You are on page 1of 9

Akli Nechache

Ph.D. Student
e-mail: anechache@mec.etsmtl.ca

Abdel-Hakim Bouzid
Professor
e-mail: hakim.bouzid@etsmtl.ca
Ecole de Technologie Superieure,
1100 Notre-Dame Ouest,
Montreal, QC, H3C 1K3, Canada

The Effect of Cylinder and Hub


Creep on the Load Relaxation in
Bolted Flanged Joints
The leakage tightness behavior of bolted flange joints is compromised due to the high
temperature effects and, in particular, when creep of the materials of the different components of the bolted flanged joint takes place. The relaxation of bolted flanged joints is
often estimated from the creep of the gasket and the bolts. The creep behavior of the
flange ring, the hub, and the cylinder is often neglected. Apart from an acknowledgement
of relaxation due to the creep, the designer has no specific tools to accurately assess this
effect on the bolt load relaxation. The objective of this paper is to present an analytical
approach capable of predicting the bolt load relaxation due to the creep of the flange
ring, hub, and cylinder. The proposed approach is compared to the 3D finite element
models of different size flanges. An emphasis will be put toward the importance of including creep of the hub and cylinder in high temperature flange designs.
DOI: 10.1115/1.2937739

Introduction
Bolted flanged joints are widely used as a means of connection
between pressure equipments such as those found in oil refineries,
chemical, and power plants. A major problem encountered with
this type of connection is its inability to remain tight over a long
period of time. The load relaxation in high temperature applications is the major contributing factor and is due partially to the
creep of the different joint members. Other than an acknowledgement of this effect, the actual ASME Code flange design 1 does
not give a methodological assessment of the load relaxation and
its impact on the actual joint tightness 2. With the new strict
environmental protection regulations and the ever-increasing
safety and hazard requirements, some code design procedures
35 are being revised to include this effect. In addition, however, these code designs incorporate the load loss due to the thermal expansion difference of the joint members that is recognized
to cause a major sealing problem in some high temperature applications 6,7. Other than the mechanical loads, the ASME Code
flange design does not give a specific calculation procedure to
account for the relaxation of the bolt load due to the creep. In
some applications, it was found that creep has induced a loss of
gasket load of more than half of its initial preloading value.
Due to the unavailability of a more complete design procedure,
the designer is often required to make decisions on the basis of
incomplete information and takes a considerable amount of judgment gained from experience and codes of practice. As an aid to
decision making, rational analysis, which attempts to take into
account the most important features of the problem, is an essential
part of the design process.
Although recognized, the long-term creep relaxation in bolted
gasketed joints remains a subject with little research. In the literature, very few papers address analytically the effect of creep to
help engineers estimate accurately the load relaxation in bolted
joints. Creep analyses of bolted flange connections were presented
in Refs. 810. Steady creep was assumed for the flange and
bolts. The stiffening effects of the hub and the cylindrical shell
portion together with gasket creep were ignored. The paper preContributed by the Pressure Vessel and Piping Division of ASME for publication
in the JOURNAL OF PRESSURE VESSEL TECHNOLOGY. Manuscript received April 22, 2006;
final manuscript received January 4, 2007; published online August 7, 2008. Review
conducted by William Koves. Paper presented at the 2005 ASME Pressure Vessels
and Piping Conference PVP2005, Denver, CO, July 17-21, 2005.

Journal of Pressure Vessel Technology

sented in Ref. 11 extends the analytical approach of Ref. 10 by


using the strain-hardening rule to estimate the bolt load loss due to
the flange creep. However, the flexibility of the gasket and attached structural components of the joint assembly was not taken
into account. Finally, a model based on the elastic interaction of
all joint elements presented in Refs. 12,13 accurately predicts
load relaxation due to the gasket creep only. A previous paper 14
presents an extended work by considering the creep of bolts and
flange ring in the axial direction only. The present paper deals
with multiaxial creep behavior of the shell, the hub, and the
flange. The analysis of creep is investigated at a uniform temperature. In this study, the effect thermal expansion difference due to
the temperature on the load relaxation is not accounted for since
this was the subject of previous papers 15,16.

Theoretical Analysis
The current ASME Code flange design rules are based on a
rigid model and do not account for the flexibility of the different
elements of the joint. Figure 1 shows the proposed model used
that is based on a previous work 15,16. The flexibility of the
flange ring, the cylinder, the hub, the gasket, and the bolts and
their mechanical interaction is considered. To this model, strainhardening creep law is applied to the flange ring, the cylinder, and
the hub. The bolt and gasket creep is not considered as this was
the subject of the previous paper 16. The theoretical calculation
procedure that considers this effect is presented in detail here
after.
Multiaxial Creep Model. The BaileyNorton equation representing the uniaxial creep law of the steel materials is given by the
following equation:
c = A c mt n

where Ac, m, n are the constants that are functions of temperature.


The value of m is greater than 1; n is usually a fraction. This law
is intended to model only the primary and secondary creeps. This
equation can be extended to the multiaxial case by introducing the
effective quantities as
n
ce = Acm
et

where the effective stress is

Copyright 2008 by ASME

AUGUST 2008, Vol. 130 / 031211-1

Downloaded 06 Oct 2008 to 129.5.16.227. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Sz = 3 2z r

The above creep model is applied to the flange ring, the cylinder,
and the hub.
Flange Creep Analysis. For large diameter flanges, ring theory
is applied to the flange annular ring section. The ring is subjected
to the tangential and radial stresses caused by the pressure and the
rotation of the flange. It should be noted that the axial and shear
stresses are neglected.

z = r = rz = z = 0

Therefore, for this situation of plane stress,


r =

du f 1
= r + rc
dr E

uf 1
= r + c
r E

z =

r + + zc
E

Considering that tangential planes remain plane after deformation,


the radial displacement of the flange is given as follows:
u f = u fr + f z

where u fr is the extensional radial displacement independent of z


and f is the flange ring rotation. Therefore, the radial and tangential stresses are given by

r =

E
1 2

E
=
1 2

uf
du f
rc + c
+
r
dr

10

du f
uf
c + rc
+
dr
r

The internal bending moment acting the flange ring is given as


follows:


A/2

M=

t f /2

B/2

Fig. 1 Bolted flange model

t f /2

zdzdr

Substituting in the Eq. 11 gives

e =

r + z + z r
2

2 1/2

ce =

rc c 2 + c zc2 + zc rc21/2

B/2

t f /2

t f /2

E
1 2

du f
uf
rc + c
+
dr
r

E f t3f lnA/B
E

121 2
1 2


A/2

B/2

t f /2

t f /2

c + rczdzdr

3
m/n1 c n1/n
rc = 2 SrA1/n
e
c ne

3
m/n1 c n1/n
zc = rc + c = 2 SzA1/n
e
c ne

13
5

Differentiating M with respect to time t and taking small time


interval T so that the variation of M is not significant, the increment of rotation f can then be obtained:
f =

with
1

Sr = 3 2r z
1

S = 3 2 r z
031211-2 / Vol. 130, AUGUST 2008

zdzdr

Substituting the radial displacement u f of Eq. 9 and noting that


f is independent of r and z and u fr is independent of z give
M=

The corresponding strains creep rates are as follows 17:

3
m/n1 c n1/n
c = 2 SA1/n
e
c ne

12

The effective creep strain is


A/2

M=

11

12t
t3f lnA/B


A/2

B/2

t f /2

t f /2

c + rczdzdr

14

where c and rc are given previously by Eq. 5. The creep axial


displacement of the flange ring at the bolt circle relative to gasket
reaction location is therefore obtained:
Transactions of the ASME

Downloaded 06 Oct 2008 to 129.5.16.227. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

w fc =

CG
f
2

15

The radial displacement variation due to creep is analyzed. Taking


Eq. 8 into consideration:
uf =

r
r + rc = u fe + rc
E

Et3c d2uc
= M z M + Ec
12 dz2

M = M z E

Differentiating u f with respect to time t gives


17

Neglecting the variation of during a small time interval, the


creep radial displacement increment of the flange ring is given as
follows:
u f = r c t

zexdx +

tc/2

1
=
E

tc/2

tc/2


tc/2

tc/2

tc/2

1
uc
xdx =
R
E

tc/2

xdx +

tc/2

tc/2

xdx
E
t /2
c

tc/2

21

tc/2

N E
pR
+
Eze =
tc tc
2tc
and
uc N
pR E
=

+
R
2tc tc
tc

Equation 23 gives
N =

pR
Euc
E
tc +
2
R

tc/2

tc/2

zcdx

22

12
t4c

d2zc
2 xdx
dz

tc/2

tc/2

tc/2

tc/2

tc/2

tc/2

tc/2

tc/2

d2zc
xdx +
dz2

d2c
xdx
dz2

tc/2

tc/2

d2c
xdx dx
dz2

31 2

and

R2t2c

Dc =

t3c E
121 2

d 4u c
4
1 p
4 + 4 uc =
dz
Dc
2

uc =

29

2 pR2
e z
+ 3 P1 cos z + M 1cos z sin z
2Etc
2 Dc
30

Considering the rate of radial displacement and strain at time t


0, the differential equation 28 becomes
121 2
d4uc
4

4 + 4 uc =
dz
Rt3c

c dx

23

12
t3c
12
t4c

tc/2

tc/2


tc/2

tc/2

c dx

d2 zc
xdx +
dz2

tc/2

tc/2

tc/2

tc/2

tc/2

tc/2

d2 zc
xdx +
dz2

d2 c
xdx
dz2

tc/2

tc/2

d2 c
xdx dx
dz2
31

tc/2

tc/2

c dx

24

Substituting Eq. 24 into Eq. 22, the extensional strain ze is


1 2pR
uc
ze =
+
R tc
2Ectc

t3c

tc/2

tc/2

12

tc/2

c dx

At time t = 0, the creep effect does not start, and the creep strain
terms vanish so that Eq. 28 becomes the differential equation of
a shell subjected to pressure p and edge loads P1 and M 1:

tc/2

tc/2

c xdx

Referring to shell theory and expressing the membrane force N


and bending moment M in terms of and Nz and M z in terms of
z and noting that Nz = pR / 2 gives

zcxdx

tc/2

zxdx +

27

The solution of this differential equation has the form of

tc/2

tc/2

c xdx

where

d 2u c
x 2 xdx
dz

zxdx
E
t /2
c

tc/2

28

20

where z consists of two components; a uniform extensional strain


ze and a bending strain. Equations 19 and 20 are multiplied by
dx and integrated over the shell thickness. The resulting equations
are then multiplied by xdx and integrated over the shell thickness
to give
tc/2

uc 1
= z + c
R E

26

tc/2

Cylindrical Shell Creep Analysis. The analysis is based on


thin-walled cylinders subjected to internal pressure and edge loads
developed at the junction with the hub. Neglecting the radial
stress, the stress-strain equations from Ref. 18 are used with the
creep components as follows:

1
121 2
d 4u c

4
1 p+
4 + 4 uc =
dz
Dc
2
Rt3c

19

zcxdx

tc/2

Combining Eqs. 21 and 2527 and after some mathematical


manipulation, the following differential equation is obtained:

18

d 2u 1
z = ze x 2 = z + zc
dz
E

tc/2

and

16

r
du f
= r c + r
dt
E

tc/2

1
c dx +
t
c
t /2
c

tc/2

zcdx


d2uc
dz2

=
z=0

25

tc/2

After simplification of the first equation of the Eq. 21 gives


Journal of Pressure Vessel Technology

Appling the boundary conditions,

12
M
1

Dc t3c


12
t4c

tc/2

zcxdx +

tc/2

tc/2

tc/2

zcxdx +

tc/2

tc/2

tc/2

tc/2

c xdx

z=0

c xdx dx

z=0

AUGUST 2008, Vol. 130 / 031211-3

Downloaded 06 Oct 2008 to 129.5.16.227. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm


d3uc
dz3

tc/2

P1 12
=

Dc t3c
z=0

tc/2

tc/2

tc/2

12
t4c

tc/2

tc/2

d zc

tc/2

d c

tc/2

d zc
xdx +
dz

dz

dz

tc/2

tc/2

d c
xdx
dz

derivative parts of K0zei/4.


At time t 0, introducing the effect of creep Eq. 36 becomes
z=0

121 2
d2 3 d2uh
4

z
+

zu
=
h
dz2
dz2
a h 3

xdx

32

xdx dx

Equation 31 has no known analytical solution and is solved using finite difference method. However, it is necessary to know the
stresses to be able to calculate the strain rates.
Due to the redistribution of load as a result of creep of the shell
material, the stress after each time increment is

z = ze +
= e +

t
z

33

pR
Ex d uc

2tc 1 2 dz2
34

Ex d uc
pR Euc
+

tc
R
1 2 dz2

Ex d2uc
E
c + c
z =
2
2
1 dz
1 2 z
+

E
1 2 tc

tc/2

tc/2

c dx +

1
tc

tc/2

zcdx

tc/2

Euc Ex d2uc
E
c + zc
2
2
R 1 dz
1 2
+

2
E
1 2 tc

tc/2

tc/2

c dx +

tc

tc/2

zcdx

tc/2

121 2

d 2 2 d 2u h
4
z
+

zu
=
p 1
h
2
2
3
dz
dz
E h
2
where

121 2

2a2b

12
M
1,2

Dh
3z 3
z/2

12
4z 4

z/2

z/2

d2 c
xdx
dz2

zcxdx

c xdx dx

38

zcxdx +

z/2

z/2

z/2

z/2

z/2

12
4z 4

z/2

z/2

d c
xdx dx
dz

c xdx dx

z=z1=z2

39

z=z1,z2

z/2

z=z1,z2

z/2

z/2

z/2

c xdx

z/2

z/2

z/2

zcxdx +

z/2

z/2

d zc
xdx +
dz
d zc
dz

z/2

d c
xdx
dz

z/2

z=z1,z2

xdx

z=z1,z2

The stresses resulting from the redistribution of load due to the


creep of the hub material after each time increment are given by

36

z = ze +

= e +

40

where

1/4

ze =

The particular solution of this differential equation according to


coordinate z, with its reference taken starting from the point where
the thickness of the flange is zero, is given as 15
37
where = 2z. Equation 37 represents the radial displacement
of the hub at any position. The thickness of the hub at any axial
position is given by th = z, C1, C2, C3, C4 are the integration
constants, which are given from the boundary conditions. ber and
bei are, respectively, the real and imaginary derivative parts of
J0ze3i/4, ker and kei are, respectively, the real and imaginary

pah
Ex d2uh

2z 1 2 dz2
41

Ex d2uh
pah Euh
e =

+
z
ah 1 2 dz2

uh = z1/2C1ber + C2bei + C3ker + C4ker

031211-4 / Vol. 130, AUGUST 2008

z/2

z/2

12
P1,2

Dh
3z 3

z/2

z/2

z=z1,z2

Hub Creep Analysis. Based on the theory developed above,


the differential equation of the radial displacement uh of the hub
considered as a cylindrical shell with a linear variation of thickness through the axial position can be obtained 19:

d 3 d2uh
z
dz
dz2

35

z/2

with the boundary conditions given as follows:

and

d2 zc
xdx +
dz2

z/2

e =

z/2

12 d2 1
4 dz2 z

d2uh
dz2

c dx



z/2

where

ze =

z/2

z/2

z/2

12
3

z=0

and
z =

d2uh
E

Eh
+
2 x
1 dz2 1 2 z
1
z

z/2

z/2

zcdx

z/2

z/2

c dx

E
c + c
1 2 z
Transactions of the ASME

Downloaded 06 Oct 2008 to 129.5.16.227. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Ex d2uh
Euh
E

c + zc
R
1 2 dz2 1 2

2
E
1 2 z

z/2

z/2

c dx +

z/2

zcdx

z/2

f =
42

Compatibility and Elastic Interaction. The radial displacement and the rotation due to the creep, calculated at the flange to
hub and hub to cylinder junctions, are added to those given by the
elastic effect. The total radial displacement and rotation are introduced on the flexibility model to determine the redistribution of
the edge loads at each junction of the hub and the cylinder. Finally, the new bolt load and subsequently the new gasket contact
stress are evaluated after each time interval.
Radial Displacements, Rotations, and Edge Loads. The radial displacement, the rotation, and the edge loads of the individual bolted joint components of Fig. 1 needs to be considered in
the flexibility analysis in order to determine the bolt load relaxation. It is to be noted that the creep terms are added for the total
displacements and rotations.
The radial displacement and the rotation of the cylindrical shell
at the hub junction are
1
1
2 R
P1 + 2 M 1 +
p + ucc
3
2 Dc
2 Dc
2Ectc
2

uc =

c =

1
1
P1
M 1 + cc
2
2 Dc
Dc

43
44

The radial displacement, the rotation, the bending moment, and


the shear force at any axial position are given by the following
equations:
uh = z1/2C1ber + C2bei + C3ker + C4kei
+

pa2h
2 h + uch
2Ehz

45

h = z3/2C1S1 + C2S2 + C3S3 + C4S4


+

Mh =

2 hpa2h
+ ch
2Ehz2

Eh3z1/2
481 2h
+

Ph =

46

C1S9 + C2S10 + C3S11 + C4S12

2 hpa2h2

47

121 2h
Eh32z1/2
241 2h

C1S5 + C2S6 + C3S7 + C4S8


48

It should be noted that the Bessel functions are detailed in Ref.


14. The radial displacement and the rotation of the flange ring
are given in terms of the applied loading such that
uf =

B
P2
P
2E
tf

K2 + 1
tf
+ + f + ucf
2
2
K 1

Journal of Pressure Vessel Technology

49

M f + cf

50

M f is the moment acting on the flange ring and given as follows:


Mf =

The finite difference method is also used to solve the differential


equations 31 and 38 of the cylinder and hub, respectively. In
order to achieve a good convergence, it is important to choose the
appropriate finite difference time step especially in the stage of
primary creep because of the high strain rates. The time step is
adjusted according to two conditions. The amount of stress relaxation cannot be greater than 500 psi and the time step cannot be
greater than 500 h. Typically, a starting time step of 0.01 h is used
and increased gradually until 500 h after only a few hours when
secondary creep is reached.

6B
Et3f lnA/B

C G
G B 2
B
Bt f
M2
P2 +
Fb +
G + B2p
Do
2Do
2Do
16Do
51

Axial Displacement Compatibility. The system obtained from


considering the compatibility of displacement, rotation, and edge
load between the shell, the hub, and the flange ring is statically
indeterminate. To solve the system, an additional equation that
considers the axial displacements is required. This is called the
relation of axial compatibility and is obtained by considering the
axial displacement of the nut. It represents the number of turns
carried out by the nut during initial tightening. This displacement
remains unchanged during operation and any other working condition. It is calculated by the sum of all axial displacements of the
joint individual elements 15,16.
n =

w =w =w
i
e

p
e

52

c
e

n = wig + wib + 2wif = wcg + wcb + 2wcf

53

with
we =

Fe
Ke

Fib Fig
C G i Fcb Fcg
CG
+
+2
f =
+
+2
Kb Kg
2
Kb Kg
2

cf +


c
f

54
where Kb is the bolt stiffness and Kg is the gasket stiffness that
depends on the level of stress reached during bolt-up and is obtained by linear interpolation of unloading curves as described in
Ref. 20. Finally, at each interval of time, a system of 13 equations is formed to solve for the 13 unknowns, namely, C1, C2, C3,
C4, P1, M 1, P2, M 2, uc, c, uh, f , and Fb 15.

Finite Element Model


To validate the analytical model that estimates the relaxation of
the bolt load, the result from three-dimensional numerical finite
element FE modeling of four bolted gasketed joints with different sizes used in pair were compared. Because of symmetry with
respect to a plane that passes through the gasket midthickness as
well as the geometry and loading, it is possible to model only an
angular portion that includes half of the bolt and half of gasket
thickness, as shown in Fig. 2. The program developed using ANSYS 8.1 22 was used to treat a 36 in., a 52 in., and a 120 in. heat
exchanger welding neck flange and NPS3 slip-on class 150 flange.
The geometrical dimensions of the flanges are given in Table 1
while the material creep constants are given in Table 2. These
creep properties are taken from Refs. 11,21 for which the temperature is 1200 F in one case and not known in the other case.
The material creep properties are beyond the ratings of the flange
examples and exaggerate the creep effect for the purpose of comparison between the developed analytical model and the FE results. Other material characteristics used are Youngs modulus of
30 106 psi and Poissons ratio of 0.3 in the elastic range and are
supposed not to vary with temperature for simplicity.
The bolts are made of A-193 B7 material. The loading is applied in three stages. The initial bolt-up is first achieved by applying an axial displacement to the bolt to produce the initial target
bolt stress of 30 ksi to the 36 and 120 in flanges and 40 ksi to the
3 and 52 in flanges. The pressure is then applied. The hydrostatic
end thrust is simulated by an equivalent longitudinal stress applied
to the shell. Finally, the last stage, which is of the most interest in
AUGUST 2008, Vol. 130 / 031211-5

Downloaded 06 Oct 2008 to 129.5.16.227. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Fig. 3 Mechanical behavior of gasket materials

Fig. 2 3D FE model

this study, is the application of creep and the relaxation of the bolt
load over time is evaluated. It is to be noted that for the case of the
120 in heat exchanger flange creep was considered with no pressure applied. To quantify the effect of creep and the contribution
of each element of the bolted joint, the shell, the hub, and the
flange ring were subjected to creep individually and simultaneously together. To emphasize the importance of the creep be-

Table 1 Description of flange geometry


Dimension
B in.
A in.
tc in.
th in.
Hub lengh in.
t f in.
C in.
G in.
Bolt nominal diameter in.
No. of bolt

NPS 3 slip-on

36 in.
HE

52 in.
HE

120 in.
HE

3.138
7.5
0.216
0.216
1
0.8775
6
4.5
0.625
4

35
48.525
1.5625
1.8125
2.491
6.625
44.562
36.75
2
28

51
58.375
0.625
0.823
1.25
5.625
56.25
52.625
1
76

120.25
127
0.625
1.125
3.125
2.9375
124.5
122.5
1
84

Table 2 Creep constants based on NortonBailey law 11,21


Creep constants
m
n

T F

Flanges

Material

Ac

36 in. HE
52 in. HE
NPS 3 cl 150
slip-on
120 in HE

SS316
SS316
Forged
steel
Forged
steel

1.04 1026
1.04 1026
9.36 1029

5.35
5.35
5.5

0.22
0.22
1

1292
1292

9.36 1029

5.5

031211-6 / Vol. 130, AUGUST 2008

havior of only the above mentioned elements, the creep of the


gasket and bolt were purposely not considered as these were the
subject of previous papers 12,13.
Two types of gaskets were used: corrugated metal sheets
CMSs for the 52 in. and NPS3 class 150 flanges and compressed
asbestos fiber CAF for the 36 in. and 120 in. flanges. The mechanical behavior of the gasket is represented by nonlinear curves
of gasket contact stress versus axial displacement. These curves
are obtained from load-compression tests conducted on rigid platens. Figure 3 presents the room temperature test data of the two
types of gasket used in the analysis.

Discussion of the Results


The results obtained from the proposed analytical approach are
compared to those of FE models for the four different flange sizes
using the creep properties of Table 2. Again as stated in the finite
element model FEM section, the creep properties were selected
to exaggerate creep deflections and compare both the analytical
and FEMs and may not be representative of the actual long-term
creep behavior. Figures 4 and 5 show the distribution of tangential
stress across the flange ring thickness at the flange OD and its
variation with time when only the flange ring creeps. These graphs
indicate that, in general, the analytical and FEA stresses of the
36 in. and 52 in. HE flanges are in good agreement. The higher
difference is observed with the linear stress distributions present
before the creep takes place and is due to the flange ring behaving
more like a plate. There is approximately 15% difference between
an annular plate and a compact ring.
To illustrate the influence of the creep of the hub and shell,
Figs. 69 show the bolt stress relaxation caused by the creep of
the flange ring, the shell, and the hub taken separately as well
combined for the four flanges. It can be stated that in general the
results between FEM and the proposed analytical model compare
well for the larger diameter flanges. The general trend of load
relaxation compares well. Table 3 summarizes the results by giving the percentage of load relaxation of the two methods after a
total creep time of 10,000 h. This is just a little bit over a year but
is not only enough to illustrate the potential effect of including the
creep of the flange hub and shell in the analysis but also sufficient
to validate the analytical model. Up to 50% difference in load
relaxation is obtained with the NPS 3 class 150 flange. It is observed that although the shell creep causes 36% of load drop as
compared to 1450% when the creep of all members takes place,
this represents 1225% of the relaxed load after 10,000 h. 58%
Transactions of the ASME

Downloaded 06 Oct 2008 to 129.5.16.227. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Fig. 4 Stress relaxation in 52 HE flange

Fig. 7 Bolt stress relaxation in 36 in HE flange

Fig. 5 Stress relaxation in NPS 36 flange

Fig. 8 Bolt stress relaxation in 52 in HE flange

Fig. 6 Bolt stress relaxation in NPS 3 class 150 flange

Fig. 9 Bolt stress relaxation in 120 in HE flange

Journal of Pressure Vessel Technology

AUGUST 2008, Vol. 130 / 031211-7

Downloaded 06 Oct 2008 to 129.5.16.227. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Table 3 Bolt stress relaxation percent due to creep of bolted


flanged joint elements
Percent bolt relaxation
due to creep
Cylinder
Hub
Ring
Ring, cylinder and hub

NPS 3 in 36 in. 52 in. 120 in.


slip-on
HE
HE
HE
FEA
Analytical
FEA
Analytical
FEA
Analytical
FEA
Analytical

5.88
7.40

47.20
29.80
51.60
38.03

3.70
2.96
5.70
5.08
9.20
10.10
14.80
14.40

3.69
2.43
5.66
4.84
11.70
12.80
20.54
22.03

4.46
3.33
8.15
7.62
23.20
18.98
27.80
29.70

of load drop is obtained when only the hub creeps. This represents
2633% of the relaxed load. Therefore, the total contribution to
load relaxation when considering the shell and hub to creep represents about 3060% of the total combined creep.
In addition, Figs. 10 and 11 show the average axial gasket stress
variation over time caused by the creep effect of the different
elements that compose the flange. Those elements are made to
creep individually or attached to the structure and combined to-

gether to show the influence on the relaxation of the contact stress


of the gasket. These results can obviously be used to assess the
leakage tightness of the joint assembly. Nevertheless, the results
obtained from the proposed analytical method are shown to match
reasonably well with those found by finite element analysis
FEA. Table 3 summarizes the results of the relaxation of the two
methods after 10,000 h. It is observed that the creep of the hub
and the cylinder causes 15%, 24%, 30% and 50% relaxations of
the average gasket contact stress, respectively, for the 36 in.,
52 in., 120 in., and NPS 3 class 150 flanges.
It is to be noted that after 10,000 h, 30% difference is found
between the FE model and the proposed analytical theory when
only the flange ring creep is considered in the case of the NPS 3
class 150. This difference is attributed to the fact the ring theory
was used for this case instead of plate theory, which is recommended for small diameter flanges. In fact, creep analysis used in
conjunction with plate theory should give good results for small
diameter flanges as well as bolted joints used with blind cover
plates that are not treated in this paper.
Finally, this study emphasizes the importance of including the
creep of the hub and the cylinder in the analysis of the bolt load
relaxation and subsequently the gasket load relaxation in comparison to the gasket and bolt creep that were investigated in Refs.
12,13.

Conclusion
A study on the importance of including the shell and hub in the
creep analysis of bolted joints has been conducted. An analytical
model was developed to evaluate the bolt load relaxation over
time. It was found that the shell and hub contribute to up to 60%
of the total relaxed load for the examples presented in this paper.
Creep of these elements including the flange ring has been
coupled to the axial deflection compatibility equations to determine the resulting gasket and bolt load relaxations. The proposed
analytical approach based on the flexibility of the joint components has potential for possible incorporation in flange designs
once simplified.
The developed analytical models were compared to the more
accurate 3D FEA on three different size flanges. The results of the
flange ring, the bolt, and gasket stresses and their relaxation over
time compare reasonably well to those of FEA.

Nomenclature
Fig. 10 Gasket stress relaxation

Fig. 11 Gasket stress relaxation

031211-8 / Vol. 130, AUGUST 2008

t
f
Ac, m, n
ah
A
B
C
C1 C12
D
Do
E
Kb
Ke
Kg
Fb

cylinder flexural rigidity in.1


strain
strain rate
dimensionless position on hub
Poissons ratio
hub flexural rigidity in.1
stress of joint element lb/ in.2
stress rate
shearing stress lb/ in.2
time increment, h
flange rotation rad
creep constants of joint elements
hub mean radius in.
outer diameter of flange in.
inner diameter of flange in.
bolt circle diameter in.
hub constants
flexural rigidity lb in.
diameter to flange centroid in.
Youngs modulus psi
bolt uniaxial stiffness lb/in.
stiffness of joint element
gasket uniaxial stiffness lb/in.
bolt force lb
Transactions of the ASME

Downloaded 06 Oct 2008 to 129.5.16.227. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Fe
Fg
G
M
Mf
M1

M2
N
P1
P2
p
R
r
S
S1 S12

T
tf
tc
th
u
u
w
w
x

force on joint element


gasket force lb
gasket reaction force diameter in.
bending moment on shell in. lb/in.
flange moment in. lb/in.
hub to cylinder discontinuity moment
in. lb/in.
flange to hub discontinuity moment in. lb/in.
membrane force on shell lb/in.
hub to cylinder discontinuity force lb/in.
flange to hub discontinuity force lb/in.
internal pressure lb/ in.2
shell radius in.
radial position on flange ring in.
stress deviator, lb/ in.2
function depending on ber, ker, bei, kei, ber,
ker, bei, kei
exposure time h
thickness of the flange in.
thickness of the cylinder in.
thickness of the hub in.
radial displacement of the joint element in.
rate of radial displacement in./h
axial displacement of joint element in.
rate of axial displacement in./h
coordinate through the thickness of shell or
hub in.
axial coordinate in.

Superscript
c refers to creep
f refers to final state
i refers to initial state
Subscript
1
2

b
c
e
f
g
r

refers
refers
refers
refers
refers
refers
refers
refers
refers

to
to
to
to
to
to
to
to
to

cylinder to hub junction


flange to hub junction
circumferential
bolt
creep and cylinder
effective
flange
gasket
radial

Journal of Pressure Vessel Technology

z refers to axial or longitudinal

References
1 ASME Boiler and Pressure Vessel Code, 2001, Section VIII, Division 2, Appendix 2, Rules for Bolted Flange Connections with Ring Type Gaskets.
2 Payne, J. R, 1985, PVRC Flanged Joint Users Survey, Weld. Res. Counc.
Bull., 306, pp. 139.
3 EN 1591-1:2001 E, Flanges and their jointsDesign rules for gasketed circular flange connections Part 1: Calculation method.
4 EN 1591-2:2001 E, Flanges and their jointsDesign rules for gasketed circular flange connections Part 2: Gasket parameters.
5 EN 13555 2001, Flanges and their jointsGasket parameters and test procedures relevant to the design rules for gasketed circular flange connections.
6 Nechache, A., and Bouzid, A., 2002, The Redistribution of Load in Bolted
Gasketed Joints Subjected to Steady State Thermal Loading, Proceedings of
the Tenth International Conference on Nuclear Engineering-ICONE 10, Arlington, VA, ICONE10-22194, pp. 19.
7 Nechache, A., and Bouzid, A., 2003, The Determination of the Load Changes
in Bolted Gasketed Joints Subjected to Elevated Temperatures, Proceedings
of the 2003 ASME-PVP Conference, PVP-Vol.457, Analysis of Bolted Joints,
Cleveland, OH, Paper No. PVP2003-1883, pp. 139148.
8 Bailey, R. W, 1933, Flanged Pipe Joints for High Pressure and Temperatures, Engineering, 1453771, pp. 674676.
9 Marin, J., 1938, Stresses and Deformations in Pipe Flanges subjected to
Creep at High Temperature, J. Franklin Inst., 226, pp. 645657.
10 Waters, E. O, 1938, Analysis of Bolted Joints at High Temperatures, Trans.
ASME, 60, pp. 8386.
11 Kraus, H., and Rosenkrans, W., 1984, Creep of Bolted Flanged Connections,
Weld. Res. Counc. Bull., 294, pp. 28.
12 Bouzid, A., Chaaban, A., and Bazergui, A., 1995, The Effect of Creep Relaxation on the Leakage Tightness of Bolted Flanged Joints, ASME J. Pressure Vessel Technol., 117, pp. 7178.
13 Bouzid, A., and Chaaban, A., 1997, An accurate Method for Evaluating Relaxation in Bolted Flanged Connections, ASME J. Pressure Vessel Technol.,
119, pp. 1017.
14 Bouzid, A., and Nechache, A., 2004, Creep Modeling of Bolted Flange
Joints, Proceedings of the 2004 ASME-PVP Conference, PVP-Vol.478, San
Diego, CA, Paper No. PVP2004-2621, pp. 4956.
15 Bouzid, A., and Nechache, A., 2005, Thermally Induced Deflections in
Bolted Flanged Connections, ASME J. Pressure Vessel Technol., 127, pp.
394401.
16 Bouzid, A., and Nechache, A., 2005, An Analytical Solution for Evaluating
Gasket Stress Change in Bolted Flange Connections Subjected to High Temperature Loading, ASME J. Pressure Vessel Technol., 127, pp. 414427.
17 Kraus, H., 1980, Creep Analysis, Wiley, New York.
18 Burgreen, D., 1979, Pressure Vessel Analysis, C.P., Jamaica, NY.
19 Timoshenko, S. P., 1930, Theory of Plate and Shells, Wiley, New York.
20 Bouzid, A., and Champliaud, H., 2004, Contact Stress Evaluation of NonLinear Gaskets Using Dual Kriging Interpolation, ASME J. Pressure Vessel
Technol., 126, pp. 445450.
21 Betten, J., 2002, Creep Mechanics, Springer, Berlin.
22 ANSYS, 2004, ANSYS, VERSION 8.1

AUGUST 2008, Vol. 130 / 031211-9

Downloaded 06 Oct 2008 to 129.5.16.227. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm