. VDI Heat Atlas, 2nd ed., chap. J1.
[185] Spirax Sarco GmbH. Grundlagen der Dampf- und Kondensattechnologie; 2014. Available at:
www.spiraxsarco.de.
[186] Available at: www.spiraxsarco.com/Resources/Pages/steam-engineering-tutorials.aspx.
[187] Spirax Sarco GmbH. Kavitation ade! CALORIE 2011;79:10–11.
[188] Glück A, Hunold D. Oil-based and synthetic heat transfer media. VDI Heat Atlas, 2nd ed.,
chap. D4.3. Berlin/Heidelberg: Springer-Verlag; 2010.
[189] Krakat G. Cryostatic bath fluids, aqueous solutions, and glycols. VDI Heat Atlas, 2nd ed.,
chap. D4.2. Berlin/Heidelberg: Springer-Verlag; 2010.
[190] Kleiber M. Exhaust air treatment in chemical industry. Gierycz P, Malanowski SK, editors,
Thermodynamics for Environment. Warszawa: Information Processing Centre; 2004.
[191] Bundesministerium für Umwelt NuR. Technische Anleitung zur Reinhaltung der Luft. Köln:
Carl Heymanns Verlag KG; 2002.
[192] Herzog F, Schulte M. Abluftreinigung durch Kryokondensation. UMWELT 1998;1/2:49–53.
[193] Messer Group GmbH. The DuoCondex process; Available at: www.messergroup.com/the-
duocondex-process.
[194] Domschke T, Steinebrunner K, Christill M, Seifert H. Verbrennung chlorierter Kohlenwasser-
stoffe – Die Deacon-Reaktion in Rauchgasen während der Abkühlung. Chem-Ing-Tech 1996;
68(5):575–579.
[195] Görner K, Hübner K. Gasreinigung und Luftreinhaltung. Berlin/Heidelberg/New York:
[196] Kugeler K, Phlippen PW. Energietechnik. Berlin/Heidelberg/New York: Springer-Verlag; 1993.
[197] Kolar J. Stickstoffoxide und Luftreinhaltung. Berlin/Heidelberg/New York: Springer-Verlag;
1990.
[198] Hevia MAG, Perez-Ramirez J. Assessment of the low-temperature EnviNOx variant for catalytic
N2 O abatement over steam-activated FeZSM-5. Appl Catal B 2008;77(3/4):248–254.
[199] Schwefer M, Siefert R, Groves MCE, Maurer R. Verfahren zur gemeinsamen Beseitigung von
N2 O und NOx – Erste großtechnische Installation im Abgas der HNO3 -Produktion. Chem-Ing
Tech 2003;75(8):1048–1049.
[200] Venkatesh M, Woodhull J. Pick the right thermal oxidizer. Chemical Engineering 2003;67–70.
[201] Müller G. Absorption organischer Lösemittel mit Glykolethern. VDI-Berichte 1989;730:
373–394.
[202] Bay K, Wanko H, Ulrich J. Biodiesel – Hoch siedendes Absorbens für die Gasreinigung.
Chem-Ing-Tech 2004;76(3):328–333.
[203] Available at: www.desotec.com.
[204] Sörensen M, Zegenhagen F, Weckenmann J. State of the art wastewater treatment in pharma-
ceutical and chemical industry by advanced oxidation. Pharm Ind 2015;77(4):594–607.
Bibliography | 391
[205] Available at: www.enviolet.com.

[206] Onken U, Behr A. Chemische Prozesskunde. Stuttgart: Georg Thieme Verlag; 1996.
[207] Available at: https://en.wikipedia.org/wiki/West_Fertilizer_Company_explosion.
[208] Available at: https://en.wikipedia.org/wiki/2015_Tianjin_explosions.
[209] Nolan DP. Application of HAZOP and What-If-Safety Reviews to the Petroleum, Petrochemical
and Chemical Industries. Park Ridge, NJ: Noyes Publications; 1994.
[210] Stephan D. Sicher ist Sicher: Warum SIL keine Pflicht, aber trotzdem ein Muss ist;
Available at: www.process.vogel.de/sicherheit/articles/483503/.
[211] Bozoki G. Überdrucksicherungen für Behälter und Rohrleitungen. Verlag TÜV Rheinland
GmbH; 1977.
[212] Renfro J, Stephenson G, Marques-Riquelme E, Vandu C. Use dynamic models when designing
high-pressure vessels. Hydrocarb Process 2014;93(5):71–76.
[213] Feuerstein A. Dynamische Berechnung von Abblasevorgängen. master’s thesis, TU Darmstadt;
2015.
[214] Venting Atmospheric and Low-Pressure Storage Tanks. American Petroleum Institute, 5th ed.;
1998. API Standard 2000.
[215] Pressure relieving and depressuring systems. American Petroleum Institute, 5th ed.; 2007.
ANSI/API Standard 521.
[216] LESER. Engineering handbook; Available at: www.leser.com/en/tools/engineering.html.
[217] Yeh G, Griman J, Najrani M. Recover from a steam reformer tube rupture. Hydrocarb Process
2013;92(6):85–88.
[218] Elliott B. Using DIERS two-phase equations to estimate tube rupture flowrates. Hydrocarb
Process 2001;8:49–54.
[219] Leung JC, Grolmes MA. A generalized correlation for flashing choked flow of initially sub-
cooled liquid. AIChE J 1988;34(4):688–691.
[220] Leung JC. A generalized correlation for one-component homogeneous equilibrium flashing
choked flow. AIChE J 1986;32(10):1743–1746.
[221] Staak D, Repke JU, Wozny G. Simulation von Entlastungsvorgängen bei Rektifikations-
kolonnen. Chem-Ing-Tech 2008;80(1/2):129–135.
[222] Smith D, Burgess J. Relief valve and flare action items: What plant engineers should know.
Hydrocarb Process 2012;91(11):41–46.
[223] ISO 4126. Safety devices for protection against excessive pressure. Beuth Verlag, Berlin;
2010.
[224] Schmidt J, Westphal F. Praxisbezogenes Vorgehen bei der Auslegung von Sicherheitsventilen
und deren Abblaseleitungen für die Durchströmung mit Gas/Dampf-Flüssigkeitsgemischen –
Teil 1. Chem-Ing-Tech 1997;69(3):312–319.
[225] Fründt J. Untersuchungen zum Einfluß der Flüssigkeitsviskosität auf die Druckentlastung.
Aachen: Shaker-Verlag; 1997. Thesis.
[226] Schmidt J, Westphal F. Praxisbezogenes Vorgehen bei der Auslegung von Sicherheitsventilen
und deren Abblaseleitungen für die Durchströmung mit Gas/Dampf-Flüssigkeitsgemischen –
Teil 2. Chem-Ing-Tech 1997;69(8):1074–1091.
[227] Brodhagen A, Schmidt F. Berechnen von kritischen Massenströmen. VDI-Wärmeatlas,
10th ed., chap. Lbd. Berlin/Heidelberg: Springer-Verlag; 2006.
[228] Sizing, Selection, and Installation of Pressure-Relieving Devices in Refineries. American
Petroleum Institute, 7th ed.; 2000. API Recommended Practice 520.
[229] Bauerfeind K, Friedel L. Berechnung der dissipationsbehafteten kritischen Düsenströmung
realer Gase. Forschung im Ingenieurwesen 2003;67(6):227–235.
[230] Westphal F, Christ M. Erfahrungen aus der Praxis mit dem 3 %-Kriterium für die Zuleitung von
Sicherheitsventilen. Technische Sicherheit 2014;4(3):28–31.
392 | Bibliography
[231] LESER. Chattering safety valve; Available at: www.leser.com/en/news-about-leser/media-

center/videos.html.
[232] Klapötke TM. Chemistry of High-Energy Materials. Berlin/New York: Walter de Gruyter; 2011.
[233] Pfenning D. Inertisieren im Sekundentakt; 2014. Presentation, FH Aachen.
[234] Thess A. The Entropy Principle. Berlin/Heidelberg: Springer-Verlag; 2010.
[235] Kittredge CP, Rowley DS. Resistance coefficients for laminar and turbulent flow through one-
half-inch valves and fittings. Trans ASME 1957;79:1759–1766.
[236] Gersten K. Einführung in die Strömungsmechanik, 3rd ed. Braunschweig: Vieweg; 1984.
[237] Kast W. Druckverlust bei der Strömung durch Leitungen mit Querschnittsänderungen. VDI-
Wärmeatlas, 8th ed. Berlin/Heidelberg: Springer-Verlag; 1997. Abschnitt Lc.
A Some numbers to remember
It seems to be a bit outdated to know simple numbers by heart. Nevertheless, many projects are
founded by having quick ideas in the open talk with colleagues, a plant manager or a plant engineer.
Often complicated process simulations must be made plausible to practitioners by rough calculations.
Knowing some often used numbers makes you a candidate for the pole position in process engineering
meetings.
Without claiming to be complete, here are some numbers considered to be worth to know them
by heart. On purpose, only the even approximations and not the exact numbers are given, as the target
of learning numbers by heart is their application without any tools.
Molecular weights
Nitrogen 28 g/mol
Air 29 g/mol
Water 18 g/mol
Chlorine 71 g/mol
Methanol 32 g/mol
Ethylene 28 g/mol
Oxygen 32 g/mol
Propylene 42 g/mol
Hydrogen 2 g/mol
Ammonia 17 g/mol
Methane 16 g/mol
Carbon Dioxide 44 g/mol
Ethanol 46 g/mol
Standard cubic meter
Essentially, the standard cubic meter is not a volume but a mass unit. It refers to the amount of gaseous
substance in 1 m3 at standard conditions p = 1.01325 bar, t = 0 °C. It can be calculated with the ideal
gas equation of state. For nitrogen with M = 28.013 g/mol, one gets
pVM 101325 Pa ⋅ 1 m3 ⋅ 0.028013 kg/mol

mN = = = 1.2498 kg ≈ 1.25 kg.
RT 8.31447 J/(mol ⋅ K) ⋅ 273.15 K
It is easy to perform this calculation for other substances as well, but undoubtedly, it is difficult to
perform without at least a pocket calculator. However, the only number that refers to the substance
in this calculation is the molecular weight. Thus, knowing the even number for nitrogen by heart, the
mass of a standard cubic meter can easily be scaled with the molecular weight, e.g.
– for air (M = 29 g/mol): m N,air = 1.25 kg ⋅ 29
28 = 1.295 kg;
– for hydrogen (M = 2 g/mol): m N,H2 = 1.25 kg ⋅ 28 2
= 89 g;
– for oxygen (M = 32 g/mol): m N,O2 = 1.25 kg ⋅ 28 = 1.43 kg;
32
– for carbon dioxide (M = 44 g/mol): m N,CO2 = 1.25 kg ⋅ 44

28 = 1.96 kg.
394 | A Some numbers to remember
Some other useful physical property data
cid
p Nitrogen 1 J/g K
cLp Water 4.2 J/g K
cid
p Water 1.9 J/gK
∆h v Water 2250 J/g (at t = 100 °C)
ρL Water 1000 kg/m3
κ Nitrogen 1.4
cp Steel 0.4–0.5 J/g K
ρ Steel 7800 kg/m3
λ Carbon Steel 50 W/mK
λ Stainless Steel 15 W/mK
Critical temperatures
Methanol 240 °C
Ethanol 241 °C
Ethylene 9 °C
Propylene 91 °C
Propane 97 °C
Nitrogen −147 °C
Ammonia 132 °C
Methane −83 °C
Water 374 °C
Carbon Dioxide 31 °C
Normal boiling points
Methanol 64 °C
Acetone 56 °C
Benzene 80 °C
Toluene 111 °C
Ethylene −104 °C
Propylene −48 °C
Propane −42 °C
Chlorine −34 °C
Water 100 °C (In fact, the exact value is 99.97 °C according to ITS-90.)
Nitrogen −196 °C
Ammonia −33 °C
Methane −161 °C
Carbon Dioxide None The triple point is at t = −56.5 °C, p = 5.2 bar.
Ethanol 78 °C
A Some numbers to remember | 395
Rough values for the vapor pressure of water
30 °C 0.04 bar
40 °C 0.075 bar
60 °C 0.2 bar
70 °C 0.3 bar
80 °C 0.5 bar
90 °C 0.7 bar
100 °C 1 bar
120 °C 2 bar
150 °C 5 bar
160 °C 6 bar
170 °C 8 bar
180 °C 10 bar
190 °C 12.5 bar
210 °C 19 bar
230 °C 28 bar
250 °C 40 bar
Some values for heat transfer
Heat transfer by natural convection to air α = 4–5 W/m2 K

Heat transfer by wind α = 10 W/m2 K
Plate heat exchanger liquid/liquid k = 2000 W/m2 K
Maximum possible solar radiation (solar constant) S = 1367 W/m2
B Pressure drop coefficients
For the evaluation of the ζ-values in Equation (12.18), the following instructions according to [149] can
be applied. In the tables, one can interpolate between the given values. If the cross-flow area changes,
the velocity always refers to the outlet of the element, i.e. to the large cross-flow area for expansions
and to the small cross-flow area for restrictions. Pictures of the particular elements can be found in
[149]. For laminar flow, the ζ-values listed here cannot be used. For small Reynolds numbers, they can
be up to 1000 fold higher. The problem is described in [235].
90° bend
r/d 1 2 4 6 10
ζ90 , smooth (k Re < 65 d) 0.21 0.14 0.11 0.09 0.11
ζ90 , rough (k Re > 65 d) 0.51 0.3 0.23 0.18 0.20
Bend with arbitrary angles ϕ ≠ 90°
ϕ 0° 30° 60° 90° 120° 150° 180°

ζ/ζ90 0 0.4 0.7 1.0 1.25 1.5 1.7
Elbow pipe with circular cross-flow area
ϕ 0° 22.5° 30° 45° 60° 90°

ζsmooth 0 0.07 0.11 0.24 0.47 1.13
ζrough 0 0.11 0.17 0.32 0.68 1.27
Elbow pipe with rectangular cross-flow area
ϕ 0° 30° 45° 60° 75° 90°

ζ 0 0.15 0.52 1.08 1.48 1.6
Corrugated expansion joint
ζ = 0.2n with n as the number of corrugations

398 | B Pressure drop coefficients
U-bend
a/d 0 2 5 10
ζ 0.33 0.21 0.21 0.21
a is the length of the straight lines [149].
Sharp-edged tube entrance
ζ = 0.5
Smooth tube entrance
– ζ = 0.01 (smooth)
– ζ = 0.03 (transition smooth–rough)
– ζ = 0.05 (rough)
Tube entrance with orifice
(d/dorifice )2 1 1.25 2 5 10
ζ 0.5 1.17 5.45 54 245
Discontinuous transition from A1 to A2 > A1
ζ = (A2 /A1 − 1)2
Continuous cross-flow area expansion (diffusor)
ϕ/2 4° 6° 8° 10° 12°

ϕ 8° 12° 16° 20° 24°
ζ(d2 /d1 = 1.2) 0.0 0.0 0.0 0.1 0.2
ζ(d2 /d1 = 1.4) 0.0 0.15 0.25 0.3 0.7
ζ(d2 /d1 = 1.6) 0.25 0.6 0.85 1.05 1.65
ζ(d2 /d1 = 1.8) 0.8 1.15 1.75 2.15 3.1
ζ(d2 /d1 = 2.0) 1.25 2.0 2.75 3.5 5.0
The values have been taken from diagrams in [149]. ϕ is the opening angle of the diffusor. The cross-
flow area expansion is characterized by the diameter at the inlet of the expansion piece d1 , the diam-
eter at the outlet of the expansion piece d2 and its length, which yields the opening angle ϕ. Because
B Pressure drop coefficients | 399
of the extremely steep gradients, the extrapolation beyond 24° should be omitted. Instead, the dis-
continuous transition should be used.
Discontinuous cross-flow area restriction from A1 to A2 < A1
The relationship between the area ratio (A2 /A1 ) and ζ can be evaluated according to the relationship
in [169], which reproduces the curve in [149] and the tabulated values in [236] sufficiently well:
1−μ 2
ζ = 1.5 ( ) , (B.1)
μ
with the restriction coefficient

0.39309023 (A2 /A1 )2 − 0.86544149 (A2 /A1 ) + 0.61790739
μ= . (B.2)
1 − 1.4837414 (A2 /A1 ) + 0.62929722 (A2 /A1 )2
Continuous cross-flow area restriction
The pressure drop is comparably low and can be described by
ζ = 0.05
with a sufficient accuracy and in a conservative way [149]. If the angle of the restriction is > 40°, the
formula for the discontinuous restriction should be applied. For very small angles, the pressure drop
of the pipe itself should not be neglected [237].
Index
γ-φ-approach 25, 47, 63, 64 aerobic process 326

ω-method 351, 358, 366, 377 aerosol 186, 312, 330, 373
φ-φ-approach 25, 47, 49, 63, 64 agitator 247, 253, 266
ζ-value 280, 294, 397 agro chemicals 89
10 % pressure drop criterion 364 air cooler 133, 308
1st generation random packing 146 air line 369
2-phase flash 61 amine 300
2nd generation random packing 146 ammonia 300, 315
3-phase flash 61 ammonia reaction 258, 260
3 % pressure drop criterion 363 ammonium nitrate 327–330
3rd generation random packing 146 ammonium sulfate 327
4th generation random packing 146 annular flow 286, 288
antifoaming agent 171
abnormal heat input 352 API-2000 343
abrasive effect 127 API-521 343
absolute pressure 296 apparent component approach 58
absorption 139, 185, 186, 310, 319 apron 162, 167
absorption column 153 aqueous phase 191
absorption/desorption unit 319 aqueous system 148, 149, 173
absorptive agent 319 area plot plan 11
acentric factor 36 asymmetric rotating disc contactor (ARDC) 200
acetylene 330 auto-refrigeration case 271
activated carbon 211, 314, 325 autoignition temperature 367
activation energy 257 axial turbo compressor 235
active area 160, 161, 167, 373 azeotrope 30, 45, 195, 266, 373
active fire protection 15 azeotropic condition 46
activity 59 azeotropic distillation 181
activity coefficient VII, VIII, 26, 30, 41, 45, 58, azeotropic point 30
373
activity coefficient at infinite dilution 73 back mixing 198
actuation case 347 back pressure 338, 364
actuation pressure 337 backward feed arrangement 91
adiabatic throttling 81, 359 baffle 115, 136
adsorbate 373 baffle cut 115
adsorbent 373 baffle orientation 115
adsorption 140, 183, 210, 310, 314, 322, 325 baffle separator 253
adsorption capacity 210 baffle spacing 116, 138
adsorption equilibrium 212 bagatelle limit 309
adsorption isotherm type 213 ball valve 291, 294
adsorption twin plant 214 BASF 327
adsorptive 373 basic chemicals 89
adsorptive agent 211 basic engineering 4, 5, 7, 353
advanced cubic equation of state 70, 74 batch distillation 188
advanced oxidation processes (AOP) 325 batch process 89, 97
advanced process control 373 batch process recipe 98
402 | Index
batch reactor 263 Carnival Monday accident 329, 349

batch simulation 99 cartridge column 160, 172
batchwise 308 catalyst 261
battery limit 6, 373 catalyst exchange 267
Bayer-Flachglocke 158 catalytic combustion 310, 315, 316
bellow 365 cause & effect matrix 16, 332, 373
bellow expansion joint 290, 291 cavitation 5, 225
BEM 121 cell method 110
Bernoulli equation 223 centrifugal extractor 201
Bhopal 329 centrifugal pump 223, 229
Billet and Schultes model 152 certified capacity 356
binary interaction parameter (BIP) 44, 71, 72 chaotic flow 286
binary parameter estimation 72 chattering 357, 365
biochemical product 189 check valve 6, 293, 373
biodiesel 319 chemical oxygen demand (COD) 324
biofilter 321 chlorine 313
biogas 326 choking 281
biological exhaust air treatment 310 chromatography 215
biological oxygen demand (BOD) 324 circulation pump 225
biological waste water treatment 326 Clausius–Clapeyron equation 29, 49, 68
bioscrubber 321 clear liquid height 166
biotrickling filter 321 closed balance point 158
Bitterfeld 328 coalescence 192, 193
BJ21T 121 coalescer 373
Blasius equation 277 COD 373
blocked discharge line 347 coil module 205
blowdown 336 cold water aggregate 308
blowdown process 341 Colebrook 277
BOD 373 collector 153, 172
boiler feed water 305, 373 collector/distributor unit 144
brainstorming 373 combinatorial part 73
brine 308 combined heat and mass transfer 121
brittle 271 combustion 310, 313
Broyden method 86 compact approach 143
bubble cap tray 157 composite curve 94
bubble column 319 composite membrane 205
bubble flow 286, 287 composition control 174
bubble point curve 29 compressibility factor 53, 373
bubble regime 154 compressible fluid 280
built-up back pressure 338 compressor 13, 81, 231, 278
butterfly valve 292 conceptual design phase 2, 3, 173
by-product 21, 82, 373 condensate 299, 304
condensate lifter 304
c-concentration 257, 259 condensate line 305
calcium carbide 330 condensation 139
CAPEX 1, 18 conservative assumption 356
capillary module 205 contingency 373
car sealed valve 293 continuity equation 360
carboxylic acid 53, 69 continuous reactor 263
Index | 403
control 292 demister 251

control cycle 332 DeNOx 315
control engineering 174 denoxation 318
control valve 223, 291, 293, 333 density 111, 149, 150
control valve failure 352 depreciation 374
convergence 143 depressurization 309, 336
conversion 255 design basis 4, 5, 270, 301, 343, 374
cooling brine 110 design mode 111
cooling system failure 352 design pressure 270, 306, 342, 345, 347, 348,
cooling water 95, 307, 374 351, 374
coproduct 21, 82, 374 design temperature 270, 342, 374
Coriolis flow meter 296 desuperheating 299
corresponding states principle 36, 37 detail engineering 6, 7, 353
corrosion 157, 240, 273, 303, 307, 318, 320, detailed approach 143
333, 339 detonation 367
corrosion allowance 269 deviation 374
corrosivity 138 dew point curve 29
COSTALD equation 49 diffusion 121, 186, 204, 312
coverage 127 diffusion coefficient 185, 197, 204, 210
critical flow phenomenon 358 dimethylether 328
critical mass flux density 359 dioxine 328
critical point 34, 49, 51, 67, 344 dip-pipe 13, 247, 374
critical pressure 34 direct steam 300
critical temperature 34 direct substitution method 84
crossflow filtration 206 dischargeable mass flow 356, 358, 363, 364
crud 193 discontinuous mode (batch) 263
cryo-condensation 310–313 disk-and-donut baffle 116
Cryosolv process 312, 313 disperse phase 195
crystal growth 218 dispersion 191
crystallization 61, 217 distillation 76, 139
CSTR cascade 265 distillation column control 173
cubic equation of state 34 distributor 146, 148, 153, 172
cyclohexane 328 dividing wall column 186
cyclone 253 double azeotrope 32
double jeopardy 333, 353, 374
data management 68 double pipe 132, 272, 290
data sheet 332 double-segmental baffle 116, 137
databank 71 downcomer 166, 171
Deacon reaction 314, 319 downcomer backup flooding 167, 169
dead-end filtration 206 downcomer choke flooding 166, 169
deaeration problem 122 downcomer clearance 162, 167
Debye–Hückel term 58 downcomer cross-flow area 167
decanter 58 draft tube baffle crystallizer (DTB) 219, 220
decomposition 160 drain 290, 294
dedicated plant 97 driving pressure difference 362
deflagration 367 driving temperature difference 91, 92, 123, 128,
degradation 172, 262, 315 131, 238, 303, 352
delivery head 223, 224 droplet formation 239
demineralized water 306, 308 droplet size 239
404 | Index
dry pressure drop 153 eutectic system 62, 217

dualflow tray 160 evaporation 139, 325
ductile 271 excess enthalpy 48, 65, 67, 73, 374
DuoCondex process 313 excess heat capacity 73
durability 206 excess volume 49, 374
dynamic simulation 2, 79, 341, 342, 346, 352, exergy analysis 97
355 exhaust air 100, 308
dynamic viscosity 75, 77, 111 exhaust air treatment 308, 313
exhaust air treatment with membranes 321
economy-of-scale 19 exothermic reaction 263
effect of temperature on a chemical expansion loop 290
reaction 260 explosion 336, 366, 368
efficiency 142 explosion limit 368
electrodialysis 210 explosion zone 367
electrolyte 46, 55, 56, 69 extended Antoine equation 51
electrolyte model 69 external heat exchanger 254
electrolyte NRTL model 57 external reactor 268
electronegativity 56 extraction 76, 194
elementary charge 56 extraction column 197
ellipsoidal head 248 extractive 194
energy balanced configuration 177 extractive distillation 182
Engel model 152 extrapolation function 51
enthalpy 374
enthalpy of adsorption 212 F-factor 150
enthalpy of fusion 63 failsafe closed (FC) 294
enthalpy of mixing 48 failsafe open (FO) 294
enthalpy of reaction 65, 261, 266 falling film crystallizer 219
enthalpy of vaporization 27, 29, 49, 65, 67, 70, falling film evaporator 126, 127, 131
344 fan 231
entrainment 128, 129, 156, 158, 163, 164, 168, filtration 323
169 fin 133
entropy 374 fine chemicals 89, 97, 189, 308
equation of state 374 fire 333
equilibrium 263 fire case 342, 347
equilibrium calculation 142 fire point 367
equilibrium constant 258, 261 fire prevention 15
equilibrium reaction 258 fixed bed reactor 262
equilibrium reactor 268 fixed costs 18, 374
equipment arrangement drawing 11 fixed valve tray 158
equipment content visualization 100 flame ionization detector (FID) 368, 369
equipment diagram 99 flammability diagram 369
error message 143 flare load 355
estimation of vapor pressures 52 flare system 336
ethanol 181, 215 flash chamber 180
ethanol–water 181, 216 flash fire 367
ethylene 70, 368 flash point 367
ethylene glycol 183 flashing 366
ethylene oxide 368 flood point 148, 150, 152, 153
eutectic 73 Flory–Huggins 70
Index | 405
flotation 323 glycol ether 319

flow 332 grand composite curve 95
flow path length 162, 163 Grashof number 375
flow pattern 286 grid baffle 117
fluid code 289 group contribution method 27, 72
fluidelastic instability 136, 138 guideword 332, 375
flushing 309
foam factor 170 half-pipe coil 254
foaming 126, 127, 145, 160, 170, 172, 357 hardness component 133, 134, 218, 307
forced circulation crystallizer (FC) 219, 220 hastelloy 306
forced circulation evaporator 131 HAZID 3
forced circulation reboiler 126, 127 HAZOP 16, 333, 375
formal kinetics 257 HAZOP analysis 330
forward feed arrangement 91 HAZOP leader 331, 332
fouling 118, 119, 123, 127, 128, 131, 133–135, HCl 186
141, 145, 148, 156–158, 160, 172, 174, 240, heat adsorption 310
251, 302, 307, 339, 374 heat capacity 55, 70
fouling factor 4, 102, 135 heat curve 110
fouling layer 135 heat exchanger 13, 76, 102, 107
fouling resistance 135 heat exchanger block 80
fractional hole area 156, 158, 162 heat integration 90, 95, 174, 303, 325
free cross-flow area 361 heat of formation 27
frequency converter 228, 238 heat of reaction 263
friction factor 276, 278 heat of vaporization 26, 27
friction pressure drop 81 heat pump 95
Friedel equation 283, 285 heat transfer 266
froth 154 heat transfer oil 306
froth height 163 heat transition 108
froth regime 154 heater block 80
Froude number 285, 374 height of inlet weir 162
fuel gas 375 height of outlet weir 162
fuel NO 314 height over the weir 168
fugacity 35, 258 helium 77
fugacity coefficient 35, 41, 375 Henry coefficient 47, 347
Henry component 47
Gantt Chart 100 Henry concept 69
gas chromatography (GC) 174, 298 Henry’s law 64
gas permeation 209 Henschke model 194
gas solubility 64, 347 heteroazeotrope 32, 59, 181
gas-liquid reaction 265 heterogeneous reaction 265
gasket 131, 289 HETP 148, 184, 375
gate valve 291, 294 high performance liquid chromatography
gauge pressure 296 (HPLC) 298
g E mixing rule 48, 59, 63, 70, 74 high vacuum 129, 242
gear pump 230 high-precision equation of state 70
geysering 123 Hoechst AG 329, 330
Gibbs–Duhem equation 45 Hoffmann–Florin equation 52
GIGO principle 82 holdup 150, 354, 375
globe valve 291 hollow fibre module 205
406 | Index
hollow stirrer 266 kettle reboiler 127

hydrodynamics 141, 149, 152 Kister–Gill correlation 152
hydrogen 77, 272 knock-out drum 253
hydrogen fluoride (HF) 53, 69 Kühni column 199
hydrogen peroxide 325 KV -value 294
hysteresis 338
laminar flow 276
ideal gas 34, 65, 360 Langmuir approach 212
ideal gas equation 32 laughing gas 315
ideal gas heat capacity 65 Laval nozzle 240, 280
ideally mixed batch stirred tank reactor 264
law of mass action 258
ideally mixed continuous stirred tank reactor
layer crystallizer 219
(CSTR) 264
layout pattern 137
ignition 366, 368
LDPE 70, 361
ignition source 318
Le Chatelier mixing rule for LEL 368
impingement plate 121
Le Chatelier’s principle 258, 259
inclination 121
leak before burst 269
inclusion 217
leakage 122, 131
individual α-function 63
leakage rate 246
induced ignition 366
Lee–Kesler–Plöcker 70
industrial fan 238
level 333
inert gas 121
level alarm 248
inert gas supply 371
level control 124
infinite dilution 47
level indicator 297
influence of the pressure on a distillation
level switch 249, 297
column 141
lever rule 62, 375
inlet weir 161
limiting activity coefficient 27
inner surface 211, 212
limiting liquid level 358
insulation 289, 306
liquid density 27, 48
interlock 5, 375
liquid heat capacity 27, 65
interlock description 332
liquid load 147, 150
internal energy 375
liquid nitrogen 311, 312
internal rate of return 23
liquid ring compressor 236, 244
inverted batch column 189
liquid seal 167
investment 18
liquid-liquid equilibrium 58, 73, 195
ISBL 375
liquid-liquid separator 191
isentropic change of state 360
liquidus line 62, 375
isentropic efficiency 232
liquiphant 298
isentropic expansion 361
LLE 104
iso-activity criterion 59
load data list 16
isolation valve 291
load diagram 163
jacket 254 load point 152
jacket water 308 LOC 369
jet flood 164, 169 local composition 42
jet pump 92, 240, 244 local composition model 44
Joule–Thomson-effect 371 locked valve 293
lower explosion limit (LEL) 368
Karl-Fischer-titration 298 lubricant 292
Katapak 267, 268 lubrication oil 244
Index | 407
Ludwigshafen 328 minimum downcomer residence time 166

Ludwigshafen-Oppau 327 minimum liquid load 148
minimum safety distance 13
Mach number 280, 282, 375 minimum vapor load 168
magnetic flow meter 297 minimum vapor velocity 169
makespan 101, 102 miniplant 197
maldistribution 146, 148, 149, 153, 163 miscibility gap 32, 46, 58, 195
man hole 160, 247 mixer 81
mass balance 2 mixer-settler arrangement 197
mass balanced configuration 177 mixing rule 47, 49, 76, 77
mass flow to be discharged 340, 348, 356–358, model change 75, 104
363, 364 model choice 68
mass transfer 186, 265 Modified UNIFAC 72–74
material 272, 289, 306, 320 modularization 14
material balance 332 molecular sieve 183, 210, 211, 216
material test 273 Monday fire 323
materials of construction 271 Moody diagram 276
matrix project management 17 motive steam 93
maximum allowable overpressure 337, 338 multieffect evaporation 90, 325
maximum allowable pressure 341 multipass tray 168
maximum downcomer velocity 166 multiple condensers 143
maximum flux density 360 multiplier 82
maximum mass flux density 246, 377 multipurpose plant 263
maximum operating pressure 338 multipurpose unit 97
maximum relief amount 358 Murphree efficiency 162, 171
Maxwell criterion 34
measurement 275
mechanical efficiency 232 nanofiltration 203, 323
mechanical foam deletion 171 narrowest cross-flow area 359, 366
mechanical stability 289 natural circulation 122
mechanical strength 4 natural frequency 138
mechanical tension 269 net positive suction head (NPSH) value 225, 247
mechanical vapor recompression (MVR) 92, 237 net present value 22
medium vacuum 242 Newton method 85
melt crystallization 219 Newton number 266
melting temperature 63 Nikuradse 277
membrane 183, 203, 325 no-tubes-in-window baffle (NTIW) 117, 137
membrane compressor 233 node 332, 375
membrane process 310 noise 133
membrane pump 229, 230 non-Newtonian flow 275
membrane valve 292 non-reasonable risk 333
MESH equation 142, 190 normal boiling point 27
methyl isocyanate 329 normal operating condition 270
Michaelis–Menten kinetics 261 notched weir 168
microfiltration 203, 323 nozzle 138
micropore 211 nozzle size 249
middle vessel 189 NPSH value 375
minimum bypass 227, 347 NRTL 42, 44, 45, 59, 63
minimum design metal temperature (MDMT) 271 NRTL electrolyte model 58
408 | Index
Nußelt number 375 pipe module 205

nutritient 320 piping 13, 15, 275
piping and instrumentation diagram (PID) 6, 15,
o-nitroanisol 330 247, 332, 375
Ohm’s law 107 piping class 15
oil diffusion pump 244 piping element 223, 286
open balance point 158 piping isometric drawing 11
operation costs 18 piping layout drawing 11
operational characteristics of a jet pump 241 piston compressor 233
OPEX 1, 18 piston pump 229
orifice 358 plant characteristics 224
OSBL 375 plant layout 4, 8
oscillating displacement pump 229 plate heat exchanger 107, 130, 135
Oslo type crystallizer 220 plate module 205
osmotic pressure 206 plot plan 332
out of balance 143 plug valve 292
outlet line 363–366 plug-flow reactor with recycle 265
outlet vapor fraction 125 plug-flow tubular reactor (PFR) 264
outlet weir 161, 167 Podbielniak extractor 201
overall plot plan 11 polymer 70, 89, 189
overdesign 118, 123 polymerization 251
overpressure from outside 272 power consumption 232
oversaturation 218 power failure 352
ozone 325 Poynting factor 41
Prandtl number 376
package unit 375 Prandtl/v. Karman 279
packed column 141, 144, 354 preheating zone 122
pair parameter 58 pressure 333
Pall ring 148 pressure control 178
parallel baffle cut 121 pressure drop 120, 150, 165, 171
parallel orientation 115 pressure drop calculation 275
passive fire protection 15 pressure drop correlations for two-phase
PC-SAFT 71 flow 366
Peng–Robinson equation 35, 63, 69, 70 pressure drop of the irrigated packing 153
permeability 203, 204 pressure hydrolysis 326
permeate 203 pressure rating 289
perpendicular orientation 115 pressure relief 309, 336
pervaporation 209 pressure relief device 270, 336
PFR with recycle 265 pressure-swing adsorption (PSA) 212, 215
pH value 375 pressure-swing distillation 181
pharmaceuticals 89, 97, 189, 308 preventive maintenance 5
phase equilibrium 266 principle of countercurrent flow 140
phase inversion membrane 205 process control system 295, 334, 376
phosphorus 317 process description 3, 332
pilot plant 197 process flow diagram (PFD) 3, 332, 375
pinch 94 process simulation 2, 79
Pinch method 93 process water 376
pipe 81 profile 143
pipe class 289 pseudo-critical pressure 111
Index | 409
pseudo-critical temperature 111 reverse reaction 257

pseudo-stream 143 Reynolds number 250, 284, 376
PSRK 37, 70, 72, 74 Richardson’s law 176
Pt-100 295 risk parameter 334, 335
pulsation 198 rod-baffled exchanger 137
pump 13, 81, 223, 278, 333, 347, 356 rod-type baffle 117
pump characteristics 224, 347, 353, 356 rotary piston compressor 235
pump efficiency 223 rotary vane pump 244
pump failure 353 rotating disc contactor (RDC) 199
purge stream 84 rotating displacement pump 230
pv diagram 33 rough vacuum 129, 242
pxy diagram 29 roughness 276
runaway chemical reaction 349
Rackett equation 48 runaway reaction 262, 266
radial turbo compressor 235 rupture disc 270, 332, 337, 339, 363
random packing 145, 146, 148, 184
Raoult’s law 30, 41, 59 safeguard 333, 376
rate of relief 336 safety integrity level (SIL) analysis 334
rate-based 79, 142, 184, 186, 195, 197, 319 safety valve 6, 15, 270, 278, 332, 337, 340, 341,
rating mode 111, 118 345, 347, 357, 363, 376
reaction equilibrium 266 saturation zone 212
reaction factor 257 scale-up 197, 267
reaction force 355 schedule view 100
reaction kinetics 256, 259, 261, 266 SCR process 315
reaction rate 256, 260, 265 screw compressor 234
reactive distillation 185, 266–268 sea water 306
real continuous stirred tank reactor 264 sea water desalination 206
reboiler 122 sealing strip 118
recipe 99 Second Law 359
recommended velocities in a pipe 278 sedimentation 323, 326
rectification 139 sedimentation curve 194
rectifying section 188, 376 seed crystal 218
recycle stream 84 segment 184
redistribution 153 segment height 184
reflux 140 selectivity 210, 255
reflux ratio 141, 143, 177, 189 self-ignition 367
refrigerated water 308 semicontinuous mode 263
regenerative thermal oxidizers (RTO) 317 semipermeable membrane 206
relief amount 346, 347 sensitivity spider 23
relief pressure 346 separation block 376
repulsive 355 separation factor 26, 145, 162, 376
repulsive force 337 separation sequence 143
residence time 124, 126, 129, 160, 162, 166, separator 82
169, 170, 173, 201, 221, 263, 264, 267, 315, separator vessel 126
316 Seveso accident 328, 330
resistance thermometer 295 shell orientation 114
restriction orifice 336 shell-and-tube heat exchanger 350, 351
retentate 203 shell-and-tube type 107
reverse osmosis 90, 206, 325 short path evaporator 129
410 | Index
shut-off valve 293 stratified flow 287

sieve tray 156, 354 stripping section 145, 188, 376
SIL classification 334, 353 structured packing 147, 184
silicon 317 subcooling 122
simplifying assumption 356 subcritical flow 295
simulated moving bed 140, 214, 215 sulfuric acid 186
simulation mode 111 sun radiation 348
single segmental baffle 115 supercritical 47
siphon 193 supercritical flow 295
siphon breaker 249 superheating 300, 301
six-tenth-law 19 superimposed back pressure 338
sloped downcomer 166 surface tension 25, 77, 149, 185, 195
slug flow 286, 288 suspension crystallizer 218
SMB 215 system flooding 152, 169
SNCR process 315
Soave–Redlich–Kwong equation 35, 63, 70 TA Luft 186, 309–311, 315
sodium cyanide 330 tangent line 376
solid-liquid equilibrium 61 tear stream 84, 376
solubility 204 technical high-precision equations of state 39
solubility in the membrane polymer 204 TEMA 112, 116
solubility of gases 48 temperature 332
solvation 57 temperature class 367
space demand 133 temperature peak 262
specialty chemicals 89, 97, 189, 308 temperature profile 174
specific enthalpy 65, 111 temperature-swing adsorption (TSA) 212
specific surface 211 ternary azeotrope 181
speed of sound 280, 282, 359, 361, 363, 365, tetrahydrofurane–water 181
366, 371, 375, 376 Texas City 328
spiral wound module 205 TH diagram 94
split balance 79 theoretical stage 142, 149
split block 376 thermal combustion 310, 315
splitter 81 thermal combustion unit 316
spool piece 193 thermal conductivity 25, 26, 77, 111, 134, 135,
spout 253 185
spray regime 155, 166 thermal engine 95
spray tower 319 thermal expansion 290, 347
standard enthalpy of formation 65 thermal expansion valve 348
static crystallizer 219 thermal NO 314
static liquid head 122, 124 thermal oil 110
static payback period 23 thermal resistance 108, 119
steam 95, 299 thermal stability 127
steam control valve 352 thermal stress 127
steam trap 302 thermal vapor recompression (TVR) 92, 240
steamout 270, 376 thermocouple 295
Stichlmair model 150 thermodynamics 141
stochiometric line 369 thermosiphon reboiler 122, 124, 125, 127, 131,
stoichiometric coefficient 257, 258 136, 179, 283
stoichiometric reactor 267 THF 181
stoichiometry 255 thin film evaporator 129
Index | 411
three-phase-flash 104 vacuum 124

throttle valve 125 vacuum application 153
Tianjin accident 330 vacuum column 165, 172
tie-in points 376 vacuum distillation 141, 160
tie-rod 118 vacuum pump 231, 242
total organic carbon (TOC) 324, 376 value creation 21, 88, 90
Toulouse accident 329 value engineering 4, 376
trace element 320 valve 81, 278, 291
training simulator 173 valve tray 158
transport property 75 van-der-Waals equation 33
tray column 141, 154 van-der-Waals property 73
tray efficiency 184 vapor cross-flow channeling 169, 170
tray fixing 355 vapor horn 180
tray spacing 161, 164, 166 vapor phase association 69, 70
triple point 49 vapor pressure 27–29, 49, 70, 76, 377
troubleshooting 103 vapor pressure shifting 65
true component approach 58 vapor quality 28
tube arrangement 138 vapor recompression 131, 325
tube layout angle 114 vapor-liquid equilibrium 139
tube pass 115 vapor-liquid separator 239
tube pitch 114, 137 vapor-liquid-liquid equilibrium 59
tube rupture 333, 350 variable costs 18, 377
turbo compressor 235 vent 290
venturi scrubber 319
turbomolecular pump 245
vessel 81
turbulent flow 276
vessel breathing 308
turndown 156, 172
vibration 116, 127, 136, 137, 171, 228, 291, 296
twin plant 311
viscosity 25, 26, 127, 149, 185, 217, 275, 357
two-pass tray 163
viscous 339
two-phase factor 283
VLLE 104
two-phase flow 122
volume balance 340, 341
two-phase flow in the safety valve 366
volume concentration 257
two-phase flow pressure drop 363
volume translation 38, 70
two-phase pressure drop 283
volume-translated Peng–Robinson equation
two-phase through the safety valve 357
(VTPR) 38
Twu-α-function 38
vortex breaker 247
Txy diagram 30
vortex flow meter 296, 297
typical 376
vortex shedding 136–138
VTPR 70, 74
U-bend 290 VTPR equation state 259
U-type heat exchanger 115
ultra-high vacuum 242, 245 Wagner equation 51
ultrafiltration 203, 323 waste water incineration 326
ultrasound 326 waste water treatment 323
ultraviolet radiation 325 water 55
UNIFAC 73, 74 water hammering 306
UNIFAC consortium 74 waterspout 247
UNIQUAC 42, 44, 45, 59, 63, 70, 73 wavy flow 287
upper explosion limit (UEL) 368 Weber number 285, 377
412 | Index
weeping 156, 158, 168, 169, 174 wired packing 147

weeping rate 169 wispy annular flow 286
Wegstein method 85 working capital 377
weir height 157, 354
weir length 168 yield 255
weir load 168, 169 yield reactor 267
wetted area 345
Wilson 42, 44, 45, 59, 63 zeolite 211
wire mesh layer 251

Chemical Engineering August 2014

Uploaded by

Copyright:

Available Formats

Chemical Engineering August 2014

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Chemical Engineering August 2014

Uploaded by

Copyright:

Available Formats

Michael Kleiber

Reactive and Membrane-Assisted Separations.

Chemical Reaction Technology.

Advanced Process Engineering Control.

Corrosion for Engineers.

Reviews in Chemical Engineering.

Library of Congress Cataloging-in-Publication Data

Bibliographic information published by the Deutsche Nationalbibliothek

© 2016 Walter de Gruyter GmbH, Berlin/Boston

There is always much more to learn than can ever be taught!

A fool with a tool is still a fool.

2 Thermodynamic models in process simulation | 25

4 Heat exchangers | 107

4.8 Air coolers | 133

5 Distillation and absorption | 139

6 Two liquid phases | 191

7 Alternative separation processes | 203

8 Fluid ﬂow engines | 223

9 Vessels and separators | 247

10 Chemical reactions | 255

11 Mechanical strength and material choice | 269

12 Piping and measurement | 275

13 Utilities and waste streams | 299

14 Process safety | 327

List of Symbols | 379

A Some numbers to remember | 393

B Pressure drop coefficients | 397

1.1 Process engineering activities

be considered whether the plant can be operated at reduced or increased capacity,

Fig. 1.1: Example for the detailing in a PFD.

VBC125A 3"/2" FC 3"/2" VBC125A

Fig. 1.3: Example of the detailing in a PID.

1.2 Realization of a plant

Fig. 1.4: Example of a detailed plastic model.

– it was practically impossible to implement major changes;

More details can be found in [4].

Fig. 1.5: Example of a 3D representation.

Fig. 1.6: Example for an overall plot plan.

– Consider the minimum safety distances. E.g., compressors or units containing a

Fig. 1.7: Typical pump arrangement.

Fig. 1.8: Fully modularized process blocks.

Further documents which are compiled for the construction are:

Fig. 1.9: Example of an isometric drawing of a line.

For the commissioning, an operating manual is prepared where all activities in

It is amazing that even in Germany there is no chair for

Project Process Engineering Procurement Construction

Fig. 1.10: Matrix organization [2].

1.3 Cost estimation

Table 1.1: Factors for evaluation of the capital costs.

Table 1.2: Example for the structure of the operation costs.

Capital costs 200 000 000 $

Item Consumption Unit Price Unit Costs ($/t)

Production costs 1 152.10

Product sales 1 500.00

3 An example is given on p. 88.

There are a number of characteristic values to assess the feasibility of a project.

Cash ﬂow = (revenues − ﬁxed costs − variable costs) ⋅ (1 − tax rate)

Fig. 1.11: Example of a sensitivity spider chart.

Physical property % error % error capital cost