Broad Lane, Sheffield, S3 7HQ Telephone: +44 (0) 114 289 2000 Facsimile: +44 (0) 114 289 2500
Broad Lane, Sheffield, S3 7HQ Telephone: +44 (0) 114 289 2000 Facsimile: +44 (0) 114 289 2500
Broad Lane, Sheffield, S3 7HQ Telephone: +44 (0) 114 289 2000 Facsimile: +44 (0) 114 289 2500
HSL/2006/21
Objectives
Main Findings
Operators of metallic and reinforced plastic (GRP) storage facilities have well
established requirements for inspecting their tanks. Thermoplastic tanks are
an attractive alternative to metal and (GRP) for many products but are often
installed with the view that they will not deteriorate and do not need to be
inspected or assessed. There are no established procedures for inspecting
these tanks and consequently they are often used without due consideration
of their condition and fitness for service.
This report offers guidance on the specification, design, fabrication and in-
service inspection of thermoplastic storage tanks in general. The emphasis is
on tanks installed over the last 20+ years although it is acknowledged that a
standard will soon be published which addresses many of the concerns and
issues raised in this document.
Page
1 INTRODUCTION 1
2 MATERIALS OF CONSTRUCTION 3
3 DESIGN & SPECIFICATION 4
3.1 Introduction 4
3.2 DVS 2205 5
3.3 European standard prEN 12573 11
3.4 NACE Publication 5A171 (1995 Revision) 12
3.5 Compatibility data 13
4 FABRICATION 13
4.1 Welding 13
4.1.1 Butt fusion welding 13
4.1.2 Extrusion Welding 13
4.1.3 Hot gas welding 14
4.2 Methods of fabricating thermoplastic tanks 14
4.2.1 Sheet forming 14
4.2.1.1 Background 14
4.2.1.2 Fabricators 15
4.2.2 Helix winding 16
4.2.2.1 Background 16
4.2.2.2 Fabricators 17
4.2.3 Rotational moulding 17
4.2.3.1 Background 17
4.2.3.2 Design Considerations 18
4.2.3.3 Fabricators/suppliers 18
5 TRANSPORT & INSTALATION 18
6 INSPECTION & QUALITY ASSURANCE 19
6.1 Pre-service inspection and quality assurance 19
6.2 In-service inspection 20
6.2.1 Techniques 20
6.2.2 Frequency 21
6.2.3 Acceptance standards 22
6.2.4 In-service testing 22
7 SERVICE EXPERIENCE: FALIURES & DEFECTS 23
7.1 Fabrication defects 23
7.1.1 Weld defects 23
7.1.2 Delamination of wall materials 23
7.1.3 Winding defects 24
7.2 Chemical degradation 25
7.2.1 Nitric acid 25
7.2.2 Sodium hypochlorite 26
7.2.3 Sulphuric acid 27
7.3 Other chemical effects 27
7.3.1 Hydrofluoric acid blistering 27
7.3.2 Hydrofluoric acids, leaching of additives of HPDE 28
7.4 Aspects of tank design 28
7.5 Operational defects and failures 29
7.5.1 Lifting of wall 29
7.5.2 “scuffing” of base 30
7.5.3 Embrittlement of outlet 30
7.5.4 Perforation of tank base by level sensor 31
7.5.5 Splitting of seal welds 31
7.5.6 Failures of sheet formed tanks 31
7.5.6.1 Splitting of seams 31
7.5.6.2 Internal bulging 32
7.5.6.3 Uneven discolouration of sheet tanks in PVDF 32
7.6 Effects of under and over-pressure 33
7.7Installation issues 33
7.8 Maintenance issues 34
7.9 Exposure to excessive temperature 35
7.10 Bund issues 36
8 CONCLUSIONS 37
9 RECCOMENDATIONS 37
APPENDIX 1 Acknowledgements 38
APPENDIX 2 Guidelines for the operation of HDPE tanks 39
APPENDIX 3 References 41
1 INTRODUCTION
A review has been requested of the specification, design, fabrication and in service
inspection and assurance of fitness for service of thermoplastic storage tanks to provide
HSE personnel with an overview and guidance notes. It must be emphasised that
thermoplastics have a good record, established over more than 30 years, of safe storage of
a wide range of chemicals and with few reported incidents of major failure. An estimated
5500 High Density Polyethylene & Polypropylene helix wound tanks of different sizes have
been supplied over the last 20 years; of these more than 60 having typical capacity of
~23 m3 have been supplied for HF duty. Maximum tank capacities of around 70 m3 are
supplied by fabricators of helix wound tanks. One manufacturer of sheet formed
thermoplastic tanks, Niplast Ltd, indicates a capability of fabricating tanks up to 100m3.
Fabricators and users of these tanks consulted during preparation of this report are
acknowledged in Appendix 1.
Table 1 Specific guidance was requested for helix wound high density
• aluminium sulphate polyethylene tanks used to store hydrofluoric acid but an
• beers additional need has been identified to consider the hazards of
other products such as sodium hypochlorite particularly when
• caustic soda
stored in large tanks of non-homogeneous forms of
• cider
construction.
• detergents
• dyes & bleaches
• ferric chloride Operators of all storage tanks have a well established duty to
inspect them periodically to assess their fitness for service;
• fertilisers
standards and guidelines covering specification, installation
• hydrochloric acid
and inspection of tanks fabricated from steel and GRP have
• hydrogen peroxide
been published. The design code for most thermoplastic tanks
• phosphoric acid used in the United Kingdom, DVS22051, is outdated and will
• sodium hypochlorite shortly be replaced by a European Standard; neither document
• sulphuric acid gives guidance on in service inspection and there are no
• wines established procedures or guidelines for inspection of these
tanks.
Many operators install thermoplastic storage tanks with the view that they will not
deteriorate and therefore do not need inspection or maintenance. Other operators may
take the view that thermoplastic tanks should simply be replaced at the end of their design
life but may not implement any formal intermediate inspection. There is often economic
pressure for tanks to be used beyond their design life sometimes without due consideration
of the condition and fitness for service.
This report concentrates on tanks which are currently in service and which should have
been designed and fabricated in accordance with the appropriate standard, DVS2205 and
1
best practice at the time. Many of the issues raised here have been addressed by the
European Standard which will be issued shortly and which reputable fabricators will be
applying now.
Two examples illustrate the importance of inspection and assessment of storage tanks and
the potential hazards of incorrect design and/or installation and emphasise the need for
operators to understand their duty to assure themselves of the correctness of a design and
installation. Both incidents happened within the last 10 years and are from the author’s own
records .
The first is an example of how a ‘small’ error during installation can negate the most careful
and detailed pre-service controls and subsequent in-service inspection that included
detailed thickness checks at 3 monthly intervals:
2
2 MATERIALS OF CONSTRUCTION - POLYOLEFINS (Polyethylene&
Polypropylene)
H H H
( C ) ( C C )
H Polyethylene H CH 3 Polypropylene
Polyethylene properties depend strongly on the molecular weight of the polymer and on the
degree of branching (number of branches along a given length). The crystallinity increases
the modulus of the material. The main types of interest for fabrication of storage tanks are:
• High Density (HDPE), (density ~0.96). High Density polyethylene is linear, has high
molecular weight and high crystallinity, typically 95%.
• Low Density (LDPE), (density ~0.92) Low Density polyethylene is branched. The higher
the crystallinity, the more dense the material. Branching of a polymer chain decreases
its ability to crystallise and therefore decreases the density.
Although PE offers good resistance to chemical attack from acids and bases, it appears
that such environments may accelerate fatigue crack growth in PE3.
The properties of PE are primarily affected by the crystallinity and the molecular weight.
Modulus increases as density increases, so does melting point, strength, and hardness.
These properties are not fixed. PE offers an excellent example where properties of the
same polymer can vary widely due to molecular weight and molecular weight distribution.
The use of high and low density polyethylene (PE) and polypropylene (PP) in the industrial
area is extensive. These comprise the lowest cost polymers for piping, smaller structural
shapes, containers, and a host of other applications which generate an industrial usage of
these materials of millions of pounds a year. Of the polyolefin products, the low and high
density polyethylene and polypropylene constitute the vast majority of uses with the low
density product being predominant.
Low density polyethylene is used extensively for piping, because of its ease of handling and
fabrication on the site. The chemical resistance of the product is outstanding, although not
as good as high density polyethylene or polypropylene. Fusion welding of the material is
readily accomplished with appropriate equipment, and installations of piping made in this
manner are the least expensive and most durable of any material available for waste lines,
water lines, and other miscellaneous services not subjected to high pressures or
temperatures. The upper temperature limit for the material is 60°C. The product can be
compounded with a wide variety of materials to increase strength, rigidity, and other
mechanical properties.
Chemical resistance is excellent within the temperature region allowed by the material. It is
now available in a fibre reinforced product to further increase the mechanical properties.
The low modulus of the material must be recognised to provide successful installations.
Straight-chain hydrocarbons, aromatic hydrocarbons, chlorinated hydrocarbons, and strong
3
oxidants must be avoided. All other materials are handled readily by the product. Stress
cracking can be a problem without careful selection of the basic polymer used in the
product or proper compounding to reduce this effect. Compounding of the product is also
recommended to reduce the effect of atmospheric exposures over long periods.
High density polyethylene and polypropylene are in many ways similar for chemical
applications. Considerably improved mechanical properties are obtained by the use of
these materials, and the chemical resistance is also greatly increased. Only strong
oxidants will attack the materials appreciably within the appropriate temperature range and
stress cracking of HDPE can be a problem if proper selection of the resin is not made.
The modulus of the polypropylene is somewhat higher, which is beneficial in certain
instances. The coefficient of thermal expansion is less for polypropylene than for high
density polyethylene.
Both materials are readily welded.
Other materials which are used and are covered by the relevant standards for design and
construction of thermoplastic tanks are Polyvinylidene Fluoride (PVDF) and Poly Vinyl
Chloride (PVC). Neither material is used extensively for large tanks and will generally be
used as a liner with a GRP shell.
All tank design procedures are required to take into account materials compatibility, working
temperatures, anticipated loading by the contents, wind and seismic factors and must also
define a design life. For metallic tanks an allowable stress in the wall is calculated4;
reduction of wall thickness due to corrosion will increase the working stress and thus
service life is defined by an allowable loss in wall thickness. Designers of GRP tanks will
define the service life by the number of fill / empty cycles as cyclic loading causes fracture
of glass fibres and the resin5. Thermoplastic tanks use creep deformation as the design
basis for service life.
Some early work6 published prior to issue of the design code used over the last 20 years,
DVS2205, discussed the use of creep as the design basis of thermoplastic tanks and
confirmed its applicability by testing a deliberately under-designed tank to destruction. This
work made the observation that load history is likely to be important as it was shown that
approximately ten times the loading (i.e. fill) time had to elapse before the deformation of
the tank could be said to be under static load. Equally on emptying the tank a considerable
time period was required to recover the original dimensions. Consequently, apart from a
few constant head tanks most tanks in normal operation will always be cycling and
therefore never come under the ‘static load’ which is the basis of DVS2205. It is possible
that the cyclic loading of some tanks will reduce service life by encouraging fatigue cracking
at discontinuities in the structure.
4
Materials compatibility of thermoplastics allows their use for a very broad range of
chemicals and they are often used to minimise product contamination which even a small
amount of corrosion of a metallic tank could cause. Thermoplastics are, in general, not
brittle and are less prone to catastrophic failure. The main drawbacks when compared with
other materials are:
• thermoplastics are inherently weaker and therefore require a much thicker construction
than GRP or metals which limits the size of tank which can be fabricated.
• thermoplastics tend to loose strength very rapidly with temperature which limits the
maximum allowable operating temperature
3.2 DVS 2205: ‘Calculation of Containers and Apparatus made from
Thermoplastics’ March 1974
This document drafted by the “Deutscher Verband für Schweißecnik” under the
International Institute of Welding has been the basis for design and construction of
thermoplastic storage tanks since ca 1975.
DVS 2205 is based upon creep of the material, i.e. permanent strain or ‘stretching’ over a
period of time, under constant load, until failure occurs. The results of tests on pipes are
presented in a series of creep strength curves as a function of temperature and time for a
particular grade of thermoplastic materials, Figure 3. This type of graph is used by the tank
designer to establish the creep strength at the design temperature for the required design
life. For example for a design temperature of 20°C and a design life of 10 years the creep
strength, K is ~8.4 N.mm-2.
5
Figure 3
Data given in DVS2205 and usually quoted in other sources is for pipe grades of
polyethylene. Helix wound tanks have always used a small number of extrudable, high
modulus pipe grade materials for which creep data exists or for which the data in DVS2205
would be valid. Sheet tanks are fabricated in pressed or calendered material which will
have different creep properties and the use of graphs in the standard may not be valid.
It should be noted that the strength of HDPE in particular is lost rapidly as temperature
increases; in general, this leads to a lower maximum working temperature for
thermoplastics when compared with GRP. There are examples of vessels fabricated by
Chem Resist Ltd which have been designed to handle processes which cycle well above
the ‘maximum allowable temperature’ because thermal conductivity of HDPE is poor and
the bulk of the wall does not reach the process temperature. These vessels are understood
to have wall thickness of ca 100 mm.
6
Figure 4
level probe
A maximum design temperature based upon the
temperature of the tank contents will not take into account
the high and variable wall temperatures where for example
a tank is exposed to direct sunlight. This factor can result
in a wall temperature estimated to be higher than 60°C and
a considerable temperature differential between the
exposed and the shaded part of the tank. Potentially this
will result in considerable distortion: a temperature rise
from 20-60°C would result in an increase in height of ca 70
mm on a tank 10m high. In one situation ‘faulty’ readings
by liquid level sensors have been attributed to the effects
of this distortion, Figure 4. Sheet formed tanks which have
a non-homogeneous wall are even more susceptible to
non-uniform expansion and separation of the outer layers
which will place additional loads on the inner layers and on external seal welds.
The value K is used to calculate an allowable stress (σzul) in the tank wall as shown below
where:
K = creep strength , N.mm-2
fs = factor for the method of welding used to fabricate the tank barrel
A2 = factor for the chemical to be stored
A4 = factor to allow for the specific strength of material at the design temperature
S = safety factor
• fs, factor for the method of welding used to fabricate the tank barrel
Material
Process PP PE
Heated tool butt welding, 0.8 0.8
normally considered to be machine controlled
Hot gas extrusion welding 0.6 0.6
This factor relates particularly to the welds used to fabricate the barrel; for a helix wound
tank which has no structural welds it is normal to use a factor of 1 when calculating wall
thickness. Sheet formed tanks have a number of welds to form the barrel and it must be
considered best practice to use machine controlled butt welding for the vertical seams
which will be exposed to the maximum hoop stress.
7
DVS 2205 makes the observation that these weld factors ‘presuppose complete mastery
and application of welding processes’. A regular check on welding operations is vital to
confirm for example that a machine butt welder is consistently able to produce welds having
a strength which is 80% of the parent material or that a manually controlled extrusion
welder is correctly set. For tanks on hazardous duties it would be appropriate to require
weld test pieces to be prepared during fabrication.
∗ DVS 2205 lists a number of chemicals and values for a chemical resistance factor,
FCR. A2 is the reciprocal of FCR and is used in equation (1). Water is given a FCR of 1
and an aggressive chemical such as nitric acid, for example, is given a value of 0.01,
the implication is that the allowable stress when exposed to nitric acid should be just
0.01X that allowed in contact with water to get the same life. This factor is critical to
the correctness of the design and, ideally, should be related to the specific grade of
polymer used particularly where compatibility is known to be doubtful. There is
generally insufficient time available to complete anything more than a cursory test and
although other sources of compatibility data may be available, none specify a factor
for the calculation and the ultimate choice is at the discretion of the designer
The data presented in DVS2205 should be viewed with caution as the chemical resistance
can vary significantly between specific grades of material. For example the very low value
for sodium hypochlorite implies that this would not be an acceptable combination although it
has been used frequently.
8
This factor takes into account the specific strength of the material as a function of
temperature and incorporates to some extent a ‘shape’ factor and notch sensitivity.
Temperature °C
Material -10 20 40 60
HDPE 1.2 1.0 1.0 1.0
PP - homopolymer 1.8 1.3 1.0 1.0
PP - copolymer 1.5 1.1 1.0 1.0
• S , a safety factor chosen by the designer. The value of this factor depends upon the
loading conditions as shown below:
Case
1 2 3
Mode of Static Load at ambient Static load under variable Dynamic Loads, i.e.
loading temperature conditions of temperature & frequently fluctuating
No danger to personnel, fill level. No danger to fill/empty cycles, temperature
equipment or environment personnel or objects in the variations
event of failure. Hazard to personnel,
equipment or environment
Factor S 1.3<------------------------------------------------------------------------------------------->2.0
The value chosen could be between these two extremes and could be greater than 2.0 if a
particular assurance of adequacy of the design is required.
The maximum permissible stress, σzul is used to calculate the required tank wall thickness;
this is determined at a number of points down the tank height to generate a thickness
profile for the whole tank.
s = D.ρ mm
2.σzul
Where:
D = tank diameter mm
ρ = static head pressure at point N.mm-2
considered
σzul = permissible stress N.mm-2
ρ = h.SG N.mm-2
1.02x105
Where:
h = depth to position in wall to be considered mm
SG = specific gravity of contents
Whatever method is used to fabricate the tank it is inevitable that the thickness required will
be generated in a series of bands each of a discrete thickness. A typical 24000l tank for
hydrofluoric acid would have a lower wall thickness of ca 50 mm. DVS 2205 makes no
comment on the method of creating the required wall thickness.
9
DVS 2205 accepts that the total design life should be the aggregate of time under load. The
calculation assumes constant loading over the design life to determine the creep strength -
in reality tank loading will almost inevitably cycle and the average loading could be well
below maximum experienced. This means that with the built in safety factors the wall
thickness will be significantly in excess of that required to simply resist the load.
The factors and values can be chosen to ‘over-design’ tanks, for example one operator of
HF storage tanks, Laporte Fluorides, specified a higher SG (1.9 instead 1.5) for the product
and a higher service temperature (40°C instead of 20°C) to arrive at an ‘acceptable’ design.
Figure 5
This calculation allows the fabricator produce a plan to create a cylinder of the required
height and thickness profile.
10
Additional calculations will generally be required to cover other factors that are not covered
specifically in DVS2205:
• wind loading
• compensation for manways and large branches
• additional restraint i.e. extra thickness for the lower wall to accommodate attachment to a
rigid base
Base Thickness: DVS2205 allows a base thickness between 6 mm and the thickness of
the lower wall. Many early tanks even on hazardous duties had thin bases, typical 6-8 mm
attached to a lower wall ca 50 mm thick.
Tie Down: If windloading calculations indicate that up lift is possible it will be necessary to
provide tie down arrangements. Different fabricators offer differing arrangements from steel
girdle/legs to plastic brackets welded to the lower tank wall. In the author’s experience the
steel girdle offers the best support without excessively restraining movement of the lower
wall arising from thermal and load effects and shrinkage.
11
• the importance of a specification which has been agreed between the customer and the
fabricator is recognised in the form of a sample questionnaire
• the importance of correct transport and installation is addressed with specific
recommendations, e.g.
∗ tanks should not be loaded or transported in ambient temperatures below 5°C
∗ tanks should only be installed on continuous horizontal flat surfaces, e.g. concrete
bases. Unstable materials such as sand should NOT be used
• sheet formed or multi-layered tanks can only have a wall made up of two layers
This is perhaps one of the main changes in the design and construction of these tanks
and acknowledges the shortcomings of this category of tank. One fabricator, Niplast, is
considering the use of thicker sheet to allow him to continue making relatively large
tanks; this may prove to be an inappropriate interpretation of the standard as one of the
advantages of having multiple layers is the number of welds and layers which must be
penetrated in the event of failure of an internal seal weld. With only two layers the
possibility of catastrophic failure by complete fracture of a vertical seam appears to be
greater. It is the author’s opinion that would be better to specify a limit to the size of tank
fabricated by this method.
• additional factors shall be incorporated in the allowable thickness calculations resulting in
a lower wall thickness 1.25x thicker than a helix wound tank
• the requirements for welding of layers in a multi-layer tanks are specified
• the allowable levels of residual stress in sheet material are specified
• a maximum allowable nozzle connection of 300 mm in the lower wall of the tank
It is most important that the considerable improvements made by this standard are
recognised by its adoption for all thermoplastic tanks and vessels; reputable fabricators are
already applying it to their products but other suppliers particularly of sheet formed tanks
may be tempted to continue to offer tanks fabricated to DVS2205. Operators who are
unaware of the CEN Standard will not realise they are ordering an inferior product.
12
3.5 Compatibility data
Numerous sources of compatibility information exist for a range of chemicals; however few
give a relative rating which can be used directly in the calculation of tank dimensions. The
following list is by no means exclusive
• Schweitzer - Corrosion Resistance Tables
• NACE CorrSur2 - Computer based graphical database
• Manufacturers Literature
• RAPRA Publications
As described above prEN 12573 cites EN1778 for procedures to establish the compatibility
of materials of construction and where time permits it would always be advisable to test a
material where its compatibility is uncertain.
4 FABRICATION
4.1 Welding
Calculations under DVS 2205 include factors for different types of welding; these are:
4.1.1 Butt fusion welding
Sheets of materials to be joined are clamped to a flat bed, the cut edges of sheet material
are softened by contact with an electrically heated platen which is withdrawn and the two
edges pushed together, Figure 6. The excess weld bead which is exuded from the joint will
be removed following welding. The process is under full mechanical control and results in
the minimum disruption of the material and maximum weld integrity. One fabricator,
Niplast, reports weld strengths which are consistently well above the 80% of parent material
strength required by DVS2205.
Figure 6
13
4.1.3 Hot gas welding
The work pieces are heated with a flow of hot gas Figure 7 Hot gas welding
and a rod of filler material is pushed under
manual pressure though a guide on the welding
gun into the weld preparation, Figure 7.
Softening rather than melting of the material
allows fusion between the filler and the work
piece. The integrity of the weld is very dependent
upon the skill of the operator and voids and
14
Figure 10 The required wall thickness is achieved using several hoops
of material which, in a large tank, could require 5 or more
layers. For large tanks temporary supporting rings may be
used to maintain the cylindrical shape and are partially
removed following completion. Figures 21 and 30 show
remnants of stiffening ring
which have been partially Figure 11
(and crudely) removed.
The barrel is welded to a
base and branches
inserted through the wall
where required. There is û
failure of internal welds
significant potential for can expose other layers to
problems with these tanks excessive head of liquid unless
because of the importance of strict dimensional control all layers are seal welded at each
interface
on each band; if too slack relative to the inner layer the
band will not give proper support and if too tight it will
either collapse the inner layer or put excessive load on
the vertical seam. Differential expansion and contraction
of the layers must also place non-uniform loads on the
layers. The potential for separation of the layers is significant particularly if the layers are
not sealed together
4.2.1.2 Fabricators
The following two well established and well equipped fabricators are known to use this
method of making tanks but because of the relative simplicity of this form of fabrication
many such tanks will be fabricated by small workshops often without the required design
capabilities or quality procedures. Implementation of draft standard prEN 12573 should limit
the size of tank which can be fabricated by this method; the use of thicker sheet material
may be considered but should be approached with caution, weld failure will not be
restrained by secondary layers as it is currently with multi-layer tanks
Niplast Thermoplastic Engineering supply tanks in HDPE, PP and PVDF for a broad
range of chemicals in many different industries. Their publicity material indicates that tanks
can be fabricated in capacities up to 100m3 by this method.
Forbes: Forbes specialises in GRP tanks and also fabricates sheet formed tanks but they
generally limit the size of unreinforced tanks made by this method to 15m3. They
emphasise the importance of sealing the layers so that liquid cannot penetrate to the
interfaces.
15
4.2.2 Helix winding
4.2.2.1 Background
A band of material is extruded on to a rotating steel mandrel and is moved parallel to the
mandrel axis to form a spiral winding of plastic which builds up into a cylinder, Figure 12.
The thickness is built up with more windings and graduated along the cylinder length (or
tank height). Pre heat is applied to the cylinder to ensure adequate softening to allow
fusion of subsequent windings.
Figure 12
After cooling the cylinder is removed from the
mandrel, the ends are trimmed and welded to a
base. Originally all bases were made by winding a
cylinder of material, slitting lengthways and
opening it out to form a ‘flat’ sheet. This method is
still used for less critical duties. Where the
proposed duty is potentially hazardous the base is
made of flat sheet. The cylinder is welded to the
base initially using a rod weld; where required by
the customer this rod weld is tested for integrity
using a water test before the vessel is emptied,
Figure 13
ã Allibert Ltd
16
4.2.2.2 Fabricators
The following UK fabricators are known to use the helix winding process
Allibert: for many years the only fabricator operating the helix winding process in the UK
they have over 25 years experience in this form of construction. Allibert also have factories
in France and Spain and have an operation in China. Their publicity material indicates the
capability of fabricating tanks in HDPE and Polypropylene in capacities up to 68m3.
Chem-Resist: operate a slight variation on the basic process and use a sheet formed
inner layer on to which the outer layers are extruded. This is understood to avoid the need
to heat the steel mandrel and allows the use of a less robust mandrel. The surface of the
sheet formed inner lining is heated to allow windings to fuse. This inner ‘lining’ is not
included in the structural calculations. A relative newcomer to helix winding Chem-Resist
has set up a similar operation in Prague.
4.2.3.1 Background
Rotational moulding is a method of moulding complex hollow items with low residual stress
and is used to make tanks, drums and bulk containers.
• powdered or liquid
plastic is placed in a
hollow, shell-like mould
at room temperature
• the mould is heated to
the melting temperature
Figure 14 ©Institut for Procesteknik, Danmarks Tekniske Universitet 1996 of the plastic and
rotated around two
rotational axes so that the different surfaces of the mould come down into the ‘pool’ of
plastic lying in the bottom of the mould. The plastic adheres to its inside surface building
up the desired wall thickness distribution.
• When all the plastic has coated the mould and sufficient time has elapsed for the plastic
to become well consolidated, the heating is stopped and the cooling stage begins.
• Once the plastic has cooled to the point where it is sufficiently solid to retain its shape,
the mould rotation stops and the plastic product is removed from the mould. Finishing
operations, such as removal of runners, are minimal.
Cross-linkable grades of polyethylene accounts for over 90% of the total tonnage of plastics
used in rotomoulding. This is mainly because the powder flow, melt flow and thermal
stability of this material are ideally suited to the process. Any shortcomings in mechanical
properties such as stiffness can be readily overcome through the use of geometrical
features, some of which are unique to rotomoulding. When using reactive or cross-linking
17
plastics, the cooling stage is not necessary to consolidate the part but the moulding must of
course be reduced to a temperature at which it can be handled easily.
The design of rotationally moulded tanks is covered by ASTM 19988. The vessels are
moulded in one piece including the nozzles and consequently there are no welds that can
act as weak points in other tanks. When designing above ground chemical storage
systems, there are a number of design considerations used in determining which materials
of construction are most suitable including:
4.2.3.3 Fabricators/suppliers
Tycon Containers Ltd: suppliers of tanks up to 45 m3 capacity. The tanks are fabricated
in the USA where this has been the preferred method of making thermoplastic tanks for
many years. The superior mechanical properties of cross-linkable grades of polyethylene
allow a thinner wall than say a helix wound tank. It is understood that by careful
manipulation of the moulding process it is possible to graduate the wall thickness up the
tank wall: it is also possible to design in moulded ribs which impart greater stiffness.
TD Ltd: are understood to supply rotationally moulded containers but have not been
approached for further information.
Appropriate lifting arrangements must be incorporated at the design stage e.g. a slinging
band, lifting eyes or lugs. Tanks must NOT be lifted by nozzle connections
Handling and transport of thermoplastic tanks should only be assigned to experienced
contractors. All reputable fabricators give guidance on lifting and transportation. Appendix
2 is the author’s compilation of this guidance.
Base - flat bottomed tanks should only be installed on a smooth, flat base, manufacturers
typically specify a deviation from nominal level of ±3 mm.m-1. Care must be taken to
ensure that the base is free of debris.
Flat polyethylene sheet pad - suppliers of rotationally moulded tanks offer a flat
polyethylene sheet as a ‘pad’ between the support base and the tank
Most fabricators now offer integral bunding systems with their tanks and, even where a tank
is installed in a pre-existing bund, it is essential that procedures are in place to ensure that
18
the bund cannot collect excessive rainwater that will reduce the bund capacity in the event
of a spillage but can also cause tanks to ‘float’. Bunds should be checked regularly for
product leakage.
Where windloading calculations require the use of tie-down bolts these must be correctly
installed using the brackets or steel girdle attached to the tank.
19
• facilities should be available to the customer to inspect and/or witness any tests at any
stage in the manufacture subject to reasonable notification.
Welders should always be suitably trained and qualified and tested regularly (6 monthly)
with weld qualification tests to maintain standards. There are now formal qualification
procedures for plastic welding offered for example by TWI (The Welding Institute).
20
dye penetrant systems could affect the material would need to be assessed before
this could be accepted as recommended procedure
∗ circumferential measurements - have been used over a period of ca 20 years to
assess helix wound tanks on the basis that the design life is based upon creep
properties and over loading would cause a permanent strain, it could also be
anticipated that degradation and/or overheating could cause the material to loose
strength and the tank to expand. Although it is unlikely that a helix wound tank will fail
by creep rupture of the walls this has proved a useful technique to give assurance
when accompanied by visual examination and destructive examination of a tank
sample
∗ high voltage spark testing is applied to welded joints. It is essential that there is an
earthed path behind or within the joint which the spark discharge can leak to. It is very
doubtful that all penetrating defects will be identified by this method
6.2.2 Frequency
Inspection intervals should be selected on the basis of identifying anticipated degradation or
deterioration in normal service before it causes a serious problem.
∗ In the author’s opinion the basic requirement should be regular visual examination by
the plant operator. At least one operator of hydrofluoric acid tanks, Rhodia Ltd,
Avonmouth, includes this on a daily checklist of plant equipment. Experience of the
author and the fabricators of helix wound tanks is that any problems will initiate as a
weep at the base weld; even when large sheet formed tanks have ‘failed’ it has been
observed as a small jet of liquid from the base weld. If identified soon enough the
tank can be drained before a more serious leak develops. Consequently ‘daily’ would
be the recommended interval for this operator’s inspection particularly for tanks on
hazardous duties. Remotely located tanks, those on less hazardous duties and tanks
which are located in bunds should at least be checked visually before each filling
operation.
∗ A formal inspection by a competent person will also be required periodically to give
assurance of fitness for service and the following guidelines may assist with selecting
an appropriate interval
• HSE Guidance note PM75 for GRP tanks recommends inspection within the first 2
years service’ to confirm the correctness of the design and fabrication’. In the authors
opinion this should be applicable to tanks of all materials of construction. In most
cases if there are problems they will be apparent within this time-scale although in the
worst cases failures have occurred within the first 12 months. The findings from this
initial assessment should then allow selection of an on going inspection programme.
Assuming no problems are identified it may be appropriate to postpone the next
formal inspection to a defined event such as:
∗ end of design life
∗ before a change of use
∗ before a change in operating conditions
∗ following an unscheduled event
21
• Where the duty is particularly hazardous, when the design life has been exceeded or if
an inspection method such as circumferential measurement is used, a more regular
inspection will be appropriate.
6.2.3 Acceptance standards
Any inspection procedure must establish acceptable and unacceptable defects and criteria
appropriate to the findings of that method. The following guidance may assist with defining
these standards but ultimately it will depend on the experience and judgement of the
inspecting authority.
• Visual Inspection:
∗ No leakage or penetrating defects
∗ Presence of cracking in welds or wall may require immediate retirement or more
frequent inspection
∗ Degradation of material of construction may require immediate retirement or more
frequent inspection
• Circumferential measurement
∗ DVS2205 applies a strain limit of 1%; for the HF tanks operated by Laporte Fluorides
a maximum allowable increase in circumference of 0.75% has been used. The
evidence of an early piece of work was that considerably higher strains were possible
before rapid failure occurred 5.
22
7 SERVICE EXPERIENCE – FAILURES AND DEFECTS
7.1 Fabrication defects
7.1.1 Weld defects
The base welds of very early helix wound tanks were formed by a ‘block’ welding technique.
These were a potential weak spot and one tank from the author’s experience did suffer a
premature leak within days of going into service. This type of weld was replaced by
extrusion welding and the water test described earlier was implemented to prevent
recurrence. Although there may still be tanks with this form of weld in service they will be
ca 20 years old and due for retirement.
7.1.2 Delamination of wall
material
Although care is taken with
helix wound tanks to ensure
complete bonding of windings
there will be occasions where
surface chilling, inadequate
roller pressure, extrudate
temperature variations could
cause a small area to remain
unfused with the adjacent
layer. This often occurs at the
extremities of a winding and
the top and bottom sections of
a cylinder will be always be
discarded. However tanks can
go into service with inherent Figure 16 Delamination of base material (source: sms)
poorly fused zones which could
open slightly during service,
Figure 16, Figure 17. These
areas are likely to be very small
and are not considered by the
fabricator be a problem. Newer
tanks would have a pressed
sheet base rather than helix
wound sheet for hazardous
duties and this problem should
not occur.
23
7.1.3 Winding defects
The internal surface finish of
Allibert helix wound tanks is
dependent upon the condition
of the steel mandrel. Minor
defects can arise for example
if a cold spot or surface
contamination prevents good
adhesion to the mandrel,
Figure 18, Figure 19.
Some early tanks were made
on mandrels where the
opening panel of the
collapsible mandrel did not
close fully leaving a step
discontinuity on the surface.
This can give the visual
appearance of a crack and it Figure 18 Winding defect (probably a cold spot on the mandrel)
could act as a stress raiser (source: sms records)
although the fabricators and
the author have not
reported any problems
arising from this defect.
24
7.2 Chemical Degradation
The following three examples of chemical incompatibility between HDPE and the stored
product indicate that caution is required with strongly oxidising species and HDPE.
7.2.1 Nitric Acid
A number of helix wound
HDPE tanks were used to
hold 70% nitric acid; HDPE is
known to be incompatible with
this strength of nitric acid.
The tanks were
decommissioned after ca 10
years service and examined
destructively. Preliminary
visual inspection confirmed
the potential for attack in the
form of transverse cracking of
the base weld, Figure 20.
Subsequent destructive
examination showed severe Figure 20 Transverse cracking of base weld (source: sms records)
cracking of the base which
could have resulted in a
catastrophic failure, Figure 21.
Mechanical testing confirmed
the degradation of the
material which had been in
contact with nitric acid.
25
7.2.2 Sodium hypochlorite
A large number of HDPE tanks
have been supplied for sodium
hypochlorite and although not
generally regarded as
incompatible there have been a
number of occurrences which
emphasise the need for early
inspection. A number of tanks
used for electrolytically
generated low strength (~1%)
sodium hypochlorite are
reported by Allibert Ltd have
failed by cracking of the base
weld after a short service period;
failure has been attributed to the
existence of ‘nascent’ chlorine in Figure 22 Hairline cracks on lower seam (source: sms records)
the solution rather than the
normal hypochlorite solution
where the chlorine has reacted
with sodium hydroxide. Sheet
formed tanks storing 12%
hypochlorite have been retired
after ca 5 years service
following weeping from the
base weld and an inspection
which revealed cracking of the
horizontal seams. This
cracking was found to be worse
towards the top of the tank
where cracking extended into
the parent material, Figure 22,
Figure 23. Figure 23 Cracking of seam and sheet towards top of tank
(source: sms records)
A process vessel made of
HDPE and used to blend cleaning solutions based on hypochlorite has also ‘failed’ after a
short service life with extensive cracking around base to wall seam and at baffle support
bracket welds and at inlet nozzles in the roof. Major repairs were required within 2 years
and recent inspection by the author after a further ca 2 years service has shown that
cracking has reinitiated and it is likely that the service life will be considerably less than the
10 year design life. Premature failure has been attributed to the combination of
hypochlorite and detergent; in addition, excessive stress has almost certainly been imposed
on the base weld by uneven support of the base, (see 7.5.1, lifting of base)
26
7.2.3 Sulphuric Acid
Allibert have reported cracking in a concentrated
(96%) sulphuric acid tank fabricated by them:
cracking was noted in the tank barrel immediately Location of cracks
above the base weld and as radial cracks around
an outlet nozzle, Figure 24. No further
documentation was available on this failure.
27
7.3.2 Hydrofluoric Acid, leaching of additives of HDPE
One operator of Allibert HDPE HF tanks in SE Asia, reports leaching of a component of the
HDPE resin resulting in coloration and precipitation in the acid 10. This phenomenon was
noted in ‘higher strength acids’ and these tanks will generally operate at high ambient
temperatures. Cases have been investigated by the author of ‘contamination’ of HF by anti
oxidant and heat stabilisers used in rotational moulding grades of HDPE for small
packages. It is likely that a similar effect has caused the observed contamination of HF in
these tanks although it has not been reported, to the author’s knowledge, elsewhere with
similar tanks. These additives are mainly present as processing aids to minimise
degradation of the HDPE resin during fabrication and will have minimal effect on behaviour
during service.
28
7.5 Operational ‘defects’ & failures
7.5.1 Lifting of wall
a) Older tanks in particular were fabricated with
relatively thin bases - it has been noted after
more than 15 years service that there is a
tendency for the tank walls to lift from the
plinth; this has been attributed to shrinkage of
the tank body which has been confirmed by
long term circumference measurements, Figure
27, Figure 28. This effect presents potential for
foreign bodies to become trapped under the
tank and also will put additional loading on the
internal seal weld and the adjacent material.
b) it is understood that non-uniform longitudinal
thermal contraction of the cylinder following
removal from the mandrel can result in a
similar phenomenon in which the tank base is
not fully supported
Figure 28
Lifting of base
caused by shrinkage of Figure 27 Lifting of tank walls
barrel?
(source: sms records)
c) a relatively new vessel has been noted in which
additional loading placed on support of the base was inadequate; this has a similar
internal weld effect to lifting of the wall and the stress generated has
contributed to initiation of cracking at the toe of the
internal base/wall in a situation which is prone to
environmental stress cracking.
29
7.5.2 ‘Scuffing’ of base
Slight movement of the tank as it
expands and contracts appears to
cause ‘scuffing’ of the base, Figure
29. This is exacerbated if the
concrete plinth becomes
roughened due to acid attack from
spillage or dilution of fume. Taken
to the extreme there is potential for
the base to be perforated. Newer
tanks having thicker bases are
less susceptible to this form of
degradation. There is some scope
for improvement of concrete
plinths to minimise this type attack,
e.g. by the use of a polyethylene
pad under the tank or a
polyethylene lining system which is
Figure 29 ‘Scuffing’ of tank base
mechanically keyed into the
concrete. Sand and bitumen are not recommended as a bed for the tank to sit on as they
are relatively unstable and their gradual loss will result in uneven stressing of the tank base.
30
7.5.4 Perforation of tank base by level sensor
A tank leaked soon after reassignment to a new duty following a number of years on the
original service without problem. The base was punctured by a level indicator sensor rod
which should have been suspended ~15 cm above the base; it was assumed that it had
dropped onto the tank base but another option might be thermal expansion and contraction
of the tank and flexing of the base - installation of sensors and pipework from the top of the
tank should be considered carefully
7.5.5 Splitting of seal welds
Over a period of time movement of the tank and associated pipework etc. can result in
failures of nozzle connections particularly to the roof.
31
7.5.6.2 Internal bulging
Access of liquor to the wall
laminates will force any air in the
wall to the top of each ‘cavity’.
Exposure to the full head of
liquid will compress this air
which, while balanced by the
liquid in the tank, causes no
disruption, Figure 32. When the
tank is drained the liquor will not
flow out of the wall quickly
enough to relieve this pressure
and wall laminates can be
exposed to a higher pressure
than they can accommodate
causing bulging.
7.5.6.3 Uneven discoloration
of sheet tanks in PVDF:
A phenomenon, which has been
noted and gave cause for
concern was uneven
discoloration of a sheet formed
vessel used to handle high purity
Figure 32 Internal bulge in large, sheet formed phosphoric acid tank HF. It is almost certain that
material from different sources
(source: sms records)
was used to make up a
‘patchwork’ sheet to fabricate the tank. Absorption of HF into the plastic reacted with
processing compounds and impurities in the plastic to give each panel a different colour.
This problem was more cosmetic than hazardous there being no apparent degradation of
the material.
32
7.6 Effects of under and overpressure
Thermoplastic tanks are not designed to
Figure 33
resist vacuum or pressure. Failures
have occurred when transfer from
tankers is made using compressed air -
the release of air at the end of transfer
can impose a sudden pressure wave. If
venting is inadequate it can cause the
failure of the roof to wall weld. Tanks
which are attached to a fume extraction
system have be subjected to partial
vacuum if vents are blocked. Sufficient
suction can easily be created when the
product is drawn from the tank if venting
is inadequate. In a case reported by
Allibert a vent on a tank was deliberately
blocked to reduce the release of
unpleasant smells; drawing off the
product caused collapse of the roof.
A tank which was deliberately
Figure 34
tested after about 17 years service Collapse of tank roof by exposure to partial vacuum
failed at the wall/roof weld at an
excess pressure of ca 0.1 bar. The
roof was repaired and the tank
subjected to partial vacuum by
sealing all vents and running water
out of the bottom outlet; collapse of
the roof and failure of a connection
occurred very quickly, Figure 33,
Figure 34.
The absence of welded seams
makes rotationally moulded tanks
exceptionally tolerant of under
pressure. Supplier’s literature
(Tycon Ltd) shows a tank which,
subjected to vacuum, has collapsed (source: sms records)
and distorted severely but ‘sprung
back’ into its original shape when
the vacuum was relieved.
7.7 Installation issues
A number of issues arise from mishandling during transport and installation.
• inadequate slinging caused a large tank (~50m3) to be dropped from a low loader onto its
bottom knuckle causing the base seam to fail (source sms records)
33
• problems have been reported where debris was left on the support plinth. Although no
leakage occurred considerable cost was incurred in downtime to lift and reinstall the
tank. It is generally recommended that these tanks are installed on a flat concrete base
which needs to carefully skimmed to ensure a smooth even surface (source Rhodia Ltd).
• tanks which have been stored temporarily on sites pending final installation and have
been left standing on rough ground, again with potential for damage to the base. Even
though an empty tank may weigh relatively little compared with a full one the net weight
of a large tank with a wall thickness of ~72 mm will be at least 2t.
• tanks (GRP) have been stored temporarily without consideration of the need for securing
against wind loading and were blown over and around the site before installation,
resulting in considerable impact damage, (source sms records).
• tanks have been installed on adequately designed steel stillage with thick, fully
supportive, flat plates under the tank. However consideration was not given to the
consequences of spillage into the bund and the legs of the stillage were rapidly corroded
causing it to partially collapse to the extent that the tank was leaning against its
neighbour. Excessive movement also put the associated pipework under increased
stress which could result in failure of the feed pipe, (source sms records).
34
• modification to tanks have been noted, e.g. to
install new nozzles or change flanges. For
example nozzle support fillets have been
shortened to fit a new flange, this has resulted
in ‘nicking’ and scoring of the nozzle, Figure
35.
• welding of material which has absorbed
species such as HF or HCl needs to be
approached with caution - the heat of welding
will cause the absorbed species to be
released and may result in a porous weld as
well as presenting a potential hazard to
personnel.
35
∗ tanks on duty with
hydrogen peroxide have a
simple water deluge
system to protect them in
the event of an exothermic
peroxide decomposition,
Figure 36. This might be a
reasonable addition to any
tank which contains a
hazardous products if fire
or exothermic reaction is a
real risk. Such products
could include concentrated
sulphuric acid which will
generate huge quantities
of heat if diluted.
Figure 36
ã Allibert Ltd
36
8 CONCLUSIONS
Thermoplastic storage tanks are widely used throughout the chemical industry and, as has
been illustrated in this report, in spite of a number potential deficiencies in design,
specification and fabrication they are generally tolerant of abuse. It is the authors’ opinion
that large sheet formed tanks are the most likely to suffer a sudden failure and helix wound
tanks very unlikely to suffer a major uncontrolled release. The potential deficiencies in the
design of these tanks should largely be controlled by application of a new design code and
as new tanks replace older ones the potential for major problems will diminish. It is
essential that users and specifiers of tanks are aware of this standard as it is likely that
some fabricators will only continue to offer DVS2205 as their design code. As with any tank
periodic inspection is required and again operators need to be aware of this requirement.
Both these aspects may be best addressed in the form of an HSE guidance note similar to
PM75.
There is also considerable scope for further examination and consideration of methods of
inspecting or monitoring thermoplastic tanks. Any such method must consider potential
failure mechanisms and be able to give assurance of identifying any weaknesses in a tank
in time.
9 RECOMMENDATIONS
• A guidance note similar to PM75 for GRP tanks would be of benefit to users of these
tanks to highlight the best practice in terms of specification, design, installation and
operation
• All current users of tanks on hazardous duties should consider the requirements to
inspect and assess their tanks within the early years of service, preferably within the first
2 years. Tanks which would be considered hazardous are:
∗ those containing chemicals with a known potential for incompatibility, e.g.
• sodium hypochlorite
• conc. nitric acid
• conc. sulphuric acid
∗ tanks which have been reassigned to a duty different to that originally specified
∗ tanks on any duty which are greater than 15m3.
∗ sheet formed tanks which are greater than two layers thick
∗ any tank which has a history of leaks or damage, e.g. during transport, installation or
commissioning
37
APPENDIX 1
Acknowledgements
Allibert Gloucester
Chem Resist Ltd Dewsbury
Forbes Plastics Downham Market
Niplast Stockport
Tycon Stockport
Users
Rhodia Avonmouth
Laporte plc
Laporte Fluorides Rotherham
Soderec International Pierrelatte, France
UCE St Fromond, France
Laporte Electronics Riddings
38
APPENDIX 2
Operating practice
• All storage installation should be subject to a regular periodic inspection by site
personnel - e.g. one site handling HF makes a daily inspection of its facility - the
maximum interval should be one week. Many, if not most, problems with thermoplastic
tanks will start as a weep or small leak or a bulge and regular inspection by site
personnel will identify problems before they can develop.
• All storage tanks should be subject to a formal inspection or assessment within the first 2
years service as prescribed in HSE PM75. This should be applicable to all materials in
order to verify the correctness of the design and installation
• This formal inspection should include an internal visual inspection where the contents are
known to be potentially incompatible, e.g. sodium hypochlorite, hazardous if released, or
if the tank is larger than 15m3 where sudden release of the contents could cause
disruption to adjacent tanks and facilities
• Risk assessment for all maintenance work adjacent thermoplastic tanks should take into
consideration possible impact on the tanks, e.g. weld spatter, grinding sparks, cutting
and gouging by steel work
• Tanks should not be reassigned to alternative duties without assessment of their design
and condition
39
Installation
3
• Third party assessment of tanks on critical duties and of capacities larger than 15m is
considered essential
• Tanks should be inspected fully during fabrication and after installation prior to service
• Prior to installation, tanks should only be stored in a prepared location and should be
restrained to prevent wind damage - this could often be provided by putting a small
amount of water in the tank BUT it is essential that the tank is drained before it is moved
• The installed tank
∗ should be fully supported by its plinth
∗ should be secured by its tie bolts or other supporting structure
∗ should not be reassigned to a new duty without independent assessment
∗ the suitability of large, (>15m3) sheet formed tanks is questionable and early
assessment and inspection is recommended
• If located in a bund there should a be system to ensure that the bund cannot loose
capacity by filling with rainwater
• Tank tie down points can be relatively weak and only provided location points and
restraint in the event of excessive wind loading, they will not prevent an empty tank from
floating if the bund fills with water.
• Transfer pipework other than feeding the tank should not pass over the top of a tank in
case the tank floats and disrupts it
• Tank installation should be supervised and/or carried out by fully trained personnel
• No thermoplastic tank should be allowed to operate beyond its design life without an
assessment
• Fire protection should be considered for large tanks in vulnerable areas
• Water deluge systems should be considered for tanks which could be exposed to
exothermic reactions by decomposition or dilution of their contents
40
APPENDIX 3
References
1
DVS 2205 “Deutscher Verband für Schweißecnik”
Design and construction of thermoplastic storage tanks, 1975
2
HSE PM75 Design, Specification and Inspection of GRP Tanks
3
http://www.materials.drexel.edu/Welcome.html
4
BS 2654 Manufacture of Vertical Steel Storage Tanks
5
BS 4994 Design and Construction of GRP Tanks
6
Design of fluid storage tanks from polypropylene, K Forbes et al, British Chemical Engineer Oct 1970
7
NACE Publication 5A171- Materials for Receiving, Handling and Storage of Hydrofluoric Acid
8
ASTM D1998 - Upright Polyethylene Tanks
9
The effect of hydrofluoric acid on polyethylene, Dr R Krzikalla, Kunstoffe(1978) 68,(12) 825-828
10
Communication with M Page, Laperth Ltd, Bangkok
41