You are on page 1of 8

i-sss^sJ?

- ^

Prcc^fiiiiie V< ,H K°iip}<

Basis of the Tubesheet Heat


Exchanger Design Rules Used in
the French Pressure Vessel Code
F. Osweiller For about 40 years most tubesheet exchangers have been designed according to the
Centre Technique des Industries
standards of TEMA. Partly due to their simplicity, these rules do not assure a safe
Mecaniques, heat-exchanger design in all cases. This is the main reason why new tubesheet design
Senlis, France rules were developed in 1981 in France for the French pressure vessel code CODAP.
For fixed tubesheet heat exchangers, the new rules account for the "elastic rotational
restraint" of the shell and channel at the outer edge of the tubesheet, as proposed
in 1959 by Galletly. For floating-head and U-tube heat exchangers, the approach
developed by Gardner in 1969 was selected with some modifications. In both cases,
the tubesheet is replaced by an equivalent solid plate with adequate effective elastic
constants, and the tube bundle is simulated by an elastic foundation. The elastic
restraint at the edge of the tubesheet due the shell and channel is accounted for in
different ways in the two types of heat exchangers. The purpose of the paper is to
present the main basis of these rules and to compare them to TEMA rules.

Introduction luaiawf mt*i llMt»Of DUItlt

Many process industries are increasing their use of tubular


heat exchangers in preference of plate exchangers because tu-
bular exchangers are better able to withstand service involving
both high temperature and high pressure. Nevertheless, one
a «4 Bh ta u,
must not forget that a tubesheet heat exchanger is not only an

w*-
apparatus devoted to transfer heat from one fluid to another,
but also a pressure vessel.
For a long time pressure vessel codes, and especially ASME,
=lc u
have provided rules for the design of the various parts of a •itfLLona M i l
heat exchanger (shell, tubes, heads, flanges, nozzles, etc.), but
none for the essential part of the exchanger: the tubesheet. It Fig. 1 Scheme of a tubular heat exchanger
is essential because three main parts of the vessel—the shell,
the tube bundle, and the channel head—are connected to it
(Fig. 1). Thus, in addition to pressure and thermal expansion for the design and fabrication of tubular exchangers. This well-
loads, the tubesheet is subjected to the reactive loads from known standard [I] has been widely used in the past 40 years,
these three parts. and still is.
The lack of tubesheet design rules in Section VIII of the These rules have the merit of long satisfactory experience;
ASME Code in the 1930s was due largely to the complexity but due to their simplicity, they often lead to overthickness or
of the design problem and the fact that Section VIII was still sometimes (for high pressure especially) to underthickness, and
in an early stage of development. may become unconservative. These disadvantages increase
In order to remedy this deficiency, the American manufac- nowadays as chemical and power industries need larger and
turers decided 50 years ago to create the Tubular Exchanger larger exchangers operated at higher and higher pressures and
Manufacturers Association (TEMA) to develop standard rules temperatures.
That is why CETIM [2] in connection with SNCT [3] and
SPEICHIM (SPIE BATIGNOLLES Group) decided to de-
velop, under the responsibility of the author, French design
Contributed by the Pressure Vessels and Piping Division and presented at the rules insuring greater safety and efficiency, thanks to a better
Pressure Vessels and Piping Conference, Nashville, Tennessee, June 17-21,1990,
of THE AMERICAN SOCIETY OF MECHANICAI ENGINEERS. Manuscript received by representation of the tubesheet and its connections with the
the PVP Division, August 9, 1991. shell, head, and tubes.

124 / Vol. 114, FEBRUARY 1992 Transactions of the ASME

Copyright © 1992 by ASME


Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jpvtas/28333/ on 07/25/2017 Terms of Use: http://www.asme.org/a
The first purpose of this paper is to present the main basis for the design of the tubesheet, instead of four previously, and
of these rules for fixed tubesheet, floating head, and U-tube the method is iterative.
heat exchangers. In 1969, Gardner [10] improved his floating tubesheet method
The second purpose is to show, through a comparison with by considering the unperforated annulus at the tube-
TEMA rules, why CODAP rules lead to a safer and more sheet periphery and proposed a direct formula. This method
efficient design than TEMA. was used later by various pressure vessel codes for the design
of floating head and U-tube exchangers: ISC [11] in 1973,
BS 5500 [12] in 1976, CODAP [13] in 1982, and more recently,
Historical Background for U-tube exchangers only, by ASME [14]. All these methods
The first rules concerning the design of tubesheet heat ex- have the drawback of considering the tubesheet either simply
changers were developed for the first time in the U. S. in 1941 supported or clamped at its edges, which obliges the designer
by the Tubular Exchanger Manufacturers Association (TEMA). to make an aribitrary choice between these two extreme cases
This standard proposed a simple formula to calculate the [16].
tubesheet thickness for U-tube and floating head heat ex- The third generation of design methods will overcome this
changers. This semi-empirical method was based on the max- drawback by taking into account the degree of edge fixity of
imum stress in thin circular plates subjected to uniform the tubesheet. This more realistic approach was proposed by
pressure. Rules for fixed tubesheet heat exchangers appeared Galletly [15], as early as 1959, by considering the tubesheet as
30 years later based on Gardner's work [4]. The original TEMA "elastically clamped" at its periphery. This method was
method [1] did not account for the tubes which stiffen the adopted in CODAP in 1982 for fixed tubesheet heat ex-
tubesheet, nor for the holes which weaken it. The recent 1988 changers.
Edition [5] tries to overcome this deficiency. In 1975, the ASME decided to set up a Special Working
Second-generation rules were initiated in 1948 by Gardner Group in order to implement new rules for the design of tube-
[6], who was the first to set up the basis of a more rational sheet heat exchangers, based on a more scientific basis than
approach for the design of floating tubesheet heat exchangers TEMA. The first such rules for the treatment of U-tube heat
by accounting for the support of the tubes and the weakening exchangers appeared in the Winter 1982 Addenda to ASME
effect of the holes in the tubesheet. In 1952, Gardner [4] adapted Section VIII, Division 1 as nonmandatory Appendix AA [14].
this method to the design of fixed tubesheet heat exchangers. The method proposed by Gardner [10] in 1969 was selected as
Simultaneously, and independently, Miller [7] proposed a a basis for these rules.
method which seems different in its presentation, but which Design rules for fixed tubesheet heat exchangers are in prep-
in fact is the same. aration [21] by the Working Group, based on the work of
The Miller method was adopted by the British Standard 1515 Soler and Singh [16]. A first draft of these rules is available,
[8] in 1965 and the Gardner method by the STOOMWEZEN but not yet published.
[9] in 1973. Using this approach requires about 15 parameters In these new methods, about 40 parameters are involved

Nomenclature

ME = reactive bending moment at tubesheet


C0 = coefficient depending on parameters X0 periphery
and RD N = number of tubes
d, = outside tube diameter p = tube pitch
D0 = perforated area diameter of tubesheet P = equivalent differential pressure in TEMA
D* = £ , *-e 3 /12(l-c* 2 ) = flexural rigidity of P* = equivalent differential pressure in CO-
equivalent plate DAP
e = tubesheet thickness obtained by CODAP Ps, Pt = shellside, tubeside pressure
method Q = N-k,/ks = ratio of tube-bundle/shell rig-
es, e, = shell, tube thickness idity
E* = elastic modulus of equivalent plate q(r) = effective pressure at radius r
E, = elastic modulus of shell, tubes r = radius (Q<r<R)
/ = nominal design stress in CODAP R = outer tubesheet radius
xs = l-AT(G?,/2fl)2 = shellside drilling coeffi- RD = 2R/D0
cient S = allowable stress of ASME VIII-Division
x, = l - A W , - 2 e , ) / 2 / ? ] 2 = tubeside drilling 1
coefficient T = tubesheet thickness in TEMA code
xm= (x, + xs)/2 = mean solidity factor of tubesheet VE = axial reaction at tubesheet periphery
F = TEMA coefficient X, Z - characteristic parameters of exchanger
Hh H2 = coefficients depending on parameters X <xs, a, = thermal expansion coefficient of shell,
and Z tubes
H4 = H2/H1 7 = M f l , - 2 0 o C ) - a ^ - 2 0 ° C ) ] - Z , = axial
ks = [ir-es-(2R + es)-Es]/(L/2) = axial rigidity differential thermal expansion between
of half-shell tubes and shell
k, = [ir-er(dret)-E,]/(L/2)= axial rigidity ACa = coefficient depending on xm and RD
of one half-tube 6S, 6, = mean temperature of shell, tubes
kw = N-k,/irR2 = modulus of equivalent elastic fi = tubesheet ligament efficiency
foundation wR BE = tubesheet rotation at its periphery
K6 = coefficient of elastic restraint at tube- a = maximum stress in tubesheet
sheet periphery v, vs, v, = Poisson ratio of tubesheet, shell, tubes
L = tube length v* = Poisson ratio of equivalent tubesheet

Journal of Pressure Vessel Technology FEBRUARY 1992, Vol. 114 /125

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jpvtas/28333/ on 07/25/2017 Terms of Use: http://www.asme.org/a


Table 1 Basis of tubesheet design methods used in various pressure
vessel codes
CODES Date TYPE OF EXCHANGER headl
U-TUBE FLOATING HEAD FIXED TUBESHEET

TEMA 1941 TEMA TEMA TEMA + GARDNER 52

BS 1515 1965 MILLER 1952 MILLER 1952 MILLER 1952


ISO 1973 GARDNER 1969 GARDNER 1969
STOOMWEZEN 1975 GARDNER 1960 GARDNER 1948 . GARDNER 1952
BS550O 1976 GARDNER 1969 GARDNER 1969 CODAP 1990
ANCC 1975 TEMA 1968 TEMA 1968

CDDAP 1980 GARDNER 1969 GARDNER 1969 GALLETLY 1959


ASME vm-i 1982 GARDNER 1969 SOLER 1984 SOLER 1984
(modified) (in preparation) (in preparation)

Fig. 4 Connection of the tubesheet with shell and head

¥
1 f^T
; /• V \
Fig. 2 Fixed tubesheets heat exchanger /// 1 !
/// • :

e/p<0 25 — [ ,
# s/p-O-5

alREAL MODEL blANALYTICAL MODEL


| // £• i
1 // ' ! : i

LU 7/ i : ; !

V,/ | • i

W *H ~A /
Ol M (JJ
U i 1 ; IP
04 M Oe &7 Ofl M W 0 Ol 02 03 04 05 0-6 07 OB OS 10

ME \ /ME Fig. 5 Values of E*/E and v* for triangular pattern

HT
elastic foundation ol modulus:
N
• The tubesheet is uniformly perforated over its whole area
k >t (unperforated annular rim is not considered).
» The membrane loads in the tubesheet are negligible, as
4P # -[•»•«, -Pt-'t] compared to the bending loads.
8
N lubes ol rigidity : Nk, -NEt.Si
The tubes are uniformly distributed over the whole tube-
L/2 sheet and subjected to a uniform temperature 9,. The bending
Fig. 3 Analytical model used in the design method moments in the tubes at their attachment with the tubesheet
are neglected.
• The exchanger is axisymmetrical and symmetric about
which take into account the characteristics of the tubesheets, the plane midway between the tubesheets so as to treat only
the tube-bundle, the shell, and the channel head. A historical a half-structure as shown in Fig. 3.
summary of the evolution of tubesheet design rules is given in
Table 1.
For convenience of presentation only the main steps of the
analysis are developed hereafter; for more details and complete
French Rules for Fixed Tubesheet Heat Exchangers equations, see reference [17].
This type of heat exchanger has 2 tubesheets attached to the First Step: Modeling of the Tubesheet-Shell-Channel Con-
shell, generally by welding. The channel is attached to the nection. The tubesheet is disconnected from the remainder
tubesheet either by welding or by bolting (Fig. 2). This type of the exchanger by applying at its periphery (Fig. 4):
is probably the most common because it is simple to fabricate,
and consequently less expensive than the 2 others. However, • An axial reaction VE due to the end load acting on the
it can only be used if the temperature difference between tubes head and to the axial displacement As of the half-shell
and shell is not too high. Otherwise, the differential thermal
expansion will cause excessively high stresses in various parts PrirR2-ksAs
of the exchanger. An expansion joint on the shell can solve VF = (1)
2TTR
this problem by absorbing the differential expansion.
where ks is the axial rigidity of the half-shell.
The design analysis of fixed tubesheet heat exchangers is 9
A reactive bending moment ME. The radial displacement
complex because the four basic elements (tubesheet, tubes, at the junction is supposed to be prevented (8r = 0), which
shell, channel) are interconnected to form a statically indeter- enables considering the moment ME as proportional and op-
minate structure and are subjected to different pressures and posite to the tubesheet rotation 6E at its periphery
thermal expansions. The stresses in the four elements of the
structure are determined by using thin shell theory in the elastic MF= -K, (2)
range, and are limited to maximum allowable stress values. where
However, discontinuity stresses in the shell or channel can be
partially or fully plastic. Ke = 2(8s + 8C) (3)
In order to achieve a method suitable for inclusion in a The coefficient Ke depends on the bending rigidities {bs) of
pressure vessel code, the following assumptions are made: the shell and (<5C) of the channel. It represents the degree of

126 / Vol. 114, FEBRUARY 1992 Transactions of the ASME

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jpvtas/28333/ on 07/25/2017 Terms of Use: http://www.asme.org/a


Ps-Xs — Pt'Xt
• The reactive loads of the tubes: - kw-A,(r) where A,(r)

w represents the axial displacements of the tubes.


® The differential expansion between tubes and shell which
is given, for the half-exchanger, by

a-RZ±.
r
p *=[a,-(0,-2OoC)-<v(0,-2O°C)]-| (6)
Fig. 6 Ligament efficiency p.
and acts as an equivalent uniform pressure through the equiv-
alent solid plate.
elastic restraint of the tubesheet by the shell and the head, and
varies between two extreme values: The net effective pressure acting on the equivalent plate is
given by
» 0, which corresponds to the simply supported case;
• oo, which corresponds to the clamped case.
q(r) = [Ps-xs-Prx,]-kw- A.00-2
In TEMA. Only these two extreme cases are consid-
ered: the tubesheet is assumed to be either simply supported
N
or clamped at its periphery through the coefficient Foi TEMA. -v, \ (P P).{d<-e<)2~] + »,12P,-Q\ (7)
Between these two theoretical cases, TEMA proposes for F a R2{P, PS) 2
linear interpolation depending on the ratio es/G (see Fig. 12). Two last terms in Eq. (7) account for the loads resulting from
The problem involved by this simplification is discussed in the the axial displacements of tubes and shell by the Poisson effect
chapter devoted to "comparison with TEMA." of pressures Ps and P,-Q is the ratio of tubes/shell rigidities.
Second Step: Modeling of the Tubesheet. The tubesheet Fifth Step: Analytical Treatment. The problem is now
is replaced by an equivalent solid plate of effective elastic reduced to the case of an equivalent solid circular plate of
constants E* and v*. A study undertaken by the author [18] elastic constants E* and v*, elastically restrained at its pe-
has led to setting up curves for the determination of E* and riphery by a moment ME, resting on an elastic foundation of
v* as a function of the ligament efficiency n for different values modulus kw and subjected to a net effective pressure q(r).
of the ratio e/p (see Fig. 5). From classical thin plate theory, the deflection of such a plate
The flexural rigidity of the equivalent solid plate is given by is governed by a fourth-order differential equation
E*-e3 d4w 2d^w ld2w \dw q(r)
D*=- D (4) (8)
12(1 - p*y1=v dr4 + rdr* i dr1 V dr ~ D*
2

where solution of which is

E-e3 Ps-xs-PrX,- 7
D-- is the flexural rigidity of the tubesheet before w(r) =.4-ber x + 5-bei x + s (9)
12(1 - / )
drilling
[ K A
D*
is the deflection efficiency of the tubesheet K
I ™ A
x=k-r= (10)
-MD*
In CODAP rules the ligament efficiency is based on the min-
imum ligament width when the tubes are welded on the tube- where r = radial distance from the center of the tubesheet, and
sheets (Fig. 6) ber x and bei x are Bessel functions of order 0.
From w(r), one may determine the shear force, the bending
p-d moment, and the slope at any point in the tubesheet. The 2
constants of integration A and B are obtained from the 2
where d is the tube-hole diameter. When tubes are expanded boundary conditions (1) and (2) at the periphery of the
over more than 90 percent of the tubesheet thickness, credit tubesheet (r=R). The solution results in 2 fundamental di-
is taken for half the tube wall as strengthening the ligament, mensionless parameters which characterize the mechanical be-
and the formula becomes havior of the exchanger. The first one
p-(d-et)
/* = X= (11)

Third Step: Modeling of the Tube Bundle. The tubes are represents the relative rigidity of the tube-bundle with respect
assumed uniformly distributed over the whole tubesheet and to that of the tubesheet. It may vary from 0 (no tube-bundle)
in sufficient number so as to act as a uniform elastic foundation to about 50 (very stiff tube-bundle). The second one
of modulus K,
Z =- (12)
N-k,
kw — (5) 4
vrr2 lk„-D*3
where k, is the axial rigidity of a half-tube.
represents the elastic rotational restraint at the periphery of
Fourth Step: Modeling of the Loading. The loads acting the tubesheet. It may vary from 0 (simple supported case) to
on the equivalent plate are due to: infinity (clamped case).
9
The values of these parameters control the maximum stresses
The pressures Ps and P, which must be corrected for the in the various parts of the exchanger: the tubesheets, the
tube-hole areas to calculate the uniform differential pressure tubes, the shell, and the channel head. As X and Z depend on
acting on the equivalent plate the tubesheet thickness, through D*, the method is iterative.

Journal of Pressure Vessel Technology FEBRUARY 1992, Vol. 114 /127

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jpvtas/28333/ on 07/25/2017 Terms of Use: http://www.asme.org/a


100 100 100 001 0-10 100

Fig. 7(a) Values of coefficients H, and H2 as a function of Z for various values of X

\H2(X) z 0-6
o P .=°; 5
0

180
/ ,' / O B
160 Simply supported . -v—X* / / /'

140 / ///
120 / /// / ,2

/ /// / / /3
•' / // / / //*'
8
80
//// / / / // /
^ CO
60
• ' / / / / / / / /

40

; ^ ^ ^ ^
20 ^
*3 ^
X
6 8 10 12 14 16 18 12 14 16 18 20

Fig. 7(b) Values of coefficients H, and H2 as a function of X for various values of Z

Sixth Step: Determination of the Maximum Stresses. The quantities H{ and H4 appearing in Eqs. (13) and (14)
Once the bending moment in the equivalent plate has been are dimensionless coefficients given by curves depending on
obtained from Eq. (9), it is easy to get the bending stress in two basic dimensionless parameters X and Z. These curves are
the tubesheet as a function of r and of the two parameters X plotted in Fig. 1(a) as a function of Z for different values of
and Z (see reference [17]). X. The left part of these curves ( Z < 0.5) represents a situation
A parametric treatment allows the determination of the max- in which the maximum bending stress appears inside the plate;
imum stress in the real tubesheet, which may be written in the whereas the right part (Z>0.5) corresponds to a maximum
form used for circular plates under pressure stress appearing at the periphery of the plate. For the inter-
2 mediate values of Z (Z — 0.5), Fig. 1(a) shows that the stress
1 R
•P* (13) is lower than for Z = 0 (simply supported case) and Z = oo
Ix-H^X, Z) \e (clamped case). For this optimum value Z 0 P T of Z, two max-
in which /x represents the ligament efficiency of the tubesheet, imum stresses appear simultaneously in the tubesheet (one
and P* represents the equivalent pressure acting on the plate. inside and one at the periphery) with the same value.
This is seen more clearly in Fig. 1(b) where H\ and Hi have
been plotted as a function of X, for different values of Z; the
p* = [xs + 2v,{l-xs) + 2vs-Q\ curve relative to Z 0 PT is higher than those relative to Z = 0 and
Q + H, Z=oo. This figure also shows that it is not correct to use
P y interpolation between simply supported and clamped cases,
' [xl + 2ul(l-x,) + Q] + (14)
Q + H< Q + H4 which is a current practice.
CODAP rules permit use of a thicker shell and channel near
where H4 depends on X and Z. the tubesheet so as to adjust the value of Z as close as possible
In Eq. (14), the first term represents the equivalent shellside to Z O PT = 0 . 5 . The maximum stresses in the tubes, the shell,
design pressure (called P / in TEMA), the second the equivalent and the channel can be obtained in a similar way as detailed
tubeside design pressure (called P't in TEMA), and the third in reference [17].
the equivalent thermal expansion design pressure (called P'd in When the tubesheet has an extension for bolting on the shell
TEMA). or channel, this part of the tubesheet is calculated as an annular

128 / Vol. 114, FEBRUARY 1992 Transactions of the AS ME

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jpvtas/28333/ on 07/25/2017 Terms of Use: http://www.asme.org/a


IJIJLJI

Fig. 8 Scheme of a floating head heat exchanger

1.0'
0.9 SOLIDITY FACTOR = 0.6
0.8
0.7 • SIMPLY SUPPORTED, BOTH ENDS
STATIONARY TUBESHEET CLAMPEDj Fig. 10 Various types of connections in CODAP
0.6 FLOATING, SIMPLY SUPPORTED

0.5
Rules for Floating-Head Heat Exchangers
0.4 In this case only, one tubesheet is fixed to the shell, by
welding or bolting, and the second tubesheet is free to move
inside the shell (Fig. 8). This type is more expansive than the
0.3
fixed type, but it has two advantages:
1 there are no stresses due to differential thermal expansion
between tubes and shell;
0.2 2 access into the shell is easier.
The design is much more simple as the second tubesheet is
free. The basic theory is the same as for the fixed tubesheets,
with the further assumption that the axial rigidity of the shell
is zero, which leads to P*=PS-P, in Eq. (13).
0.1
2 3 4 5
Tubesheet Design. The rule used in CODAP is based on
Fig. 9 Values of coefficient C„
the method proposed by Gardner in 1969 [10]. The tubesheet
thickness for bending is given by the direct formula

e=(C0 + AC0)-Dt (15)


plate clamped on the shell or channel and subjected to the 2nf
uniform differential pressure Ps - Pt and to the moment exerted where D0 is the diameter of the perforated area of the tubesheet.
by the adjoining flange. The coefficient Ca, given in Fig. 9, depends on 2 coefficients:
1 X0 which is similar to the coefficient X relative to the fixed
Practical Design Procedure. The design procedure is as tubesheets, with modifications so as to present a direct formula
follows: for calculating e
1 Select a value for tubesheet thickness e. 1.5(l-V*2)(x,-xs)^-^
2 Determine values of E* and v* from Fig. 5.
3 Calculate X and Z by Eqs. (11) and (12). Xn (16)
4 Determine values of Hu H2 from Fig. 7(a). \P.-P,
5 Calculate P* by Eq. (14).
6 Calculate the bending stress a in the tubesheet by Eq. (13). 2fi-f
The stresses in tubes, shell, and channel, which are not given 2 RD = 2R/D0, which accounts for an untubed rim at the
in this paper, are calculated in the same way. periphery of the tubesheet.

All these calculated stresses are limited to the maximum Values of coefficient C0 are given (Fig. 9) for a mean solidity
allowable stresses fixed by CODAP using the concept of pri- factor xm of the tubesheet equal to 0.6. The coefficient AC0
mary and secondary stresses: gives corrections for other values of the solidity factor. As
shown in Fig. 9, C0 values are only given for the two theoretical
8
bending stresses (in tubesheet, shell, channel) are limited cases of simply supported and clamped tubesheets; no credit
to is given to the rotational restraint on the tubesheet from the
1.5f for pressure loads (2f for the tubesheet, when shell and channel. Therefore, the designer must judge whether
integral with shell and channel), the tubesheet is simply supported, clamped, or halfway be-
• 3f for thermal expansion loads + pressure loads tween these two extreme cases. Such a design approach is not
9 acceptable in a pressure vessel code because 2 different de-
membrane stresses (in tubes, shell) are limited to
• f for pressure loads signers could obtain two different tubesheet thicknesses for
1.5f for thermal expansion loads + pressure loads. the same heat exchanger.

Journal of Pressure Vessel Technology FEBRUARY 1992, Vol. 114 /129

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jpvtas/28333/ on 07/25/2017 Terms of Use: http://www.asme.org/a


IJXJILJI

11
Fig. 11 Scheme of a U-tube heat exchanger
002 OJOS

Fig. 12 Coefficient F of TEMA


To provide consistent designs, CODAP decided to consider
the tubesheet as
9
PCODAP ~y2/*H, fJ- 0.3
simply supported (SS) when bolted with shell and channel
(sketch 1 in Fig. 10)
0.8 OR 1
/~i SS TEMA"
o ^~*o
8 simply supported
halfway between simply supported (SS) and clamped (CL)
in all other cases, i.e., >TEMA
•v £lamped
cj.am
• integral with shell and bolted with channel (sketch 2)
• bolted with shell and integral with channel (sketch 3)
• integral with shell and channel (sketch 4)

In these cases, C0 is taken in the rules as •z=o (


CODAP

The same treatment is applied to the coefficient AC 0 , which


also depends on the type of tubesheet-shell-channel connection. Fig. 13 Comparison of coefficient F i n CODAP and TEMA codes for
simply supported and clamped case
In CODAP, the coefficient C0 depends on the tubesheet
thickness through the coefficient Xa (Eq. (16)), which is a
function of E* and consequently of e (see Fig. 5). So the basic
ter is devoted to U-tube heat exchangers. The bending tubesheet
formula, Eq. (15), is written in the French rules as
thickness is given by the direct formula
(C 0 + AC0)> (D0\\{Ps_Pt) ( n )

e=(C0 + AC0)-D0 (20)


and a is limited to If. The method is iterative, but the direct 2M/
Eq. (15) is very helpful for finding a first evaluation of e by where C0 + AC0 is given directly by a table as a function of
using the lowest curve in Fig. 5, giving results on the safe side. the solidity factor and ratio 2R/D0.
Tube Design. The calculation of the maximum axial stresses
in tubes is quite similar to the one used in fixed tubesheets. Comparison With TEMA
Two different formulas are given for interior and outermost Comparison for Fixed and Floating-Head Heat Exchangers.
tubes TEMA rules for fixed tubesheets were elaborated by Gardner
[4] for the fifth edition of 1968, which remained unchanged
(Ps Xs- -Prx,)+FH(P, -P,)
Hi (18) in the sixth edition of 1978 [1], analyzed in reference [19]. In
xt — xs these editions, the basic bending formula for the tubesheet
(Ps -Ptx,)-F,e(Ps-
•Xs- -Pt) thickness is
°t,t (19)
x,-
The tube coefficients Fti and F,e are analogous to coefficient T=F- (21)
C0 and are given by curves, which depends on parameters X0
and RD. where F i s given by Fig. 12. (F= 1.0 for a simply supported
An inconvenience of the method is that the value of the tubesheet, F = 0 . 8 for a clamped tubesheet.) G is the shell
coefficient X0 implicitly assumes that the thickness of the diameter. P=P* for fixed tubesheets (Eq. (14)) and PSP, for
tubesheet is the one calculated from Eq. (15). If the real thick- floating tubesheets. S= allowable stress taken from ASME
ness is higher, the stresses calculated in the tubes by Eqs. (18) VIII-Division 1.
and (19) will not be correct, but generally the results are on From the basic analysis presented for fixed and floating-
the safe side. head heat exchangers, Eq. (13) shows that the bending formula
can be written
Rules for U-Tube Heat Exchangers
1
This type, which has only one tubesheet fixed to the shell
and channel by welding or bolting (Fig. 11), is simpler than the
two other types of heat exchangers, but may be more costly
r= \/2/xZ/i ( j - ^ C O D A P r.
II (22)

due to the forming of the tubes. The design method is also where Hx depends on the characteristic coefficients X and Z
simpler as the tubes do not act as an elastic foundation. So of the heat exchanger. In TEMA, Z = 0 (simply supported case)
the floating-head heat exchanger method can be used by mak- or 00 (clamped case).
ing Xo = 0 in the foregoing equations. This is done in most Comparison of tubesheet Eqs. (21) and (22) shows that the
tubesheet codes (ISO, BS 5500), but in CODAP a special chap- coefficient F of TEMA has been empirically chosen by assum-

130 / Vol. 114, FEBRUARY 1992 Transactions of the ASME

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jpvtas/28333/ on 07/25/2017 Terms of Use: http://www.asme.org/a


Table 2 Comparison of CODAP (e) and TEMA (7) thicknesses for U- ceptable for classical exchangers of medium size, subjected to
tube exchangers normal pressures, for which these rules were originally estab-
lished in the sixties and seventies.
e/T VALUES Today chemical and petroleum industries need larger ex-
changers operating under increasingly severe conditions. Due
SKETCH 1 (Fig. 10) SKETCHES2,3,4(Fig.l0) to their simplicity, TEMA standards do not ensure the same
H = 0.22 H = 0.335 H = 0.22 u = 0.335
level of safety for all these modern exchangers.
RD
This is the main reason why the French Pressure Vessel Code
Committee has elaborated from 1975 to 1980 a more scientific
1.0 1.35 1.10 1.33 1.08 design method for fixed tubesheet heat exchangers accounting
1.1 1.28 1.07 1.24 1.01 for the axial rigidity of the tubes, the presence of the holes,
1.3 1.15 0.99 1.07 0.90 and the type of connection between tubesheet, shell, and head.
Certain other countries (Australia, China, India, UK, USA,
etc.) are developing a similar approach, but as far as the author
knows, CODAP was the first to publish such rules in 1980.
ing that the stiffening effect of the tubes (accounted for by
coefficient X) is balanced by the weakening effect of the holes For floating-head and U-tube exchangers, CODAP, as do
(accounted for by coefficient /x). some other pressure vessel codes, proposes rules based on the
work of Gardner, with some modifications. These rules take
If the first stiffening effect prevails, which happens when into account the untubed rim at the periphery of the tubesheet,
Xis high (X> 7), the TEMA formula will be overconservative. but the treatment of the tubesheet-shell-channel connection
In the inverse case, which happens when Xis low (X<3), the needs further refinements.
TEMA thickness will be underconservative. This is illustrated
in Fig. 13. These two types of methods, for fixed tubesheets and float-
ing-head and U-tube tubesheets, are already in practice in some
The seventh edition of TEMA [5], published in 1988, pro- countries, but not always officially. France will propose them
vides a new bending formula for the tubesheet, which includes as a basis for the future European Unfired Pressure Vessel
a ligament efficiency i) Code, whose implementation was decided in November 1989
by the European Community.
(23)
References
with a factor 3 instead of 2 below G. This ligament efficiency 1 TEMA, "Standards of Tubular Exchanger Manufacturer's Association,"
is based on the mean ligament efficiency in the tubesheet, Sixth Edition, 1978.
instead of the minimum ligament efficiency /x in CODAP 2 CETIM, Centre Technique des Industries Mecaniques, Technical Center
for Mechanical Industry.
3 SNCT, Syndicat National de la Chaudronnerie-Tolerie, National Asso-

" = 1-4in7© ciation for Pressure Vessels.


4 Gardner, K. A., "Heat Exchanger Tubesheet Design-2: Fixed Tube-
sheets," Trans. ASME, Vol. 74, 1952, pp. 159-166.
where 5 TEMA, "Standards of Tubular Exchanger Manufacturer's Association,"
Seventh Edition, 1988.
6 Gardner, K. A., "Heat Exchanger Tubesheet Design," Trans. ASME,
0 = — for triangular pitch Vol. 70, 1948, pp. 377-385.
7 Miller, K. A. G., "Design of Tube Plates in Heat Exchangers," Proceed-
ings of the Institute of Mechanical Engineers, Vol. 113, 1952, pp. 215-231.
8 = — for square pitch 8 BS 1515—Part 1,' 'British Standard Specification for Fusion Welded Pres-
sure Vessels," 1965.
9 STOOMWENZEN (Dutch Code), "Rules for Pressure Vessels," 1973.
Values of t\ generally range between 0.4 and 0.6, whereas it 10 Gardner, K. A., "Tubesheet Design: A Basis for Standardization," First
values are usually between 0.2 and 0.3. Finally, this new for- International Conference on Pressure Vessel Technology, Part 1, Design and
mula leads to a thickness decrease on the order of 5 to 15 Analysis, ASME, 1969, pp. 621-648.
percent for a square pattern, 0 to 10 percent for a triangular 11 ISO/DIS2694, "Pressure Vessels," 1973.
pattern. 12 BS 5500, "British Standard Specification for Fusion Welded Pressure
Vessels," 1988.
13 CODAP: French Code for the Construction of Pressure Vessels, AFIAP
Camparison for U-Tube Heat Exchangers. In this type of and SNCT, 1985.
exchanger, the tubes do not provide any staying action to the 14 ASME Boiler and Pressure Vessel Code, Section VIII, Division 1, Ap-
tubesheet. TEMA takes this into consideration by increasing pendix AA, 1989.
the F value by 25 percent: F= 1.25 when the tubesheet is 15 Galletly, G. D., "Optimum Design of Thin Circular Plates on an Elastic
simply supported, F= 1.00 when the tubesheet is clamped. Foundation," Proceedings of the Institute of the Mechanical Engineers, Vol.
173, 1959, pp. 687-698.
Table 2 shows a comparison between the thicknesses ob- 16 Singh, K. P., and Soler, A., Mechanical Design of Heat Exchangers, 1st
tained by CODAP and TEMA (6th Edition) for 3 different Edition, Arcturus Publishers, 1984.
values of RD and two values of /x, for the sketches shown in 17 Osweiller, F., "Methode de calcul des echangers a deux tStes fixes pour
Fig. 10. le CODAP," Etude CETIM, No. 14B031, 1986.
18 Osweiller, F., "Evolution and Synthesis of the Effective Elastic Constants
Concept for the Design of Tubesheets," ASME JOURNAL OF PRESSURE VESSEL
Conclusions TECHNOLOGY, Aug. 1989, Vol. I l l , pp. 209-217.
19 Osweiller, F., "Analysis of TEMA Tubesheet Design Rules," Proceedings
The TEMA rules for determining tubesheet thickness are of the 1986 Pressure Vessel and Piping Conference, Chicago, Illinois, July 1986.
based on a semi-empirical method which assumes that the 20 ANNC Italian Code, VSR Collection, Chap. VSR-l-N, 1975.
stiffening effect supplied by the tubes is balanced by the weak- 21 Soler, A., Caldwell, S., Singh, K. P., "Tubesheet Analysis—A Proposed
ASME Design Procedure," Karl Gardner Memorial Session, ASME PVP-Vol.
ening effect of the holes. This simplistic approach was ac- 118, H00343, D e c , 1986.

Journal of Pressure Vessel Technology FEBRUARY 1992, Vol. 114 /131

Downloaded From: http://pressurevesseltech.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jpvtas/28333/ on 07/25/2017 Terms of Use: http://www.asme.org/a

You might also like