On The Design of The Tubesheet and Tubesheet-To-Shell Junction of A Fixed Tubesheet Heat Exchanger

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

ARTICLE IN PRESS

International Journal of Pressure Vessels and Piping 83 (2006) 714–720


www.elsevier.com/locate/ijpvp

On the design of the tubesheet and the tubesheet-to-shell junction


of a fixed tubesheet heat exchanger
K. Behsetaa,b,, S. Schindlerb,c
a
No 34, 7th Ave Nsre (Gisha) Street Tehran 14 467, Iran
b
Institute for Pressure Vessel and Plant Technology, Vienna University of Technology, Gusshausstr. 30/329, A-1040, Vienna, Austria
c
Technischer Überwachungsverein Austria, Krugerstr. 16, A-1015 Vienna, Austria
Received 31 January 2006; received in revised form 3 August 2006; accepted 3 August 2006

Abstract

For the (mechanical) design of an existing fixed tubesheet heat exchanger, a C2-Hydrogenation reactor in a petrochemical plant,
various code solutions are compared with each other and with a Finite Element solution based on the Direct Route in Design by Analysis
(EN 13445-3, Annex B). The codes and standards used in the investigation are ASME Section VIII, Division 1 and EN 13445-3, Clause
13 and Annex J.
The ASME VIII/2 and TEMA codes are not appropriate for this design.
The ASME VIII/1 and EN 13445-3 Clause 13 approaches are similar. Differences in maximum permissible pressures result partly from
different nominal design stresses.
The modern EN 13445-3 Annex J approach, being based on limit analysis theory, leads to very different, much more efficient results.
The Direct Route in Design by Analysis confirms the EN 13445-3 Annex J results, but gives, at the same time, clear insight into the
behaviour of the whole structure and the various maximum permissible pressure limiting details.
r 2006 Elsevier Ltd. All rights reserved.

Keywords: Heat exchanger design; Design by Analysis; Direct Route

1. Introduction ASME Code, Section VIII, Div. 1,2, [2], the European
Standard for Unfired Pressure Vessels EN 13445 by the
Heat exchangers are frequently called the workhorses in European Committee for Standardization (CEN) [3], and
process and petrochemical plants, and more than 65% of the Standards of the Tubular Exchanger Manufacturers
these are (shell-and-tube type) tubular heat exchangers. Association (TEMA) [4].
Tubular heat exchangers exemplify many aspects of the Large heat exchangers and heat exchangers made of
challenges in the (mechanical) design of pressure vessels. expensive materials are cost effective components in
Their design requires a thorough grounding in several industrial plants. For these, a thorough understanding of
disciplines of mechanics, and a broad understanding of the code differences is of paramount importance.
interrelationships between the thermal and the mechanical Results of code comparisons exist and a few are
performance of heat exchangers [1]. In a liberal legal published, see e.g. [5,6]. However, having been performed
framework, for the mechanical design of heat exchangers before changes in ASME with regard to the effective elastic
the requirements in several codes and standards may be constants of the perforated region of tubesheets, the
used. Amongst these are the widely known code of the comparisons of ASME/EN are based on ASME results
American Society for Mechanical Engineers (ASME), that are not up-to-date.
Unique design advantages of one or other of the
Corresponding author. No 34, 7th Ave Nsre (Gisha) Street Tehran 14 aforementioned codes are few. However, the smaller
467, Iran. nominal design stresses of ASME and TEMA compared
E-mail address: behsetaeq@yahoo.com (K. Behseta). to EN 13445 for low alloy steels in the low to medium

0308-0161/$ - see front matter r 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijpvp.2006.08.002
ARTICLE IN PRESS
K. Behseta, S. Schindler / International Journal of Pressure Vessels and Piping 83 (2006) 714–720 715

temperature range, and for austenitic steels in the whole Table 1


temperature range, are well known. Nominal design stresses used
For fixed tubesheet heat exchangers, of interest here, the Code ASME VIII/1 (MPa) EN 13445-3 (MPa)
ratio, X, of the axial tube-bundle stiffness to the effective
tubesheet bending stiffness is an important parameter: Part
Tubesheet 138.0 145.0
h i1=4 Tubes 108.2 121.0
X ¼ 6ð1  n2 ÞN t E t et ðd t  et ÞD20 =ðLE  e3p Þ Shells 138.0 150.5

with n the effective Poisson’s ratio of the perforated accordance with the requirements of the Direct Route in
tubesheet region, Nt the number of tubes, Et the modulus Design by Analysis (DBA-DR), EN 13445-3 Annex B.
of elasticity of the tube material, et the nominal tube The tubesheet thickness to tube pitch ratio of the heat
thickness, dt the nominal outside diameter of the tubes, D0 exchanger considered is slightly smaller than 2.0, slightly
the equivalent diameter of the outer tube limit circle, E  the outside the limit of applicability of the ASME VIII/2
effective modulus of elasticity of the perforated tubesheet requirements. ASME VIII/2 being not attractive for this
region, L the tube length between the tubesheet inner faces, kind of heat exchanger, this code division is not considered.
and ep the tubesheet thickness. The ratio of the (axial) tube-bundle stiffness to the
For the same nominal design stress the following applies: effective tubesheet stiffness of this heat exchanger, at 12.46,
For X less than about 2 [6] (or 3 [5]) ASME VIII/1 and being fairly large, TEMA results in tubesheet thicknesses
EN 13445-3 Clause 13 render slightly thicker tubesheets that are known to be overly thick, see above, and,
than TEMA. For X larger than 5 [6] (or 9 [5]) TEMA therefore, TEMA is also not considered in this comparison.
renders larger thicknesses than ASME and EN 13445-3 In order to make the comparison unbiased, it is based on
Clause 13, with no unique order but there are remarkable ASME VIII/1 nominal design stresses for the ASME VIII/
differences between ASME VIII/1 and EN 13445-3 Clause 1 calculations, and on nominal design stresses deduced
13 results. For X in the intermediate range all three render from ASME material (strength) parameters with EN
comparable results, in most cases in the (descending) 13445-3 Clause 6 requirements for EN 13445-3 Clause 13
thickness order TEMA, ASME VIII/1, EN 13445-3 and Annex J calculations, and also for the FE result
Clause 13. evaluation, adapted for the partial safety factor format
ASME VIII/2 does not take into account the staying used in EN 13445-3 Annex B. To emphasise the goal of the
action of the tubes. This makes this code division comparison, the comparison of code equations, and to be
unattractive for fixed tubesheet heat exchangers. able to show the influence of thermal stresses by means of
The limit analysis theory based approach in EN 13445-3 stress ratios, the nominal design stresses of Load Case 7—
Annex J offers, for fixed tubesheet heat exchangers, clear design pressures and design temperatures—have also been
advantages for all ratios X in comparison with the other used in the decisive, thickness determining Load Case 3—
three approaches discussed above. design pressures but without thermal expansion. The
ASME VIII/1 and EN 13445-3 Clause 13 are very similar corresponding nominal design stresses are compiled in
in their approach. Both are based on classical shell and Table 1, see also the respective detailed calculations. The
plate theory, and both treat the perforated region of the differences in the nominal design stresses according to
tubesheet as a solid (unperforated) plate of the same ASME VIII/1 and EN 13445-3 are, for this example and
thickness but with effective elastic constants. These are this load case, small (enough).
intended to render an equivalent overall behaviour of the
perforated region and the equivalent unperforated plate. 2. The heat exchanger
Differences that are not caused by different nominal design
stresses are the result of the treatment of the unperforated The heat exchanger considered in this paper is an
rim of tubesheets as a solid ring, without distortion of the existing C2-Hydrogenation reactor in an olefin plant. The
cross-section, in ASME VIII/1, and the assumption of a design dimensions and design data, relevant for the
tubesheet totally perforated, without an unperforated rim, investigation, are:
in EN 13445-3 Clause 13.
Part of this paper deals with just another comparison
Shell inner diameter—4250 mm1,2 (4294 mm)1,3
between ASME VIII/1 and EN 13445-3 Clause 13, but for
Channel outer diameter—4230 mm1
a very large fixed tubesheet heat exchanger with outside
Shell thickness—70 mm1,2 (32 mm)1,3
diameter 4430 mm, maximum allowable tube-side pressure
Channel thickness—90 mm1
40 bar, self-weight approximately 1.56 MN, and practically
Tube outer diameter—57 mm
the whole cross-section tubed. This comparison is com-
plemented by another with the attractive EN 13445-3 1
Nominal.
Annex J approach, and by a comparison with the results of 2
As variant SA 537 Cl.2 is also considered in the investigation.
3
an elaborate Finite Element (FE) model, with evaluation in Fluid temperatures.
ARTICLE IN PRESS
716 K. Behseta, S. Schindler / International Journal of Pressure Vessels and Piping 83 (2006) 714–720

∅4230 90
∅58
135

2
R1
∅57
29
∅4250 70

1200

∅4288 32

Fig. 1. Geometry detail (dimensions in mm).

Tube thickness—2.9 mm1


Tubesheet thickness—135 mm1
Corrosion allowance—3 mm (on tubes 0)
Equivalent diameter of outer tube limit circle
(OTL)—4136 mm
Design pressure tubeside—40 bar
Design pressure shellside—10 bar
Design temperature tubeside—4/190 1C4
Design temperature shellside—4/145 1C4
Temperatures for determination of material Fig. 2. (a) FE model: Full model. Part of whole model, (b): FE model:
properties: Tubesheet of Full model.
Tubeside shell—145 1
Shellside shell—190 1C 3. Code results comparison
Tubesheet—167.5 1C
Tubes—167.5 1C The results of the calculations according to ASME VIII/1,
Materials: EN 13445-3 Clause 13 and Annex J, are given in Table 2
Shells: SA 516 Gr.705 for Load Case 3 (Design pressures, no thermal displace-
Tubesheet: SA 266 Cl.2 ment) and in Table 3 for Load Case 7 (Design pressures
Tubes: SA 344 Gr.1 and thermal displacements according to design tempera-
tures).
Details of the tube-to-tubesheet and the tubesheet-to-shells Because of the special design, especially the small
joints are shown in Fig. 1. The tube layout is shown in unperforated rim of the tubesheet and the shellside shell
Fig. 2, the isometric of the tubesheet in the so-called Full consisting of three courses of different thickness, comments
model. on the calculations are required.
The shellside shell consists of one middle course of Shellside courses of different thickness are directly
32 mm nominal thickness, and two thicker ones near the addressed in ASME VIII/1 and EN 13445-3 Clause 13.
tubesheet, nominal thickness 70 mm, the upper one of The given shell lengths of the thicker courses near the
1200 mm length, the lower of 800 mm length. tubesheets are, at 1200 and 800 mm, longer than the
required length (for the application scope):
4
Courses near tubesheet. pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
5
Course between thicker courses near tubesheets. l s ¼ 1:4 ðDe þ eÞes ¼ 753:3 mm.
ARTICLE IN PRESS
K. Behseta, S. Schindler / International Journal of Pressure Vessels and Piping 83 (2006) 714–720 717

Table 2
Stress ratios for Load Case 3

Stress ratioa Code

Load case 3 in % ASME VIII/1 (%) EN 13445-3 Cl.13 (%) EN 13445-3 Annex J (%)

Tubesheet
Bending 296.9/275.9 ¼ 108 288.2/290 ¼ 99 0
Shear 57.3/110 ¼ 52 65.9/116 ¼ 57 0
Shellside shell
seqnb
Membrane 41.6/138.0 ¼ 30 41.2/150.5 ¼ 27 39
Membrane+bending 98.0/275.9 ¼ 36 105.8/451.5 ¼ 23 —
seqfc
Membrane 96.0/138.0 ¼ 70 93.8/150.5 ¼ 62 90

Values for EN 13445-3 Annex J are optimised for the tubesheet only, not for tubesheet and shells. The ratios given for this approach are not stress ratios
but ratios of stress resultants, and there is no limit requirement for membrane plus bending stress resultants in the shells, only for the membrane stress
resultants.
a
Ratio of relevant stress to allowable value (of Load Case 7).
b
At junction to tubesheet.
c
Far from junction.

Table 3
Stress ratios for Load Case 7

Stress ratioa Code

Load case 7 in % ASME VIII/1 (%) EN 13445-3 Cl.13 (%) EN 13445-3 Annex J (%)

Tubesheet:
Bending 253.3/551.8 ¼ 46 145.7/435.0 ¼ 33 0
Shear 33.1/110 ¼ 30 33.3/116 ¼ 29 0
Shellside shell
seqnb
Membrane 49.3/138.0 ¼ 36 51.7/150.5 ¼ 34 39
Membrane+bending 96.8/552.0 ¼ 18 84.8/451.5 ¼ 19 —
seqfc
Membrane 113.8/138.0 ¼ 82 119.5/150.5 ¼ 79 90

Values for EN 13445-3 Annex J are optimised for the tubesheet only, not for tubesheet and shells. The ratios given for this approach are not stress ratios
but ratios of stress resultants, and there is no limit requirement for membrane plus bending stress resultants in the shells, only for the membrane stress
resultants.
a
Ratio of relevant stress to allowable value.
b
At junction to tubesheet.
c
Far from junction.

Different thicknesses are not directly addressed in the other For Load Case 3 the values for Fm according to the two
code variants. However, the lengths of the shell courses code variants are close to each other but in Load Case 7
being longer than the ‘‘usual’’ influence length of rotational they differ by approximately a factor of 1.8.
symmetric bending disturbances, the other code variants That the stress ratios for Load Case 3 are large, and even
are considered to be directly applicable as well. larger than unity, does not mean that the design is not
The unperforated tubesheet rim is very small. The basic admissible. Formally and correctly for this load case, larger
assumptions in the models for ASME VIII/1 and EN nominal design stresses are allowed, and feasible—the
13445-3 Clause 13 are, with the exception of those for this nominal design stresses for ambient temperatures. If there
unperforated rim, equal. Thus, similar results according to are no thermal displacements then there is no temperature
these codes could be expected. The results do agree well, larger than ambient.
with one exception—the bending stress in the net cross- EN 13445-3 Annex J uses a totally different approach
section of the tubesheet in Load Case 7, see Table 3. The based on limit analysis theory.
cause of this discrepancy is that although the equations in The calculations according to this code variant showed
both codes are very much alike, in ASME VIII/1 the factor that the tubesheet plus tube bundle is self-supporting, the
Fm depends on the actions, whereas in EN 13445-3 Clause tubesheet does not need the support of the shells, and no
13 it depends only on the geometry and the elastic reaction between tubesheet and shell junction is a statically
properties of the tubesheet and shells, not on the actions. admissible solution. The results of this solution, which is
ARTICLE IN PRESS
718 K. Behseta, S. Schindler / International Journal of Pressure Vessels and Piping 83 (2006) 714–720

optimal with regard to the tubesheet, are included in the


tables. This solution is not the optimal one with regard to
the whole structure considered—tubesheet, tubes, and both
shells. Better solutions, which would decrease the large
stress ratio for the thin part of the shell-side shell can be
obtained, with increase of the tubesheet ratios. Further
iterations resulted in an overall optimal solution with ratios
29%, 70%, and 69%, for tubesheet bending, tubesheet
shear, and shellside shell membrane force far from the
junction.
ASME VIII/1 and EN 13445-3 Clause 13 indicate
wrongly that the tubesheet is the weak part. EN 13445-3
Annex J shows clearly that the tubesheet is self-supporting,
given that the tubes and the tube-to-tubesheet junctions are
strong enough, and that it is the thin course of the shellside
shell that is the weakest part.
This thin shell course is directly addressed in ASME
VIII/1 and in EN 13445-3 Clause 13, and, therefore,
checked in a routine application of this approach. The
check can be incorporated in applications of EN 13445-3
Annex J, but can be missed easily in routine applications.
The usual check as a cylindrical shell under internal
(shellside) pressure is only complementary, and does not
account for the large longitudinal stresses in the shell
course, due to the large pressure in the channel.
EN 13445-3 Annex J requires for heat exchangers with
fixed tubesheets a fatigue check, for which a conservative
approach is given. For the heat exchanger considered here,
a reactor in a petrochemical plant, fatigue is not a problem.
Results of a fatigue design check are given in Section 5, in
the discussion of the DBA-DR results.

4. The FE model
Fig. 3. (a) FE model: Whole Sector model. (b): FE model: Tubesheet of
The heat exchanger is symmetric in the parts of interest Sector model.
here with regard to two meridional planes, and is
considered to be symmetric with regard to a cross-section upper side of the tube-side shell (the channel), with
in the middle between the tubesheet. Pre-checks with a resultant equal to the pressure force at the end.
model representing one eighth of the heat exchanger, called
in the following the Full model and shown in Fig. 2,
showed this FE model to be too time consuming. Two 5. The design checks
other, simpler FE models were checked against this full
model, and one of these, the so-called Sector model, shown In accordance with EN 13445-3 Annex B, four design
in Fig. 3, was used for all design checks. checks have to be considered here. These are the gross
This FE model consists of 42482 elements and 82238 plastic deformation design check (GPD-DC), the progres-
nodes. The shells, tubesheet and a part of the tubes are sive plastic deformation design check (PD-DC), the fatigue
modelled with 3D-volume elements SOLID 45. Tubes not design check (F-DC), and the instability design check
modelled with volume elements are replaced by vertical (I-DC), for all relevant load cases. Of these design checks
link elements, 12 per full bore. The upper nodes of the link only the first three are of interest here, the I-DC is a
elements are the corresponding upmost nodes of the bores straightforward comparison of calculated compressive
in the tubesheet. All nodes at the meridional sides of the axial stresses in the tubes with the corresponding allowable
tubesheet and shells are assigned with symmetry boundary ones, with requirements in other clauses of the standard.
conditions. 3D-pipes, link elements, and the lower side of For the GPD-DC, in accordance with the standard’s
shellside shell, are fixed in the vertical direction. To take requirements, the pressures for Load Case 3 have been
into account the pressure force on the upper end of the heat increased proportionally, from zero up to the design
exchanger, vertical forces were applied on all nodes of the values and further. In the thermal stress free sector model
ARTICLE IN PRESS
K. Behseta, S. Schindler / International Journal of Pressure Vessels and Piping 83 (2006) 714–720 719

a linear-elastic ideal-plastic constitutive law and associated


flow law is used with first-order-theory, stress-free initial
state, Mises
pffiffiffi yield condition, and with yield stresses given
by 1:2  3=2 times the nominal design stresses listed in
Table 1. The factor 1.2 results from the p relevant
ffiffiffi partial
safety factors in the standard. The factor 3=2 has to be
applied because of the use of the Mises’ yield condition in
the model and Tresca’s in the standard’s requirements for
the GPD-DC.
At 240% of the design pressure the FEA calculations
failed to converge, and the last convergent solution
obtained at a pressure of 198% of the design pressure
showed a maximum equivalent total plastic strain well
below the limit (for equivalent structural strains) of 5%.
There result permissible pressures of 198/1.2 ¼ 165% of
the design pressures. Total plastification occurred at the Fig. 4. Maximum principal stress (normal to weld joint direction), linear-
upper end of the channel shell, with little plastic flow in the elastic model, Action State M.
tubesheet.
The tubesheet and the tubesheet-to-shell junction being
of interest, a second investigation has been performed with For the F-DC, EN 13445-3 recommends, taking account
a stronger channel shell material–SA 537 Cl.2, with of the strong influence of stress changes on the number of
nominal design stress (at design temperature) of allowed cycles, use of realistic action values, not design
188 MPa, and yield stress of 195.4 MPa. The FEA values. Therefore, the allowable number of cycles was
calculations failed to converge in this case at 260% of the determined for cyclic actions from Action State 0, up to
design pressures, with the last convergent solution at 222% Action State M, and back to Action State 0. Action State
of the design pressures with maximum plastic equivalent M is given by pressures of 85% of the maximum
total strain well below the limit (for equivalent structural permissible pressures and temperatures equal to the design
strains) of 5%. Critical sections are now the net cross- pressures, and where the above, in connection with the PD-
section of the tubesheet (between the bores), and the thin DC, described thermal stresses are included in the reaction
shell-side shell course. The maximum permissible pressures of the structure.
are 222/1.2 ¼ 185% of the design pressures, and the Fatigue is decisive in the weld joint between the channel
corresponding ratio of design pressures is 54%. This ratio shell and the tubesheet, see Fig. 4. For Action State M a
is smaller than the value obtained by use of EN 13445-3 maximum principal stress normal to the weld joint of
Annex J, after overall optimisation. 411.9 MPa was obtained. For the relevant fatigue class
In the PD-DC, shakedown to purely elastic behaviour FAT 63, the calculation of the allowable number of cycles
was proven for the cyclic load case from Action State 0, according to Clause 18 of EN 13445-3 gave a result of 2806
given by zero pressures and ambient temperatures up to cycles, a value larger than the specified number of cycles to
Action State A, given by pressures of 185% of the specified be considered to occur under reasonably foreseeable
design pressures and temperatures equal to the design conditions, specified to be 300.
temperatures, and back to Action State 0. In the resulting
reaction of the structure stationary thermal stresses 6. Conclusions
corresponding to a stationary temperature distribution,
with metal surface temperatures as specified, were included. For this special heat exchanger, application of ASME
For use in Melan’s shakedown theorem a suitable VIII/1 and EN 13445-3 Clause 13 leads to unnecessarily
residual stress field was created by a pressurisation/ thick tubesheets. Application of EN 13445-3 Annex J
depressurisation cycle of the FE model used in the GPD- results, for all parts, in thicknesses which are much more
DC, with maximum pressures close to the limit analysis economical, but the thin shell-side shell course requires
pressures of the model. additional calculation steps that are easily missed in
The superposition of the resulting residual stress field routine applications. The fatigue check, required in this
with the stress field determined in the corresponding model approach, leads to overly conservative results if the
with an (unbounded) linear-elastic constitutive law for requirements in this annex are followed, and details of
actions of Action State A rendered a stress field for which the geometry at the hot spot cannot be taken into account
Mises equivalent stress exceeded nowhere the relevant directly. The importance of these details is easily over-
limit, given in the standard by the 0.2% proof stress at the looked.
calculated metal temperature. Therefore, according to The setting-up of the FE model is time-consuming, but
Melan’s shakedown theorem, shakedown to linear-elastic the results may be worth the effort: The maximum
behaviour is proven for this action cycle. permissible pressures, as per the Direct Route in Design
ARTICLE IN PRESS
720 K. Behseta, S. Schindler / International Journal of Pressure Vessels and Piping 83 (2006) 714–720

by Analysis, are even larger than the ones obtained with an References
EN 13445-3 Annex J approach, definitely the best of the
three Design by Formula approaches discussed, and all [1] Singh KP, Soler AI. Mechanical design of heat exchangers and
pressure vessel components. Acturus Publishers, Inc.; 1984.
details can be incorporated. The various design checks give
[2] 2004 ASME boiler and pressure vessel code, Section VIII, Division
a clear picture of the limit behaviour and of critical areas 1,2–rules for construction of pressure vessels. New York: ASME Press;
important for in-service inspection (ISI)—critical regions 2004.
and required ISI intervals. [3] EN13445-3: European standard for unfired pressure vessels—part 3:
design. European Committee for Standardization (CEN), 2002.
[4] TEMA: Standards of the tubular exchanger manufacturers association,
eighth ed. Tubular Exchanger Manufacturers Association, Inc.; 1999.
Acknowledgement [5] Rao KR, editor. Companion guide to the ASME boiler & pressure
vessel code, vol. 2. New York: ASME Press; 2002.
[6] Osweiller F. Tubesheet heat-exchangers: new common design rules in
This paper is extract of parts of the dissertation prepared UVP, CODAP and ASME. In: Proceedings ninth international
at the Vienna University of Technology. Usage of ANSYS conference on pressure vessel technology (ICPVT-9), vol. 1, Sydney,
under the university license is acknowledged. 2000. p. 43–55.

You might also like