See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/260425306

Boiler-turbine simulator with real-time capability for dispatcher training

using LabView

Conference Paper · May 2010

DOI: 10.1109/OPTIM.2010.5510509

CITATIONS READS

2 506

3 authors, including:

Gheorghe-Daniel Andreescu N. Muntean

Polytechnic University of Timisoara Polytechnic University of Timisoara
67 PUBLICATIONS   1,048 CITATIONS    54 PUBLICATIONS   276 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Understanding Hearing Through Cochlear Implants. A Simulation Model of the Complete Hearing Path. View project

All content following this page was uploaded by N. Muntean on 16 April 2014.

The user has requested enhancement of the downloaded file.

 2010, 12th International Conference on Optimization of Electrical and Electronic Equipment, OPTIM 2010

Boiler-Turbine Simulator with Real-Time Capability

for Dispatcher Training Using LabView
Mihai Iacob*, Gheorghe-Daniel Andreescu*, and Nicolae Muntean**
*
“Politehnica” University of Timisoara, Department of Automation and Applied Informatics, Timisoara, Romania
**
“Politehnica” University of Timisoara, Department of Electrical Engineering, Timisoara, Romania
[email protected], [email protected], [email protected]

Abstract–This paper presents an open loop dispatcher training On the other hand, the paper also presents how to put a
simulator for boiler-turbine implemented in LabView for LabView virtual instrument (VI) – “the simulator”, into a real
COLTERM heating power plant of Timisoara, Romania. The time application, although this is done only for test purpose.
system employs real-time capability, graphical user interface
(GUI), uninterrupted operator interaction, having as The difference between running in real-time under a real-
background a low order boiler-turbine model for dynamic time operating system (RTOS) - in LabVIEW, or using a
simulation. The operator manually controls the fuel charge on general purpose operating system (i.e. Windows - GPOS), is
each of the three boilers, the turbine valve position and the that the RTOS assures determinism by executing the
steam to consumers, to anticipate parameter evolution on each simulator as a time critical priority program on the real-time
boiler and the electric power generated by turbine. target (hardware) leaving the interface to be executed by the
host (i.e. PC, touch screen).
I. INTRODUCTION The paper is organized as follows. Section II presents a
couple of tested models, general boiler schematics and means
The data problems and the need for accurate low order
of adapting model parameters and structure to the desired
models representing boiler dynamics is not new [1]. Every
plant. Section III presents the LabView code for
model has to be adapted in parameter and architectural terms
implementing the simulator and explanations regarding how
for each cogeneration plant. These models are built by using
to switch this code to real-time. Section IV presents
the fundamental principles of mass, energy and volume
simulation responses and results. Section V is for conclusions.
balances. The results are complex or simplified models.
Comparatively, the responses of these models can be
considered as equally valid [2]. II. BOILER PROCESS AND LOW ORDER MODELS
A suitable boiler model for long term dynamics simulation A general plant schematics is presented in Fig. 1. The
of a heating power plant system must accurately represent the water, entering on the boiler water walls, is transformed into
main unit outputs and capture some relevant internal variables steam due to heat transfer from the furnace to the pipes metal.
[3]. The monitored outputs are the power generated by turbine The steam is gathered in a drum, which sends it to a super
and the throttle pressure. The internal variables are the steam heater for transforming wet steam to saturated steam. After
flow, generated steam and drum pressure. leaving the boiler unit, the steam is sent to a turbine, which
A concrete case is the central heating power plant of generates electric power.
Timisoara, where the modeled system is divided into three 49
t/h coal fired boilers having the capacity of producing 100 t/h
steam at pressures up to 18.3 bars. These units use the steam
for internal and external consumers and for electric power
generation by passing the superheated steam to a 16 MW
turbine. The boilers are small cogeneration units compared to
the ones described in [2], due to their double functionality.
For this case, the paper presents an open loop simulator
based on a simplified boiler model, used for dispatcher
training to increase system safety and reliability. The
simulator is built under the hardware-in-the-loop (HIL)
concept in LabVEW to model the real system using live data.
In the next stage, it allows the integration of a rapid-control-
prototyping (RCP) system with the HIL simulated process for
testing and tuning in real time. The final integration of the
Fig. 1. General diagram of boiler-turbine unit.
control system with the real process will be in a short period.

978-1-4244-7020-4/10/$26.00 '2010 IEEE 864

 Fig. 3. LabView boiler model.

Fig. 2. Low order model of boiler system. SF = μ PT (3)

TABLE I. Nomenclature where: PSH = K SH SF 2 (4)

Parameter Description
The fuel dynamics is modeled using a dead time element
Cb Boiler storage constant with the time constant TD of the fuel, in series with a first
Ep Pressure deviation order element with time constant TF for the heat to reach the
F Fuel gain water inside the water walls.
HLM Steam generation high limit The model proposed in this paper for implementing the
Ksh Super heater drop friction coefficient open loop simulator is based on the one from Fig. 2 in [7].
Pd Drum pressure If the fuel controller is eliminated from Fig. 2 and the
P ref Throttle pressure reference model operates without pressure set point, then from the
Psh Pressure drop from drum to throttle control point of view, the boiler-turbine unit can be modeled
Pt Throttle pressure
as a MIMO 2x2 system [8], with valve and fuel demand as
inputs and throttle pressure and steam flow as outputs. The
Sg Steam generation
LabView adapted model for one of the three boiler units with
Sf Turbine steam flow
reference to the generic model in Fig. 2 [7] is presented in
Sn Net steam flow into drum
Fig. 3, where H(s) represents the fuel dynamics.
Td Fuel time delay Model coefficients can be calculated by using the methods
Tf Fuel and water wall time constant described in [4] and [2], by conducting two tests on the real
Valve Valve position plant: the first by modifying the throttle valve position with
about 10% for determining the boiler storage constant (CB),
Several boiler models were proposed for different types of
and the second by making a step change in fuel demand for
generating units, some of them having as starting point the
determining the fuel dynamics TD and TF coefficients.
low order model [4] presented in Fig. 2 with the following
main elements. The turbine valve is used to control the steam
pressure (PT) and steam flow (SF) at the turbine inlet.
The net steam flow (SN) is the difference between the
generated steam (SG) and SF. The drum pressure (PD) is
proportional to the integral of the net steam flow to the
storage volume (CB). This gives the energy balance equation:
d ΔPD
ΔSG − ΔS F = CB (1)
dt
PT of steam inlet to turbine [5-6-8] is the difference
between drum pressure and the pressure drop on the super
heater (Psh). The drop in steam pressure from the drum to
turbine is represented as the square of the steam flow rate
(SF), while Ksh denotes the friction coefficient of the super
heater tubes [5]. The paragraph above illustrates the two main
nonlinear characteristics of the system:
P −P =K S 2 (2) Fig. 4. Simulator interface for 3 boiler-turbine system.
D T SH F

865
 A simple turbine model for simulator is presented in Fig. 5.
The steam is gathered in the steam pipe and sent to the 16
MW turbine. The parameters are identified as follows:
SFgain = 0.054; Turbinegain = 0.0625.

III. HOW ITS MADE - IMPLEMENTATION

The program was constructed in LabView and its structure
consists of a sequence, which in the first tab executes the
Fig. 5. Turbine model. initializations of the controls and indicators, and in the second
tab - the logic of the simulator with associated graphical
For boiler no. 1 in Fig. 3 of the heating power plant at interface. Fig. 6 presents the simulation loop which runs
Timisoara, the parameters were identified as follows: under Runge-Kutta 4 order solver with fixed time step of 0.1
seconds in non real time mode. The loop contains three
TD = 160; TF = 55; 1/CB = 0.000461; simulation subsystems that represent each one boiler.
F = 1.998; Ksh = 0.000138; SFgain = 0.096. The subsystems are assembled as in Fig. 3, with
The above coefficients were applied to all of the three associated inputs and output. The main simulation loop calls
boilers due to the same architecture and performance. The the subsystems in order to execute the boiler algorithm and to
steam from the boilers is collected on a common pipe and pass data to the graph indicators which show drum pressure
sent to turbine inlet. Each boiler has a valve in order to vs. throttle pressure and steam flow vs. generated steam. The
balance the steam on the common steam pipe as in Fig. 4. steam flow from each boiler is then collected at the turbine
In the real situation, the problem is that this valves are inlet by means of a summation block, minus steam to
open/close types and do not allow positioning in order to consumers which is an input that the dispatcher can set.
control the throttle pressure. This is a major inconvenience Real-time capability can be implemented by choosing to
because until the boiler reaches the pressure in the throttle run only the simulation presented in Fig. 6 on a RT target
pipe, the afferent valve cannot be opened and all the steam is with a RTOS assuring determinism to the process. The
principle is presented in Fig. 7.
either lost in the atmosphere or given to consumers.

Fig. 6. Simulation loop for boiler-turbine model.

866
 IV. SIMULATION RESULTS
Several tests were done on the simulator to verify the
accuracy of the model dynamics. Each test was done after the
units reached the desired operating point. The time to reach
these points differs, depending on the load of each unit.
A. Pressure Response of Boiler to Valve Step Change
The first test employs stabilizing the unit at the following
parameters:
Boiler firing rate B=39t/h; PT=13.5 bar; SF=77.92t/h;
Valve 1=open; Turbine power N=4.23; Valve=60%.
Next, a step change in throttle valve position is made from
60% to 70% (Fig. 9). The throttle pressure is up to 10% lower
than drum pressure due to pressure drop on the super heater.
A spike is detected at turbine power and steam flow, because
of the instant excess of steam; as the pressure and steam
balance, parameters quickly return to their nominal points.
Fig. 7. Network communication protocol [9]. Similar test are done for units two and three and for other
operating points, but the Figures are omitted for brevity.
B. Responses of Boiler to Step Change in Firing Rate

Fig. 8. GUI code in host loop VI.

Fig. 9. Pressure response of boiler 1 to valve step change.

Using the real time communication wizard three VI’s are Modification at the load of the boiler and steam to
created: time critical loop (TCL), which contains the consumers makes the subject of the next test. Other operating
simulation; normal priority loop (NPL), which contains the points are chosen:
communication protocols through RT FIFO’s and calls the
TCL; and the host loop, which contains the GUI. To access Boiler 1 operating point: PT = 10 bar; PD = 10.5 bar;
data in host loop code and to construct and condition GUI B = 29 t/h; SF = 58 t/h; Valve1 = open.
(Fig. 8), local variables are used. The GUI runs in parallel, (in Boiler 2 operating point: PT = 11.2 bar; PD = 11.7 bar;
a while loop) with the host loop inside the host loop virtual B = 32 t/h; SF = 64t/h; Valve2 = open.
instrument.
The NPL is downloaded to the RT hardware PCI Throttle Valve = 60%; N = 6.64 MW – generated by two
7041/6024E and the host loop runs on the PC. The use of real boilers at close pressure operating points.
time can be justified by HIL and RCP, where the simulated For boiler two, a step change in its firing rate is made from
process runs on the RT target with a real controller, or the 32 t/h to 29 t/h. A pressure and steam flow drop is recorded
control algorithm runs on the RT target with the real plant. In (Fig. 10 and Fig. 11). The power output of the turbine also
either case, the RTOS provides determinism, which is the decreases from 6.63 MW to 6.32 MW (Fig. 12).
characteristic of a system that describes how consistently it If the steam from the common collector pipe is sent to
responds to external events or performs operations within a consumers, there is no modification in boiler parameters.
given time limit [9-13]. The power of the turbine decreases in this case due to less
mass steam flow at the inlet (Fig. 12).

867
 V. CONCLUSION
Low order boiler models have been widely used in
literature for estimating industrial furnaces and generation
units thanks to their simplicity. Such models are not perfect
and might not always match exactly the responses of the real
plant. This paper presents a simulator capable to be switched
in real time, based on static data recorded on the actual plant.
The simulator implemented in LabView employs real-time
capability, graphical user interface (GUI), uninterrupted
operator interaction, having as background a low order boiler-
turbine model for dynamic simulation.
The functionality of this simulator can be found in:
Fig. 10. Pressure drop for boiler 2. • dispatcher training for operating the boilers
• determining where improvements are needed by
analyzing times during transient responses and
parameter values at various operating point.

ACKNOWLEDGMENT
The authors are grateful for the technical support received
from COLTERM Timisoara thermal heating power plant.
Ph.D. student Mihai Iacob thanks for the financial support of
his work under strategic grant POSDRU/6/1.5/S/13/ 2008,
project ID 6998, within the Sector Operational Program for
Human Resources Development 2007-2013, Romania, co-
financed by the European Social Fund - Investing in People.

Fig. 11. Steam flow drop for boiler 2.

REFERENCES
[1] F. P. Demello and F. P. Emad, “Boiler pressure control configuration,”
IEEE Winter Power Meet., N.Y., Jan. 1967.
[2] F. P. de Mello, “Boiler models for system dynamic performance
studies,” IEEE Transactions on Power Systems, vol. 6, no. 1, pp. 66-74,
Feb. 1991.
[3] M. E. Flynn and M. J. 0’Malley, “ A drum boiler model for long term
power system dynamic simulation,” IEEE Transactions on Power
Systems, vol. 14, no. 1, pp. 209-217, Feb. 1999.
[4] E. Cheres, “Small and medium size drum boiler models suitable for
long term dynamic response,” IEEE Transactions on Energy
Conversion, vol. 5, no. 4, pp. 686-692, Dec. 1990.
[5] E. Cheres, Z.J. Palmor, and J. Tuch, “Drum type boiler following
control conFiguration,” IEEE Transactions on Energy Conversion, vol.
9, no. 1, pp. 199-205, March 1994.
[6] C. S. Chen, Y. D. Lee, C. T. Hsu, and C. K. Chen, “Transient response
of an incinerator plant by considering boiler model with dynamic
steam,” IEEE Transactions on Power Systems PWRS, vol. 23, no. 1, pp.
92-99, Feb. 2008.
Fig. 12. Power decrease firing rate and steam to consumers. [7] Wen Tan, Fang Fang, Liang Tian, Caifen Fu, and Jizhen Liu, “Linear
control of a boiler-turbine unit: Analysis and design,” ISA Transactions,
The first power decrease is given by the drop in the firing vol. 47, no. 2, pp. 189-197, April 2008.
[8] Wen Tan, Jizhen Liu, Fang Fang, and Yanqiao Chen, “Tuning of PID
rate, while the second decrease is due to 15 t/h of steam sent controllers for boiler-turbine units,” ISA Transactions, vol. 43, pp. 571-
to the internal/external consumers. 583, 2004.
The results presented in this chapter are intuitive for [9] National Instruments, “LabVIEW Real-time module user manual”,
ftp.ni.com/support/manuals/322154e.pdf, April 2004
dispatchers in order to estimate boiler responses at various [10] National Instruments, “Selecting Your LabVIEW™ Real-Time
operating points and plant maneuvers. Further expansions can Deployment Platform”, ftp://ftp.ni.com/pub/devzone/tut/wp2435.pdf,
be made to the simulator by estimating more accurately the Aug. 2004.
[11] National Instruments, “LabVIEWTM Control Design Toolkit User
responses of the plant. This can be achieved by executing Manual”, www.ni.com/pdf/manuals/371057d.pdf, Sept. 2004.
literature presented, dynamic tests, rather than static ones or [12] National Instruments, “LabVIEWTM Simulation Module User Manual”,
by using equation based models, which can be difficult to www.ni.com/pdf/manuals/371013a.pdf, April 2004.
[13] Mihaela Lascu, “Advanced Programming Techniques in LabView”, Ed.
implement due to limited data availability from the physical “Politehnica” Timisoara, 2007.
plant.

868
 Mihai Iacob received his B.E. in automation Gheorghe-Daniel Andreescu (M’03–SM’05)
and computer science from “Politehnica” received the M.S. degree in applied electronics and
University of Timisoara, Romania in 2008. He the Ph.D. degree in automatic control systems from
is currently a Ph.D. student at “Politehnica” the University Politehnica of Timisoara, Timisoara,
University of Timisoara with research interest in Romania, in 1977 and 1999, respectively. From
intelligent optimization technologies, plant 1977 to 1984, he was a Researcher and R&D Unit
simulation and PLC/SCADA/DCS concepts for Manager in the area of applied electronics and
industrial processes. industrial control with Electrotimis Company,
Timisoara. Since 1984, he has been a Member of
the Department of Automation and Applied
Informatics, University Politehnica of Timisoara,
Nicolae Muntean (M’91) received the B.Sc. where he is currently a Full Professor. Since 1993, he has also been a
degree and the Ph.D. degree from the Researcher with TIMTEH Electronics Company, Timisoara, where during
“Politehnica” University of Timisoara, 1994–2002, he specialized in automatic testing for avionics, e.g., at British
Romania. He is currently Full Professor at the Airways Avionic Engineering, Cardiff, U.K. He is the author or coauthor of
“Politehnica” University of Timisoara, over 60 papers in journals and international conference proceedings and two
Department of Electrical Engineering. He is the books. His major book is Estimators in Control of Electric Drives—
author of over 50 publications in the field of Applications to Permanent Magnet Synchronous Motors (in Romanian)
power electronics and drives, including four (EOU, 1999). His research interests are mainly in the areas of advanced
books (as single author or coauthor). control of ac drives sensorless control, electric/hybrid vehicles, observers,
His major expertise is in industrial sliding-mode control, mechatronic systems—robotics, automatic testing for
implementation of electric drives with power avionics, microcontroller and VHDL applications, real-time implementations,
electronics in various applications. He was, from 1981 to 1988, a design and monitoring and control of distributed systems.
engineer in electric drives and, for 3 years in this period, design manager. Dr. Andreescu is a member of the Technical Committee TC4.2 on
He is currently involved in research and development related to Mechatronic Systems of the International Federation of Automatic Control,
automotive electronics and renewable energy systems. the Romanian Society of Control Engineering and Technical Informatics, and
He is an Associate Editor of the internet-only Journal of Electric the Romanian Association of Robotics. He is member of the Engineering
Engineering (www.jee.ro). Sciences Commission of the National University Research Council
(CNCSIS), Romania.

869

View publication stats