Professional Documents
Culture Documents
Abstract:
The use of FRP and dual laminates has grown a lot for manufacturing vessels and piping during the
last years, due to its high corrosion resistance and cost-effectiveness compared with the expensive
metallic alloys required for handling highly corrosive fluids. However, the criteria and procedures for
designing FRP vessels all around the world are very different, giving birth to very different design
codes for every geographical zone. As a result, it is complicated comparing these design codes to
choose the one which fits better to our needs.
The purpose of this study is comparing four different design codes for FRP vessels (AD-Merkblatt N1
[1], BS 4994:1987 [2], EN 13121 [3] and ASME RTP-1 Ed. 2007 [4]). To do so, five different tanks
(two storage tanks, a scrubber and two horizontal pressure vessels) have been designed using all these
codes. Next, the equations used for calculating every part of the tanks, and the results obtained have
been analyzed and compared, showing the most important strong and weak points of every code, plus
their similarities and differences. The full study called “Global FRP standards”, has been published by
Materials Technology Institute, Inc, (MTI) in 2015.
The use of FRP and dual laminates has grown a lot during the last years, due to its high
corrosion resistance and cost-effectiveness compared with the expensive metallic alloys
usually required for handling highly corrosive fluids.
Usually big companies are operating worldwide, and every country requires applying
different codes or standards for designing FRP vessels. In addition, many different suppliers
can be chosen for manufacturing these vessels, being even possible choosing a manufacturer
from a part of the world for supplying material to any other part of the world. In addition, the
company technical staff is based in world headquarters usually, having many restrictions for
getting personally involved in the manufacturing process of these type of vessels. Considering
all these questions, the design, manufacturing, inspection and transport of a FRP vessel can
become an important challenge for any multinational company.
Several design codes are available for FRP vessels, having important differences among them.
As a result, it is very important choosing the right one to avoid problems during all the
manufacturing and installation process. The aim of this study called “Global FRP Standards”
[5] published by MTI (Materials Technology Institute), is comparing four different design
codes for FRP vessels, so the technical differences among them can be established. Once
these differences are known, it will be easier making the right decision when choosing the
design code we are going to apply.
The four design codes for FRP vessels that has been compared on the study are shown next:
To compare them, five different hypothetical vessels have been designed, using every one of
the design codes:
The geometry of the two storage tanks is equal (they only have different design conditions
and contain different fluids). The same thing happens for the horizontal pressure vessels.
Once the design of all the hypothetical vessels has been made through all the design codes,
the equations used for the design and the results obtained have been analyzed and compared,
showing the main differences among them, and their main weak and strong points.
Comparison of the safety factors
First of all, we are going to compare how the safety factor is obtained by every design code:
AD-Merkblatt N1 [1]
In this case, the safety factor is obtained as a product of four subfactors (A1 to A4). These
subfactors are related to the next questions:
BS 4994:1987 [2]
In this design code, the safety factor is obtained as a product of five subfactors (K1 to K5),
being calculated as shown next:
K1 - This subfactor depends on the manufacturing procedure chosen for the vessel
(hand lay-up, filament winding or spraying)
K2 - Models the behaviour and chemical strength of the resin against the attack of the
fluid contained on the vessel
K3 - This subfactor depends on the difference between the design temperature for the
vessel and the heat distortion temperature of the resin (HDT)
K4 - Models the behaviour of the FRP laminate against cyclic loads (fatigue
behaviour)
K5 - This subfactor depends on the type of curing applied to the laminate (depends on
whether post-cure is applied or not)
In this design code, the safety factor is obtained again as a product of five subfactors (A1 to
A5), depending on the next questions:
A1 - This subfactor gives a measure of the statistical dispersion of the real mechanical
properties obtained from proof tests on laminate samples
A2 - Models the behaviour and chemical strength of the resin against the attack of the
fluid contained on the vessel
A3 - This subfactor depends on the difference between the maximum allowable
temperature for the resin when containing the fluid considered, and the heat distortion
temperature of this resin (HDT)
A4 - Models the behaviour of the FRP laminate against cyclic loads (fatigue
behaviour)
A5 - Models the long term behaviour of the FRP laminates depending on the laminate
composition and the type of effort applied (tensile or bending)
In that case, the safety factor is a fixed figure which depends on the design and testing
conditions, as shown in the next table:
When critical service conditions are considered, the safety factors must be increased a 25 %.
Notes:
Once the safety factors have been compared, it is very important analyzing how the
mechanical properties of the laminates are obtained through every design code.
AD-Merkblatt N1 [1]
This design code proposes just one method for obtaining the mechanical properties of a
laminate, being a pure empirical procedure which consists on performing proof tests on
laminate samples. The results obtained from these tests must be treated statistically, so the
average mechanical properties can be obtained. As a result it is not possible to estimate the
mechanical properties before performing the tests, being required manufacturing samples for
obtaining their properties. The methods for performing these proof tests and for treating
statistically the results are not very clearly explained on the design code, doing more
complicated its application.
Finally, it is important highlighting that the procedures given by this design code for
obtaining the maximum allowable strength of the laminate, does not incorporate the
maximum allowable strain. As a result, this strain limit is not implicitly included in the
maximum allowable stress, and two independent verifications will have to be performed
during the design process, one for the stress and another more for the strain applied into the
laminate.
BS 4994:1987 [2]
This design code calculates the mechanical properties adding the typical strength and modulus
given for every glass reinforcement layer stacked in the laminate. These typical strength and
modulus for every type of glass reinforcement are given in tables included on the code.
The maximum allowable stress of a laminate is obtained as a function of the safety factor and
the maximum allowable strain for the resin considered. As a result, the strain limit is
implicitly included in the maximum allowable stress for the laminate, so just the strength
verification is required during the calculation.
BS 4994:1987 [2] allows calculating the tensile properties (stress and modulus), but it has not
a procedure for obtaining the bending properties (which might be considered when wind and
seismic loads are applied into a vessel).
In that case, the mechanical properties can be obtained through two different procedures,
depending on whether we are applying a basic or advanced design:
Basic design:
This procedure is exactly the same used by BS 4994:1987 [2], so the typical strength and
modulus of every glass reinforcing layer must be added for obtaining the mechanical
properties of the whole laminate. Again, the laminate strength will depend on the safety factor
and the maximum allowable strain for the resin considered, being the strain limit implicitly
included in the maximum allowable stress.
In that case there is a method available for obtaining the flexural modulus of the laminates,
based on laminate theory. This method is very similar to the one proposed by ASME RTP-1
Ed. 2007 [4].
Advanced design:
The second procedure proposed by EN 13121 [3] is very similar to the one given by AD-
Merkblatt N1 [1]. It consist on obtaining the mechanical properties empirically, performing
proof tests on laminate samples, treating later the results statistically for obtaining the average
mechanical properties. Compared with AD-Merkblatt N1 [1], the procedure is better
explained and fully included on the code, being easier applying it.
ASME RTP-1 Ed. 2007 [4]
Finally, this design code includes three different procedures for obtaining the mechanical
properties:
According to chapter M-3, the mechanical properties can be also obtained through
lamination analysis. In that case laminate theory is applied for obtaining the properties
of the whole laminate, adding the mechanical properties of every layer, which can be
previously obtained from figures, as a function of their glass content and the resin
modulus.
The last procedure available is using design basis laminates, which is again an
empirical method that obtains the mechanical properties after treating statistically the
results obtained from proof tests on laminate samples (being very similar to the
methods given by AD-Merkblatt N1 [1] and the advanced design on EN 13121 part 3
[3])
All the design codes compared requires verifying the cylindrical shell under three different
conditions:
These conditions are obtained by all design codes as a combination of the next typical loads:
The calculation procedure for a cylindrical shell is quite similar for all the design codes, but
two different approaches can be distinguished among them:
BS 4994:1987 [2] and EN 13121 part 3 [3] are applying an iterative procedure (exclusive for
FRP design codes), which requires performing the next steps:
On the other hand, AD Merkblatt N1 [1] and ASME RTP-1 [4] subpart 3A calculate the
minimum thickness required for the shell as a function of the laminate maximum allowable
stress and the load combination considered. This procedure is shared with the design codes
for metallic vessels.
For the study, a laminate composed by a combination of filament winding and woven roving
tissues has been considered for the shell of all the hypothetical vessels calculated. The
cylinders have been divided into rings (H-1, H-2, H-3, etc.) at different heights, allowing
having different thicknesses along them. The next thicknesses have been obtained for the
different hypothetical vessels:
Table 2: Wall thickness for storage tank under mild service conditions:
Note: H-1 is the ring placed at shell bottom, while H-4 is placed at shell top.
Table 3: Wall thickness for storage tank under severe service conditions:
Note: H-1 is the ring placed at shell bottom, while H-4 is placed at shell top.
Table 4: Wall thickness for scrubber:
Note: H-1 is the ring placed at shell bottom, while H-12 is placed at shell top.
Table 5: Wall thickness for horizontal pressure vessel under mild service conditions:
Note: H-1 is the ring placed at left side of the shell, while H-12 is placed at right side.
Table 6: Wall thickness for horizontal pressure vessel under severe service conditions:
Note: H-1 is the ring placed at left side of the shell, while H-12 is placed at right side.
For all the vessels, it can be seen that the minimum thicknesses are obtained by both
BS 4994:1987 [2] and EN 13121 [3]. Next we can find AD-Merkblatt N1 [1] and ASME
RTP-1 [4], which are giving higher and very similar thicknesses. Considering that the
parameter that is governing most of the vessel’s design is the vacuum, it can be said that the
procedure given by BS 4994:1987 [2] and EN 13121 part 3 [3] (which allows dealing with
stepped shells with no uniform thickness), allows obtaining smaller thicknesses than AD
Merkblatt N1 [1] and ASME RTP-1 [4].
All the design codes except AD-Merkblatt N1 [1] include a procedure for calculating a flat
bottom, based on the typical solution for a round plate with fixed edges, submitted to an
uniform load. These procedures allows obtaining the bending moment on the plate (so the
stress can also be calculated), and the maximum deflection at the center of the plate.
All the three design codes which include this procedure, are also giving a limit for the
maximum deflection allowed at the plate center. This limit is different depending on the
design code considered:
For BS 4994:4987 [2] and EN 13121 part 3 [3], the deflection limit is equal to 1.5
times the flat bottom thickness, for avoiding the second level effects.
For ASME RTP-1 Ed. 2007 [4], the deflection limit is equal to 1.5 % of the vessel
diameter
AD-Merkblatt N1 [1] does not include any method for calculating a flat bottom, so a
procedure given by any other design code must be borrowed, or it will be required adapting to
FRP the typical solution for round plates with fixed edges submitted to uniform load.
The procedures explained above are applicable for the central part of the flat bottom, but it
will be also required calculating the flat bottom knuckle. Depending on the design code
considered, a different method must be applied:
BS 4994:1987 [2] applies a very simplified procedure which consists on increasing a
50 % the loads obtained at the bottom of the cylindrical shell. The knuckle laminate
will have to be strong enough for withstanding these increased loads.
EN 13121 part 3 [3] includes a very detailed procedure for calculating the loads
applied into the knuckle, depending on the ratio between the transition radius and the
vessel diameter.
ASME RTP-1 Ed 2007 [4] only requires calculating the flat bottom knuckle when the
vessel’s diameter is higher than 144 inches. In that case, subpart 3B (stress analysis)
must be applied for calculating this part of the vessel. When vessel diameter is lower
than 144 inches, the same thickness can be applied for both the central part and the flat
bottom knuckle.
A flat bottom made by hand lay up procedure has been considered for all the hypothetical
vessels calculated, obtaining the next thicknesses:
Table 7: Wall thickness obtained for the flat bottom of the storage tanks
It can be seen that the results obtained for both vessels are equal, because the design
conditions are very similar for them. The biggest thickness is given by ASME RTP-1 [4] due
to the higher safety factor applied by this design code. The smallest thickness is given by EN
13121 part 3 [3], due to the smaller safety factor applied.
All the design codes require performing two different verifications for calculating a domed
head:
Again, two different approaches can be found on the design codes compared:
BS 4994:1987 [2] and EN 13121 part 3 [3] apply an iterative procedure (exclusive for FRP
design codes), which requires performing the next steps:
AD Merkblatt N1 [1] and ASME RTP-1 [4] subpart 3A calculate the minimum thickness
required for the domed head as a function of the maximum allowable stress for the laminate
and the internal/external pressure considered. This procedure is shared with the design codes
for metallic vessels.
It also has to be highlighted that BS 4994:1987 [2] and ASME RTP-1 [4] subpart 3A give an
uniform thickness for the whole domed head. On the other hand, AD Merkblatt N1 [1] and
EN 13121 part 3 [3] calculate independently the head knuckle and crown, giving different
thickness for both parts (usually higher stresses are applied in the knuckle, so a more
optimized design can be obtained).
A hand lay up laminate has been considered for the domed heads of all the hypothetical
vessels, obtaining the next thicknesses:
The minimum thicknesses for the domed heads are given in general by BS 4994:1987 [2].
Bearing in mind that the equations used are similar for all the design codes, these results are
due to the different constants used on these equations.
Calculation of reinforcing pads for nozzles
Two parameters must be calculated for a reinforcing pad around a nozzle, its thickness and its
diameter. Two different approaches for calculating these parameters can be found on the
compared design codes:
BS 4994:1987 [2] and ASME RTP-1 [4] subpart 3A calculate the loss of strength
caused by the hole made on the wall for placing the nozzle. The reinforcing pad
thickness will have to be enough to balance this loss of strength.
Two different approaches are also applied for calculating the reinforcing pad diameter,
depending on the design code considered:
When AD Merkblatt N1 [1] and EN 13121 part 3 [3] are applied, the reinforcing pad
diameter is obtained as a function of the vessel’s diameter and the vessel’s wall
thickness (it is independent of the nozzle diameter)
When BS 4994:1987 [2] and ASME RTP-1 [3] subpart 3A are applied, the reinforcing
pad diameter is obtained as a function of the nozzle’s internal diameter only
Being the procedures applied by the design codes so different, the results obtained for the
reinforcing pads are also very different. It has to be highlighted that the reinforcing pad
diameter is bigger for small nozzles when AD Merkblatt N1 [1] and EN 13121 part 3 [3] are
applied. The opposite effect is obtained for big nozzles, being the reinforcing pad diameter
smaller when AD Merkblatt N1 [1] and EN 13121 part 3 [3] are applied.
Only EN 13121 part 3 [3] has an specific method for calculating a support skirt for vertical
vessels. This procedure is very detailed and even considers the effects of the different pressure
and temperature applied to both parts, vessel and the support skirt.
AD Merkblatt N1 [1] has not an specific method for calculating this part of a vessel, so for the
study it has been applied the method given by AD Merkblatt S 3/1 [6] for metallic vessels,
adapting it to FRP particularities.
BS 4994:1987 [2] and ASME RTP-1 [4] subpart 3A do not have any method for calculating
the support skirt neither. For solving this question during the study, the support skirt has been
considered as an additional part of the cylindrical shell, but submitted to the specific loads
applied at this part (no pressure or vacuum and room temperature have been considered for
the support skirt).
Only EN 13121 part 3 [3] has an specific method for calculating an horizontal vessel
supported by saddles. The procedure included in this design code is based on Zick’s method
(used by most of design codes for metallic vessels), but adapted to FRP particularities.
For AD Merkblatt N1 [1] , it has been applied the procedure given by AD Merkblatt S 3/2 [7]
for metallic vessels, but adapted to FRP properties and safety factors.
For the horizontal vessels designed according to BS 4994:1987 [2] , the procedure given by
the design code BS 5500 [8] for metallic vessels has been applied, adapting it to FRP (as it
has been made for AD-Merkblatt N1 [1]). The procedure given by BS 5500 [8] is also the
Zick’s method for horizontal metallic vessels.
Finally, when the horizontal vessels have been designed according to ASME RTP-1 [4]
subpart 3A, it has been applied the method given by ASME VIII div. 2 [9] for horizontal
metallic vessels. This procedure is again the Zick’s method (the same used by BS 5500 [8]
and EN 13121 part 3 [3]).
Once the calculation procedures used by the four design codes have been compared, it is also
interesting comparing the manufacturing cost of the hypothetical vessels designed during the
study. The next figure shows these costs compared:
Figure 1: Cost of every calculated vessel depending on the design code applied
Note: The hypothetical vessels are represented by the next codes on the figure:
It can be seen that the most cost-effective design code is EN 13121-3 [3] for all the
hypothetical vessels. On the other hand, the less cost-effective design code is ASME RTP-1
[4] when subpart 3A is applied. It is important highlighting that cost-effectiveness of ASME
RTP-1 [4] can be improved applying subpart 3B, but this particular case has not been
considered for this study.
Finally the figure shows that the cost differences among the different design codes are bigger
when the vessel becomes more complicated (scrubber) and when the design pressure is
increased (horizontal pressure vessels).
Once the vessel has been manufactured, it will be required registering it, but the legal
requirements to do so are very different depending on the place where the vessel is going to
be installed. Some examples of the rules that must be applied for vessels in different countries
are shown next:
These certifications and markings usually require the approval of an independent third party
for the design, manufacturing and testing process. It is very difficult finding inspectors and
engineers trained on FRP inspection and calculation outside France and the USA, so dealing
with these third parties could become an important challenge.
Two typical examples about how the design code must be chosen for simplifying later the
vessel’s certification and marking are described next:
Example No. 1: An USA company needs a FRP vessel for one of its plants in a country inside
the European Union:
In that case the best option would be designing the FRP vessel according to EN 13121-3 [3],
which is an harmonized code in Europe. As a result, using this design code you will have
direct conformity with the Pressure Equipment Directive (PED) 97/23 [10], and it will be
easier preparing the documents required for putting the CE marking. In most cases a Notified
Body (third party authorized by the local government to perform these tasks), will have to
verify the whole design, manufacturing and testing process, before allowing the manufacturer
to put the CE marking onto the FRP vessel.
Example No. 2: An USA or European company needs a FRP vessel for one of its plants in
China:
When this case is considered, the best option will be choosing ASME RTP-1 [4], mainly
because it can be understood more easily by engineers non skilled on FRP design. As a result,
it will be easier for the third party involved in the SELO [13] certification understanding the
calculation note they have to revise.
Final conclusions
The final conclusions of this study can be summarized showing the strong and weak points
detected on every design code:
Table 9: Strong and weak points for design code AD-Merkblatt N1 [1]
It has very detailed procedures for designing The procedures given for obtaining the safety
the next parts of a FRP vessel: factor and the mechanical properties of a
laminate are complicated, and require always
Cylindrical shells (especially when performing proof tests on laminate samples.
they are submitted to vacuum)
Reinforcing pads for nozzles The inspection procedures and acceptance
Domed heads criteria for the laminates are not included
directly on the code.
It is not well known by most of engineers
outside Germany.
There is not a procedure for calculating the
shell and domed head against local loads, so a
procedure from any other design code must
be borrowed to do so.
It does not include procedures for calculating
some typical parts of a FRP vessel as:
Flat bottom
Support saddles (for horizontal
vessels)
Support skirts (vertical vessels)
Table 10: Strong and weak points for design code BS 4994:1987 [2]
It has procedures for designing most parts of The same safety factor is used for both long
a typical FRP vessel. and short term conditions, so the cost-
effectiveness is reduced.
It includes clear and simple procedures for
obtaining the safety factors and the It has been developed and issued in 1987, so
mechanical properties of a laminate. some parts are obsolete nowadays.
It is very well known all around the world The procedure included for calculating the
because it has been one of the first design flat bottom knuckle is too simple, so the cost
codes issued for FRP vessels. effectiveness is reduced.
It includes a detailed procedure for There is not a procedure for calculating the
calculating the effect of local loads in domed flexural properties of a laminate, and they are
heads and shells. usually required when bending moments are
applied.
It does not include procedures for calculating
some typical parts of a FRP vessel as:
Support saddles (for horizontal
vessels)
Support skirts (vertical vessels)
Table 11: Strong and weak points for design code EN 13121 part 3 [3]:
It contains procedures for designing all the It requires considering more design cases for
typical parts of a FRP vessel. the vessel calculation when the design code is
fully applied.
It includes clear and simple procedures for
obtaining the safety factor and the mechanical It is the most complex design code among the
properties of a laminate. four compared, so it is more difficult to deal
with it for engineers non skilled in FRP.
It includes a detailed procedure for
calculating the effect of local loads applied Some minor typing mistakes are still
on the cylindrical shells. remaining on the design code text, so they
will have to be corrected on future editions.
It is the most cost-effective design code
among the four compared. It is not very well known outside Europe
It is harmonized with the Pressure Equipment because its first edition was issued very
Directive (PED) 97/23 EC, so it gives you recently (2008).
direct conformity with this Directive. As a
result, it allows putting the CE marking
required in Europe in a very easy way.
Table 12: Strong and weak points for design code ASME RTP-1 Ed. 2007 [4]:
It contains procedures for designing most When the design temperature for the vessel is
typical parts of a FRP vessel. higher than 180 ºF (84 ºC), it is mandatory
performing proof tests on laminate samples at
Subpart 3A of this design code is the most this design temperature. These tests will
simple among the four codes compared. As a allow verifying that the mechanical properties
result, it is easier to deal with it for an considered at this temperature are correct.
engineer non skilled in FRP calculation.
Subpart 3A of this code is the less cost-
The procedure for obtaining the safety factor effective among the four codes compared.
is very simple compared with the rest of
design codes. Subpart 3B (stress analysis) is quite
complicated when compared with the rest of
A very detailed and complete procedure for design codes. As a result it is more
obtaining the mechanical properties of a complicated dealing with it, even more for
laminate (based on laminate theory) is engineers non skilled in FRP.
included.
In many cases it requires putting the ASME
Subpart 3B allows dealing with almost any stamp on the FRP vessel. However, outside
possible geometry or design case considered the USA and Canada it is very difficult
for a vessel. It also allows improving the finding FRP manufacturers allowed to put
cost-effectiveness of the code. this stamp.
It is a design code very well known all
around the world, especially in America and
the Middle East.
Acknowledgments
We want to thank the next persons and institutions, without their cooperation it will never
have been possible completing this study:
The Materials Technology Institute (MTI) for their sponsorship that has allowed
performing the Global FRP standards project
Mr. Dale Keeler for helping on the correction of calculation notes and final report
Mr. Barry Greene and Mr. Randy Scheel for coordinating all the project during more
than two years
Mr. Pradip Khaladkar for proposing and sponsoring this project within MTI
References