T.

Sawa
Associate Professor, Department of Mechanical Engineering.

A Stress Analysis of Pipe Flange Connections

The use of pipe flange connections is standardized in the codes of JIS, ASME, DIN and so on. However, these codes are almost entirely dependent on experience, and subsequently some problems concerning pipe flange connections have been encountered. In the present paper, the distribution of contact stresses which governs the sealing performance is analyzed as a three-body contact problem, using an axisymmetrical three-dimensional theory of elasticity. The effects of the stiffness and the thickness of raised face metallic gaskets on the contact stresses and the effective gasket seating width are shown by numerical calculation. Moreover, stresses produced on the hub, the load factor (the relationship between an increment of bolt axial force and an internal pressure), and the maximum stress caused in bolts are analyzed. For verification, experiments are carried out. The analytical results are satisfactorily consistent with the experimental results.

N. Higurashi
Graduate Student. Yamanashi University, Kofu, Japan

H. Akagawa
Engineer, Taikisha Co., Ltd., Tokyo, Japan

Introduction A pipe flange connection is an important part of a pipe system. The use of it is standardized in the codes of JIS (1976), ASME (1981), DIN (1964) and so on. However, these codes are almost entirely dependent on experience, and consequently some problems concerning pipe flange connections have been encountered. Up to now, many investigations (Nishioka et al., 1979a, 1979b; Kohmura, 1985, 1986; Sawa et al., 1986; Sawa and Kumano, 1985) have been carried out on the pipe flange connections; but it seems that only a few investigations, including the studies done by Nishioka (1979a, 1979b) and ourselves (Sawa et al., 1986), have been done on pipe flange connections, taking into account the entire structure which consists of hubs, flanges, gaskets and bolts, because of the difficulties encountered in attempting an analysis. For this reason, several characteristics of pipe flange connections are yet to be clarified. In order to make the design of pipe flange connections optimal, it is necessary to treat the connections as an entire structure consisting of hubs, flanges, gaskets and bolts and to examine (/) the contact stresses which govern the sealing performance, (ii) the load factor (the relationship between an increment of bolt axial force and an internal pressure), (Hi) the moment arm, and (iv) the effective gasket seating width and so on. The purpose of this paper is to establish an optimal design of pipe flange connections by treating a pipe flange connection as an entire structure. In the analysis, a hub of pipes, flanges and a raised face metallic gasket are replaced with finite hollow cylinders, and the contact stresses on the gasket, the stresses produced on the hub and the load factor are analyzed, respectively, as an elastic contact problem using an axisymContributed by the Pressure Vessels and Piping Division and presented at the Pressure Vessels and Piping Conference, Honolulu, Hawaii, July 23-27, 1989,
of THE AMERICAN SOCIETY OF MECHANICAL ENGINEERS (JSME co-sponsorship).

metrical three-dimensional theory of elasticity. The stress produced on the hub and the load factor are measured by strain gages when an internal pressure is applied to the pipe flange connections. The analytical results are compared with the experimental results. In addition, the effects of the stiffness and the thickness of raised face metallic gaskets on the contact stresses, the effective gasket seating width and the moment are discussed. Theoretical Analysis 2.1 Analysis of Contact Stresses on the Gaskets. Figure 1 shows a pipe flange connection subjected to an internal pressure, in which a raised face metallic flat gasket is interposed between a couple of pipe flanges and is fastened with an initial clamping force Ff by A^ sets of nuts and bolts. In order to analyze the distribution of contact stresses on the connection in the initial clamping, a model for analysis shown in Fig. 2(a) is adopted; namely, the hubs of pipes, the flanges and the gasket are respectively replaced with the hollow cylinders [I], [II], and [III]. The inner diameter, the outer diameter, the thickness, Young's modulus and Poisson's ratio of the hollow cylinders [I], [II], and [III] are designated as 2au 2b\, 2hx, Eu vu 2o2, 2b2, 2h2, E2, i>2, 2a3, 263, 2hit E^ and p3, respectively. The width of the bearing surface is designated as 2c and the pitch circle diameter as 2d. It is assumed that the initial clamping force Ff x N acts on the upper surface of the hollow cylinders [II] (d - c < r < d + c) as a uniform pressure Pi (Pi = F/ x N/A-wdc) annularly, and that the mean diameter between the maximum and the minimum diameter of the hub is used as the outer diameter of the hub of pipes. Expanding the uniform pressure Pi due to the initial clamping force into Bessel functions Co(ysllr), the boundary conditions are expressed as follows: (/) u is the deformation in the r direction, (//) w is the deformation in the z direction, and (Hi) the superscripts I, II, and III correspond to the hollow cylinders
NOVEMBER 1991, Vol. 113 / 497

Manuscript received by the PVP Division, May 1, 1989; revised manuscript received May 1, 1991.

Journal of Pressure Vessel Technology

Downloaded 19 Oct 2008 to 129.5.16.227. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Copyright 1991 by ASME

Z,,Z2,Z3

0; xL-i

-^rm

-nii]

w*
icir
pe flange ;Gasket Pipe flange cfe2o1/<p 2a, 4>2a3 *2b! < 2d
qb2b 3

^ ^

--nra]
->-r[n]

TT3
->-riu

3>2b2
(a) Fig. 1 A pipe flange connection with a raised face metallic gasket subjected to internal pressure The case of initial clamping

[I], [II], and [III]. The shearing forces on the contact surfaces between the hollow cylinders [II] and [III] are neglected.
ll,l2,ll

(/)

On the hollow cylinders [I] (hubs) r = ax: arl = rrzl r=bx : a/ = T'
Zi = hi : az = rrz
= 0 = 0

u.
(1) Bearing surface

P2t

= o]

0, _ . p ,
02

U) On the hollow cylinders [II] (flanges) r = a2 a f H = T n = 0 r=b2 : trra = rna = 0 22 = ^2 : <r^ =

Z2~-

~r[n]

a0+Y^a/:0(ysllr) =o

(2)

czzz.

-r[m] -r[n]

-hi-.r"

(Hi)

On the hollow cylinder [III] (gasket) r = a3:arm r=bi:or

m

= rrzm = 0> = rrzm = 0\ (3)

z 3 = A 3 :Trzm = Q (iv) On the contact surface between [I] and [II]

(<fz\=-hi = (Oz\ = h2 (i < / < * ! ) (l</-<fol) (Trz)z\~-hx=(Trzl)z2

>2al^2a?'
*2a3 <P2bi

cP2d

= h1

("')*
dw

(4) (ax<r<bx)
zi = -hl

*2b3 *2b2
(/j) Fig. 2 The case where internal pressure is applied

a7

z1=n1

Model for analysis and dimensions of pipe flange connection

498 / Vol. 113, NOVEMBER 1991

Transactions of the ASME

Downloaded 19 Oct 2008 to 129.5.16.227. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

r-

M60x3

Pressure

cP130 cP165
(a)

Pipe flange

Fig. 4

Setup 01 experimental apparatus

I !
et:>50
et:> 105
(b)

.c

0-,

2
Gasket

2.2 Analysis of Load Factor. As shown in Fig. 1, when an internal pressure P is applied to pipe flange connections fastened with an internal clamping force Fj , an increment F I of bolt axial force is produced in each bolt. The relationship between the increment F I of bolt axial force and the internal pressure P, (the load factor per bolt) is given by Eq. (6) (Sawa et al., 1986; Sawa and Kumano, 1985), using the axial force W( W = 7ra/ 2p) due to the internal pressure P.

<I>_l_~.(Keg')
WIN kl+keg Kpi

(6)

=:r

Strain gauge
M16
,-+ f}
to

In the case where a separation of contact surface increases with an application of the internal pressure P, the load factor is estimated by analyzing the values of Keg' and Kpl '
cb3

f}

10

0.6 15

20
90

38

2.3 Analysis of Bending Moment Produced in Bolts. With an application of the internal pressure P, a bending moment M b is produced, as well as an increment F I of bolt axial force. Here, the method to obtain the bending moment M b is given in the references (Sawa et al., 1986; Sawa and Kumano, 1985). The maximum increment elmax is expressed as Eq. (7), taking account of the bending moment M b
el

max

<I> W M b =-.-+A N Z
b

(7)

Fig. 3

Boll Dimension 01 pipe flange, gasket and bolt used in experiments

(e)

where A b is the cross-sectional area of bolt axis and Z is the section modulus.

3 Experimental Method
Figure 3 shows dimensions of pipe flanges, gaskets and bolts used in the experiments in order to obtain the load factor and the stress produced on the hubs. The pipe flanges shown in Fig. 3(a) are manufactured in conformity to the basic dimensions of JIS B 2216-40 kg/cm2 steel pipe flanges and they are made of steel for structural use (S45C, JIS) and the nominal diameter is 50 mm. Figure 3(b) shows dimensions of gaskets and their thickness 2h 3 is taken as 3 mm and 5 mm. The materials used for gaskets are mild steel (SS41, JIS) and aluminum (AI-H, JIS). Figure 3(c) shows dimensions of bolts used in the experiment. The size of bolts used is M16 and they are made of chromium molybdenum steel (SCM435, JIS). Two strain gages (Kyowa Electronic Instrument Co. Ltd., KFC-2C l -ll) are attached to the shank of bolts in 180-deg phase. After a couple of the pipe flanges with the gasket are fastened with eight sets of nuts and bolts with the initial clamping force Fj = 15 kN, an internal pressure is applied to the connection using an oil pressure pump (Riken Pressure Machine Co. Ltd., MP5C) as shown in Fig. 4. Then, the magnitude of the internal pressure is measured with a pressure transducer (Kyowa Electronic Co. Ltd., PO-500KU), and the axial force and the maxNOVEMBER 1991, Vol. 113/499

where
ao=
d/-d l
2

2 -

b 2 PI
1
Ph

2(d2 2 C l ('YslldZ) -dI 2 C I bslId l )] as = - l1(b 2e 2( lib) 2 2 11 'Ys 2 0 'Ys 2 -a2 CO ('Ys a2)]

d l =d-c, d 2 =d+c

The distributions of contact stresses in the case where the internal pressure is applied to the connection are estimated by superposing the case of Fig. 2(a) on the case of Fig. 2(b). It is assumed that the axial force W' (W' = 7ra12p) acts on the upper surface of the hubs as a uniform pressure P 2 , as shown in Fig. 2(b). Journal of Pressure Vessel Technology

Downloaded 19 Oct 2008 to 129.5.16.227. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Strain gouge

Fig. 5 Attaching point of strain gages for measuring stress produced on the hub

1.0 imum stress of bolts are picked up by the strain gages attached to the shank of bolts. The outputs are recorded by an X-Y recorder (Yokogawa Electric Works Co. Ltd., 3036) through dynamic amplifiers. The stress produced on hubs is measured with strain gages (Kyowa Electronic Instrument Co. Ltd., KFC-03-Crll) glued to the hub of pipes as shown in Fig. 5. They are attached on the line A-A and the middle of bolt pitch B-B. 4 Analytical Results and Comparisons with Experimental Results 4.1 Analytical Results on Distribution of Contact Stresses in Initial Clamping. Numerical calculations are done using the dimensions of the hub of pipes, the pipe flanges and the gaskets used in the experiments. Poisson's ratios pu v2, and v3 are all put as 0.3. Young's modulus EY and E2 are put as 206GPa and E3 is put as 206GPa for mild steel, 68.7GPa for aluminum. The diameter of the bearing surface is put as 2c = 24 mm, the pitch circle diameter as 2d = 130 mm, and others as 2a\ = 2a2 = 50 mm, 2bi = 79 mm, 2b2 = 165 mm, 2b3 = 105 mm, lhx = 20 mm, 2h2 = 26 mm, 2/i3 = 3, 5, 10 mm. Figures 6(a) and 7(a) show the effects of the ratio Ex/E3 of Young's modulus and the thickness of gaskets on the distribution of contact stresses in the initial clamping. The ordinate is the ratio of the contact stress uz to mean contact stress azm, and the abscissa is the ratio of the distance r from the center to the inner radius a3 of gaskets. Figures 6(a) and 7(a) show that the distributions of contact stresses tend to be uniform with an increase of Ex/E3 and of the thickness of gaskets. But it is seen that the separation of contact surface already begins at the inner circumference in the initial clamping under the boundary conditions (1)~(5). Then, in order to estimate the effective gasket seating width and the moment arm which are standardized in the codes of JIS (1976) and so on, the distributions of contact stresses are obtained through trial and error, and are shown in Fig. 6(b) and Fig. 1(b). The stresses in the initial clamping produced on the width (b3 a 3 ') of oblique lines are shown in Fig. 8, and the value (b3 - a 3 ') is defined as the effective gasket seating width b. Then, after determining a point on which the gasket force P3 acts, the distance between the bolt axes and the point on which gasket force acts is defined as the moment arm A3. The values of b and A3 are shown in Table 1. It is shown that the effective gasket seating width b and the moment arm A3
500 / Vol. 113, NOVEMBER 1991

1.2
(a)

1.4

1.6

1.8

2,02,1
r/03

The case of a, = a3

= 5,0
\=>
N

4,0

3,0

2,0

1.0

0,0 -1,0 1.0

(b)

(Without
_l I I

hubs)'

1,2

1,4

1,6

1.8

2,02,1
r/03

The case where a3 is changed

Fig. 6 Effects of ratio of Young's modulus on distribution of contact stresses

tend to become larger with an increase of the ratio Ex/E3 and the thickness of gaskets. From the codes of JIS (1976), ASME (1981) and so on, the effective gasket seating width b and the moment arm A3 are shown to remain constant, regardless of the material and the thickness of gaskets. It is seen that the difference between the results obtained from JIS (1976), ASME (1981), etc., and the results obtained by this study concerning M3(M3 = P3 x A}) being produced on the pipe flange connections, is considerably sizable. 4.2 Comparison on the Load Factor. Figure 9 shows comparisons of the numerical results with the experimental results on the load factor. The abscissa is the internal pressure P and the ordinate is the bolt axial force Ff + F,. The gaskets
Transactions of the ASME

Downloaded 19 Oct 2008 to 129.5.16.227. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Pipe flange

5.0 I

L.

1.0

1,2
(a)

1,4

1.6

1.8

2,02,
r/03 Fig. 8 Forces act on a pipe flange connection

The case of a, = a3

-' 18
IX.

r-

'v~-:

" E X P - - - ' ^ -~>---.-:-:Nlin

"['"--r r __!_ rrrrr j rL , 1-

t 17
<u 1 6 15
X

_ . . _ _ ; . - .

_^

-.. r_.._!_...
; ;
r

~:s^~ ..- : -:; ; .

-... i - ! \ - ! - -- r

\-c- 1 SkN rf I3N1

, i j-

_z=r_^y=i- ;
^ -

_-*^=.^~_.
==r

^^L~;-r-r.r--i-j--;r---

- _^;~;"jTzr

14

-^rzJ7:^

a
._..._

^^-i^
j ._ ..

?^-^^

j .

^ ^=

o
cq

" " i i

:
- " " i~

_"!_: T ~ : T Z _ : ~

r- - - -

...., 20

0 0,0 -1.0 1,0

(b) Fig. 7 (Without
I I

10

Pressure P MPa
hubs/
I l _

1.2

1,4

1,6

1.8

2,02,1
r/03

Fig. 9 Comparison of numerical results with experimental results with respect to load factor (case of 2b, = 20, 2h2 = 26, 2h 3 = 5, 2a, = 50, 2a2 = 50, 2a3 = 50, 2b, = 79, 2b2 = 165, 2b3 = 105 mm, 3 = 68.7 GPa)

The case where a3 is changed Table 1 Effective gasket seating width b and moment arm A3 in the case of initial clamping

Effects of thickness of gaskets on distribution of contact stresses

are made of aluminum and the thickness of 2h3 = 5 mm. In this connection, a separation of contact surface has already occurred in the initial clamping (7y = 15 kN). When the internal pressure P is applied to the connection, the separation of contact surface grows and the increment of bolt axial force yields a curve. With a further increase of the internal pressure P, a leakage of the internal fluid, from the inner circumference of gaskets occurs, and the internal pressure does not increase. It is seen that the numerical results are consistent with the experimental results. Here, it is assumed that the leakage is caused in the analysis when the point of 98 percent of the basic gasket seating width begins to separate. Figure 10 shows the comparison between the analytical result and the experimental result concerning the maximum stress in
Journal of Pressure Vessel Technology

(mm)
ri3/ai E1/E3 N b b (Num)
(JIS)

0,06 3 1

0.10 3 27.5 10

0,20 3

8,8

7.2

10,2 9,3

16,5

13,5

A3 (Num) A3 Uis)

16,0

15,0

16,6 .21.8

17.9

18,0

NOVEMBER 1991, Vol. 113 / 501

Downloaded 19 Oct 2008 to 129.5.16.227. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Table 2 Effective gasket seating width b and moment arm A3 in the case where internal pressure is applied (dimensions are same as Fig. 9)

(mm)

P (MPa) b (Num)
b
(JIS)

1,9 8,2

5,9 6.2 9,3

8,7 4,2

12.3
2,2

10,2

A3 (Num) ' 16,6

A3

15,0

14.6 21,8

13,8

13,3

(JIS)

Pressure

P MPa

Table 3 Stress produced on the hub of pipes in axial direction in the case of initial clamping (case of 2a, = SO, 2>, = 90, 2/), = 36, others are same as Fig. 9)

Fig, 10 Comparison in the case where bending moment is taken into consideration (dimensions are same as Fig. 9)

2ri3 (mm) Material

3 Al
32.3 40,2 65,5 29,5 37',4

5
SS41 33,3

Nm u
rj/2 (MPa)
JIS

Exp

31,9

Pressure
Fig. 11 Stress produced on the hub of pipes in axial direction in the case where internal pressure is applied (dimensions are same as Table 3)

bolts, taking into account a bending moment produced in bolts. The analytical result and experimental result are satisfactorily consistent. Also, the increment F, of bolt axial force obtained from the load factor is shown by an alternate long and short dash line. The maximum stress is larger than the stress obtained from the load factor. Therefore, when designing bolts, the maximum stress must be taken into consideration. The effective gasket seating width and the moment arm in the case of initial clamping are presented in Section 4.1. However, it is supposed that the effective gasket seating width b and the moment arm A3 will be changed when the internal pressure is applied to the connections. When the internal pressure P is applied to connections, the effective gasket seating width b and the moment arm ^43 is obtained by the same way as Section 4.1, by superposing the stress distribution in initial clamping and the stress distribution when the internal pressure is applied. Table 2 shows the analytical results concerning b and A3. It is seen that the difference between the numerical results and the results obtained from JIS (1976), is substantial. 4.3 Comparison About a Stress Produced on the Hub of Pipes. Table 3 shows a relationship among the stress produced on the hub of pipes, the thickness and the material of gaskets in initial clamping (Ff = 15 kN). From the table, the
502 / Vol. 113, NOVEMBER 1991

composition of gaskets has considerable effect on the stress produced on the hub. The stress produced on aluminum is larger than on mild steel. Also, it is seen that the stress produced on the hub becomes larger with an increase of the thickness. In contrast with this, the value from JIS (1976) code is held constant independent of the material and thickness of the gasket, and the value from JIS (1976) code deviates from the experimental and the numerical results. Here, the value of the stress obtained by the experiment is measured at A-A line (Fig. 5). Referring to JIS (1976) and so on, numerical calculations are done using the dimension of the hub of pipes as 2a} = 50 mm, 2bi = 90 mm, 2h, = 36 mm. Figure 11 shows the stress produced on the hub when the internal pressure P is applied to the connection. The ordinate is the stress produced on the hub az and the abscissa is the internal pressure P. It is seen that the numerical results, taking into account the hub of the pipes, is satisfactorily consistent with the experimental results. On the other hand, it is shown that the result obtained from the formula in JIS (1976) code increases linearly and deviates substantially from the experimental result. Conclusions This paper dealt with the characteristics of pipe flange connections with raised face metallic flat gaskets. The distribution of contact stresses, the load factor, the stress produced on the hub of pipes, the effective gasket seating width and the moment arm were all examined. The following results were obtained: 1 In order to analyze the distribution of contact stresses, a method of analysis was demonstrated using the three-dimensional theory of elasticity by replacing hubs of pipes, flanges and a gasket with finite hollow cylinders. 2 The load factor was analyzed and the analytical results were compared with the experimental results. They were satisfactorily consistent. 3 Concerning the stress produced on the hub, the analytical results were satisfactorily consistent with the experimental results. The results obtained from JIS (1976) code deviated substantially from the experimental results. 4 The effective gasket seating width and the moment arm were analyzed by using the distributions of contact stresses
Transactions of the ASME

Downloaded 19 Oct 2008 to 129.5.16.227. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

mentioned in 1 and the analytical results deviated from the results obtained from JIS (1976) code. References
ASME, 1981, Boiler and Pressure Vessel Code, Sec. VIII, Appendix II. DIN, 1964, 2505 Berechung von Flanschverbindungen. JIS, 1976, B2210-2217. Kohmura, S., 1985, "The Design of Aluminium Bolted Flange Connections (Part 1: Relationship between Leak Properties and Gasket Load in Integral Hub Flanges)," Bulletin of JSME, Vol. 28, No. 243, pp. 2181-2187. Kohmura, S., 1986, "The Design of Aluminium Bolted Flange Connections (Part 2: Relationship between Stresses and Bending Flexibility in Integral Hub Flanges)," Bulletin of JSME, Vol. 29, No. 249, pp. 1026-1032. Nishioka, K., etal., 1979a, "Strength of Integral Pipe Flanges (No. 1: Stress Distribution in Flange and the Criticism of the Conventional Standards)," Bulletin of the JSME, Vol. 22, No. 174, pp. 1705-1711. Nishioka, K., et al., 1979b, "Strength of Integral Pipe Flanges (No. 2: Gasket Seating Stress and the Influence of Number of Bolts)," Bulletin of the JSME, Vol. 22, No. 174, pp. 1712-1718. Sawa, T., et al., 1986, "On the Characteristics of Bolted Joints with Gaskets (Stress Analysis of Metallic Gaskets with Raised Face Interposed between Pipe Flanges)," Bulletin of the JSME, Vol. 29, No. 248, pp. 385-392. Sawa, T., and Kumano, H., 1985, "On the Characteristics of Bolted Joints with Gaskets (Stress Analysis of Full Face Metallic Flat Gaskets Interposed Between Pipe Flange)," Bulletin of the JSME, Vol. 28, No. 237, pp. 400-407. Sawa, T., and Maruyama, K., 1976, "On the Deformation of the Bolt Head and Nut in a Bolted Joint," Bulletin of the JSME, Vol. 19, No. 128, pp. 203211. Sawa, T., and Shiraishi, H., 1983, "Simple Method of Calculate the Force Ratio of Bolted Joints (The Case Where Clamped Parts are Circular Flanges and Pipe Flanges)," Bulletin of the JSME, Vol. 26, No. 216, pp. 1088-1095.

= Aj^

+ BjZi In r + C 0 ' ^ + f ] ^ 7 5

[Aj{A3'%r

+ A 3 7 V/-/ lr + A^Kor + AjX'rK^) + Bj{AjlI0r + A42ft/r/lf + A4nK0r + Aj4j3jrKir}]sm^JZi)

+

1=1 " * T s

2 o ^ f~

{2vsh(yshi)+ysh1ch(ysh1)}sh(yszi)
(10)
n 73

+ 7iz1sh(7i/i1 )ch(7iZi )]C0 (ysr) ^J^^JJBj^sufj' ,ys, A', A3'\ A3 , A3 , A3/4, A/ 1 , A/ 2 , A/ 3 , A/ 4 , 0/, hlt r, Zl) _.-iiW)3 , + A3nKo/+A3'%!'rKi/} + BJ{Aj[I0r' +A/%''rIlr' + A 4 7 V^ 1 /)]cos(ft/'z 1 ) + YJ =T~I [ ~ f 2 "i ch (7^i)
J=I sys

= " S - , l / , &J{Aj%/ +A/2/3/V/1/

AJ'K0/

+y^Hyshi))ch(y1zi)

APPENDIX In order to analyze the hollow cylinderes [I], [II], and [III], MichelFs stress function < is used. The stresses and displacef > ments are expressed as Eqs. (8) and (9). The stress functions *', *", * IU for hollow cylinders [I], [II], and [HI] are put as $' = $ / + * 2 [ + $ 3 l + *', *" = *," + * 2 " + $ 3 " + *4U and $'" = $i i n + $ 2 ln , respectively. $' is taken from solutions in which variables are separated of biharmonic function as follows. $ u and *"' are stress functions which replace the superscripts I with II or III.
9

+ ysZich(yshl)sh(ysz,)]Co(ysr) (11) * 3 7 =* 3 (^ 0 7 , ^ C07, A/, BJ, CJ, ft,7 , y Aj, Aj\ A/ 2 , A373, A374, A471, A472, A/ 3 , A474, Qj, vlt hu r, Zl) = A0l^~ + B0Izllnr+Co
+

Srvn^V(A3/1/0/

=i A ft, + A312ft/ //,/ +Aj3K0/+AJXr rKlr'} 72 + BJ [ A / 7 0 / + A4 ft/'//,/ + A/'Ko/ + Aj%,''rKlr')]sm((3j'zl)

+

i*.

92*

2s.

1 3*

S KT-i IKl -2^1)ch(7,A1)-7./fish(7,Ai))sh(7,i) frf fi* ys + 7iZich(7^1)ch(7sz,)]C0(7i/-) (12) (8) * / = *4(Aj, Bj, Cj, ft,7, y A/, A371, 2372, A373, A374, A/ 1 , A/ 2 , A473, A/ 4 , 3 / , K hu r, Zl) = - S J~T3
n=l A(P)

dz d

r or (2-e)V2*-

>

0e

( i - ) v ^ l + y d2 E drdz

a2*

[Aj{Aj%r

AjWrIlr

+ Aj3K0r

+ A3'WrKlr} + Aj{A/%r
s=i
u

+ Aj2l3jrllr + Aj3K0r
*7i

+ A/%IrKlr)]cos(JiJz1)+ J] -p^-j [{(1 - 2 ^ ) 8 ^ 7 ^ , ) , a2* i a* (9) -ysh^h(yshi))ch(yszi)+ysZish(yshl)sh{ysZi)]C0{ysr) where shfc): sinh(z), ch(z): cosh(z), 1^ J^nr), 1^': I^'r),
7

i +v (

(13) K^.

where the following equation must be satisfied: V 2 V 2 * = 0,

7 V

K,(Pr), * ' : Ktfjr), ft,' = ft,(A,) = ^ , ft, ' = ft,'(Al) = 'd^rdr+d?

7 1

-^j-

T,

Jo 7 , 5 0 7 , Co7, Aj, Bj, Cj, 1J, %', C/, AJ,

*, =*i(V> B0', Cj, A J, Bj, Cj, ft, , y A,, , A / ,

,2 ,3 _ A _ ,1 _ ,2 J

A37 , A/ , A 3 \ A/ , A/ , A / , Af, 0 / , vu hu r, z,)

_ A ,1

BJ, CJ, AJ, BJ, CJ, AJ, Bj, Cj, (, 5 = 1, 2, 3, ...) are unknown coefficients.

Journal of Pressure Vessel Technology

NOVEMBER 1991, Vol. 113 / 503

Downloaded 19 Oct 2008 to 129.5.16.227. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm