Sei sulla pagina 1di 22

Marine Structures 21 (2008) 138159

Stress concentration factors at welds in pipelines and


tanks subjected to internal pressure and axial force
Inge Lotsberg
Det Norske Veritas, Veritasveien 1, 1322 Hvik, Norway
Received 16 March 2007; accepted 13 December 2007
Abstract
In this paper, analytical expressions for stress concentration factors in pipes subjected to internal
pressure and axial force are derived for a number of design cases based on classical shell theory. The
effect of fabrication tolerances in simple butt welds is assessed. Analyses based on classical mechanics
are compared with results from axisymmetric nite element analyses for verication of the presented
methodology. Stress concentration factors are presented for circumferential butt welds in pipes
welded together from pipes with different thicknesses, welds at buckling arrestors, welds at anged
connections in pipelines, and welds at ring stiffeners on the inside and the outside of the pipes. It also
includes stress concentration factors at end closures in pipes for gas storage. Larger pipes are
fabricated from plates with a longitudinal weld. This fabrication process introduces out-of-roundness
in the pipes. The actual out-of-roundness is a function of internal pressure. An analytical expression
for the bending stress in the pipe wall due to this out-of-roundness is presented. The derived stress
concentration factors can be used together with a hot spot stress SN curve for calculation of fatigue
damage.
r 2008 Elsevier Ltd. All rights reserved.
Keywords: Fatigue; Stress concentration factors; Circumferential welds; Ring stiffeners; Flanged connections;
Longitudinal welds; Pipes; Fabrication tolerances; Pipelines; Cylinders for compressed gas
1. Introduction
A stress concentration factor can be dened as a stress magnication at a detail due to
the detail itself or due to a fabrication tolerance with the nominal stress as a reference
ARTICLE IN PRESS
www.elsevier.com/locate/marstruc
0951-8339/$ - see front matter r 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.marstruc.2007.12.002
E-mail address: inge.lotsberg@dnv.com
value. The maximum stress is often referred to as the hot spot stress that is used in relation
with SN data for fatigue life calculation. This hot spot stress is thus derived as the stress
concentration factor times the nominal stress. Derivation of stress concentration factors at
butt welds has been presented in the literature by Maddox [1], Connely and Zettlemoyer [2]
and Lotsberg and Rove [3].
A rather detailed assessment of stress concentration factors in pipes and tubular
structural members subjected to axial force was presented by Lotsberg [4]. Effects of
fabrication tolerances, thickness transitions and ringstiffeners and bulkheads on hot spot
stress were investigated.
The welding of circumferential welds in structural members is often performed from
outside only, which is the case for the last circumferential welds made in fabrication of
jacket structures when welding braces to the stub sections in tubular joints. It was
ARTICLE IN PRESS
Nomeclature
A
r
area of ring stiffener
D exural rigidity or stiffness of the shell
E Youngs modulus
M
0
bending moment in the shell per unit circumferential length at x 0
M
x
bending moment in the shell per unit circumferential length at x
N axial force per unit circumferential length
Q
0
shear force per unit circumferential length at x 0 acting in the radial
direction of the shell
Q
x
shear force per unit circumferential length at x acting in the radial direction of
the shell
SCF stress concentration factor
W section modulus of plate and shell per unit width
d outer diameter of pipe
e
x
exp(x)
l
e
elastic length
p internal radial pressure loading
r radius of the shell measured from the axis of the cylinder to the shell mid-
surface
t thickness of shell or tubular or pipe
w radial deection of the shell
w
h
homogenous part of the radial deection of the shell
w
part
particular part of the radial deection of the shell
b, g, j, c coefcients introduced to simplify equations for SCFs
d
m
eccentricity at butt welds
n Poissons ratio
y rotation of shell
s
t
total stress
s
b
bending stress
s
nominal
nominal stress
x reduced co-ordinate
I. Lotsberg / Marine Structures 21 (2008) 138159 139
concluded by Lotsberg [4] that transition in thickness on the outside of the pipe
was preferred at circumferential welds between tubular sections with different
thickness in order to achieve good fatigue capacity for these welds. It was also
shown that it is preferred to put ringstiffeners on the inside of the tubular
members when the pipes are subjected to a longitudinal force range to achieve
long fatigue lives. It might be questioned what is most efcient considering effect of
internal pressure in this respect. The background for this together with analytical
expressions for stress concentration factors in welded pipes is presented more in detail in
the following sections.
Pipes are being used in pipelines for transportation of oil and gas and as cylinders
for storage of compressed gas. The internal pressure in the pipes gives stresses in the
circumferential direction of the pipes. The internal pressure also gives longitudinal
stress in the pipe wall due to the end cap effect. This nominal longitudinal stress
due to internal pressure is only half that of the circumferential stress. However,
increased wall thickness or stiffeners may be required at supports and at crack
arrestors in deep water pipelines (e.g. [5]). This may introduce stress concentrations
that make the stress cycling in the axial direction of the pipes important. It should also be
noted that the longitudinal welds are normally made from both sides allowing for a
rather good SN curve while circumferential welds are normally made from the outside
only and a less good SN curve apply to the root of the weld as compared with
the outer weld toe (refer Fig. 1). Reference is made to fatigue design standards such as IIW
[6] and to DNV-RP-C203 [7] for description of applicable SN curves. Axial stresses in the
pipes may also occur due to external forces such as wave loading and due to change in
temperature.
Pipes made of high strength steel can be subjected to large stress ranges and the
welds can only sustain a limited number of full stress cycles from lling and emptying of
the pipes until failure occurs. In fabricated pipes there will always be some imperfections
left from fabrication. Residual out-of-roundness in the pipes after fabrication is
an important parameter in this respect as it together with the membrane stress in
the circumferential direction in the pipe wall introduces local bending stresses. The
membrane stress and the local bending stress, which are acting normal to the longitudinal
weld may be governing the fatigue lives of these elements due to stress cycles from
lling and emptying of the pipes. This problem is further considered in the last section
of this paper.
ARTICLE IN PRESS
Outside of pipe
t

m
Weld root
Weld toe
L
t
d
Fig. 1. Section through a single side weld.
I. Lotsberg / Marine Structures 21 (2008) 138159 140
2. Classical shell theory
2.1. Cylindrical shell
The differential equation for the cylindrical shell shown in Fig. 2 can be found in the
literature (e.g. [8])
D
q
4
w
qx
4

Et
r
2
w px
n
r
N (1)
where D is the exural rigidity of the shell dened as
D
Et
3
121 n
2

(2)
where E is the Youngs modulus, n is the Possons ratio, t is the shell thickness, r is the
radius of the cylindrical shell measured from the axis of the cylinder to the middle of the
shell surface, p(x) is the internal radial loading and N is the axial force per unit
circumferential length.
The solution of the homogenous part of the differential equation (1) can be expressed as
w
h

l
2
e
2D
M
0
g
4
x Q
0
l
e
g
1
x (3)
where an elastic length is dened as
l
e

rt
p

31 n
2

4
_ (4)
For steel with n 0.3 the elastic length becomes
l
e
0:78

rt
p
(5)
and with reference to Fig. 2 the following denitions are made:
x
x
l
e
g
1
x e
x
cos x
g
2
x e
x
sin x
g
3
x g
1
x g
2
x
g
4
x g
1
x g
2
x 6
ARTICLE IN PRESS
N
M
0
w
Q
0
X,
2r
d
p (x)
t
Fig. 2. Circular cylindrical shell loaded symmetrically with respect to its axis.
I. Lotsberg / Marine Structures 21 (2008) 138159 141
The particular part can be expressed for p(x) as a polynomial in less or equal third
degree as
w
part

r
2
Et
px
nr
Et
N (7)
The total displacement, w, is the sum of the homogenous and the particular part. Then
for constant p
qw
qx

l
e
2D
2M
0
g
1
x Q
0
l
e
g
3
x (8)
The moment (per unit circumferential length) at a section x (Fig. 2) is calculated as
M
x
D
q
2
w
qx
2
M
0
g
3
x Q
0
l
e
g
2
x (9)
and the shear force (per unit circumferential length) is derived as
Q
x

qM
x
qx

2M
0
g
2
x
l
e
Q
0
g
4
x (10)
2.2. Spherical shells
Reference is made to Flugge [9] and Spence and Tooth [10] for general theory on
spherical shells. In the following the relations between the forces and rotations at the edge
of the sphere shown in Fig. 3 are presented. From these books the following equation is
ARTICLE IN PRESS
M
0s
Q
0s
p
a
t
s
N = pa/2
w
N = pa/2
Q
0s
M
0s
Fig. 3. End sphere for gas storage pipe.
I. Lotsberg / Marine Structures 21 (2008) 138159 142
derived for radial displacement:
w
s

2Q
0s
al
Et
s

2M
0s
l
2
Et
s
(11)
and for rotation
qw
s
qx

2Q
0s
l
2
Et
s

4M
0s
l
3
Eat
s
(12)
where
l

31 n
2

4
_

a
t
s
_
(13)
The radial displacement due to pressure p is
w
p

pa
2
2Et
s
1 n (14)
There is no rotation at the edge for a constant pressure p.
3. Fabrication tolerances at a butt weld in a plate
The effect of fabrication tolerance at a simple butt weld in a plate with respect to
additional stress is illustrated in Fig. 4. The plate is subjected to an axial force per unit
width of N s
a
*t. The fabrication tolerance results in a shift in neutral axis equal to d
m
at
the butt weld. This shift in neutral axis implies a bending moment in the plate at both sides
of the weld of M N*d
m
/2, as there will be an inection point in the beam model in the
centre of the weld. From Fig. 4 it is observed that the moment changes sign over a short
length. The validity of classical beam theory might be questioned in this respect. Therefore,
this has been investigated more in detail for plate structures by Lotsberg and Rove [3]. It is
also assessed further by a ne nite element mesh analysis reported in Section 4. The
section modulus for a plate of unit width is W t
2
/6. Thus the bending stress in the plate
at the weld becomes
s
b

M
W

Nd
m
=2
t
2
=6

3Nd
m
t
2

3s
a
d
m
t
(15)
ARTICLE IN PRESS
N

m
M = N*
m
/ 2
Fig. 4. Moment distribution at a plate with an eccentric butt weld.
I. Lotsberg / Marine Structures 21 (2008) 138159 143
By denition the stress concentration factor at the weld toe in Fig. 4 becomes
SCF 1
3d
m
t
(16)
where d
m
is the fabrication tolerance as shown in Fig. 4 and t is the pipe wall thickness.
4. Circumferential butt welds in pipes at thickness transitions and with fabrication tolerances
Additional stresses at butt welds in pipes resulting from thickness transitions and
fabrication tolerances can be derived by use of shell theory as shown by Lotsberg [4]. For
the outer side of a tubular subjected to axial stress, as shown in Fig. 5, a stress
concentration factor was derived as
SCF 1
6d
t
1
1
1 t
2
=t
1

2:5
e
a
(17)
where d
1
2
(t
2
t
1
) is the shift in neutral axis as shown in Fig. 5 and
a
1:82L

d t
1
t
1
_
1
1 t
2
=t
1

2:5
(18)
where L is the length for transition in thickness as shown in Fig. 5.
For the inner side
SCF 1
6d
t
1
1
1 t
2
=t
1

2:5
e
a
(19)
Considering the axial stress in the pipe resulting from end cap effect from internal
pressure (or from other axial force in the pipes) the following stress concentration factor
can be used for the weld toe side in Fig. 1
SCF 1 3
d
m
t
e

t=d
p
(20)
This equation is derived from Eq. (17) by putting L 0.91 times the width of the weld.
This stress concentration factor should be used if the pipe is welded from both sides. It is
conservative to use the same stress concentration factor for the root side in symmetric
welds (V-shaped weld groove) as explained in Lotsberg [4] and DNV-RP-C203 [7]. The
bending stress at the root of the weld in Fig. 1 is close to zero following the moment
distribution shown in Fig. 4. An axisymmetric model is made of a pipeline section with
diameter 42
00
and wall thickness 31.0 mm in order to assess this further, refer to Lotsberg
ARTICLE IN PRESS
t
2
Outside 4
1
Inside

L
Neutral
axis

nominal
t
1
Fig. 5. Preferred transition in thickness at tubular butt weld when subjected to external axial force.
I. Lotsberg / Marine Structures 21 (2008) 138159 144
and Holth [11]. Two different geometries were analysed as shown in Fig. 6:
One single side welded connection with a fabrication tolerance equal to 2 mm as shown
in Fig. 6a, which also shows a shift in neutral axis equal to 2 mm.
One analysis with the same geometry at the notch region, but with added material on
the right pipe section such that the connection is without shift in neutral axis (Fig. 6b).
ARTICLE IN PRESS
B A
r = 1mm h = 0.05*t
r = 1mm
2 mm
h = 0.05*t
15
t
30
2 mm
t
B A
r = 1mm
2 mm
15
30
Fig. 6. Single side welded connections: (a) eccentric connection; and (b) symmetric connection (aligned neutral
axes).
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
2.2
0 5 10 15 20 25 30 35
S
t
r
e
s
s

(
M
P
a
)
Eccentric 2 mm
Symmetric
SCF eq. (17)
Distance from inner side (mm)
Fig. 7. Stress in axial direction in section AA.
I. Lotsberg / Marine Structures 21 (2008) 138159 145
The notch at the weld root and at the two weld toes in Fig. 1 is modelled by a radius
1.0 mm. This requires modelling by a ne mesh typically that recommended for the notch
stress methodology (ref. Appendix D of DNV-RP-C203 [7]). The connections were
subjected to a unit axial load over the 31 mm thick section. Stresses at different sections
AA and BB are presented in Figs. 7 and 8, respectively.
First the stress in the direction normal to line AA in Fig. 7 is considered. The left part
of Fig. 7 shows stress at the inner side of the pipe and the right part shows stress at the
outer side. At the outer side there is a peak stress resulting from the notch at the weld toe.
More interesting in this respect is the stress at the inner side which is less affected by
notches (see Fig. 6). Here the stress is compared with that resulting from Eq. (17), where L
is put equal to the width of the weld. It is seen that the calculated stress for eccentricity
2 mm is slightly larger than that from the linear stress distribution from Eq. (17). The
calculated stress distribution at the inner side seems to be slightly disturbed by the presence
of the weld notch. The same disturbance is also present for the situation without shift in
neutral axis, but with presence of the root notch (symmetric geometry). Based on this the
global stress behaviour represented by Eq. (17) is considered to be in agreement with the
numerical results from the nite element analyses. The results using different equations for
calculation of stress concentration factors are shown in Table 1.
Next the stress distribution through section BB is considered (refer Fig. 8). It is
observed that the gradients of both the stress distributions for the two geometries are small
away from the notches indicating that a shift in neutral axis does not give signicant beam
ARTICLE IN PRESS
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
2.2
2.4
2.6
0 5 10 15 20 25 30 35
Distance from inner side (mm)
S
t
r
e
s
s

(
M
P
a
)
Eccentric 2 mm
Symmetric
Fig. 8. Stress in axial direction in section BB.
Table 1
Calculated stress concentration factors
Methodology used for calculation Stress concentration factor
Axisymmetric nite element model 1.188
SCF Eq. (16) 1.193
SCF Eq. (17) 1.178
SCF Eq. (20) 1.163
I. Lotsberg / Marine Structures 21 (2008) 138159 146
bending stress at the root that needs to be included in a fatigue analysis. However, the
notch stress due to the local root geometry needs to be accounted for. The SN curves in
DNV-RP-C203 [7] for the root side that accounts for this effect can be used.
Direct read out of the maximum stress in the root in the nite element model
representing Fig. 6a gives s
notch
2.62 MPa for unit axial load and the maximum stress in
the root in Fig. 6b gives s
notch
2.45 MPa. (The maximum stress is a surface stress which
is slightly different from that of the axial stress presented in Figs. 7 and 8). This results in a
stress concentration factor SCF 2.62/2.45 1.069. As the geometry is the same except
for different shift in neutral axis one may use this result to assess an alternative equation
for stress concentration factor for the root as
SCF 1 k
d
m
t
e

t=d
p
(21)
Here only k is unknown and solving this equation gives k 1.2.
From this it is seen that stress concentration factors due to eccentricity after fabrication
at weld roots of single sided symmetric welds are signicantly less than for the weld toe
(refer Eq. (20)).
It is shown here that additional stress at the hot spot resulting from shift in neutral may
be accounted for by analytical expressions. Local stress increase due to local notches
should be accounted for in the SN curve if not a notch stress approach is used such as
presented as alternatives in IIW [6] or DNV-RP-C203 [7]. Reference may also be made to
local approaches presented by Radaj et al. [12].
In general the stress concentration factor to be used is also linked to the SN curve that
is recommended for the fatigue assessment, refer also Lotsberg and Sigurdsson [13].
5. Circumferential butt welds in pipes with different thicknesses
Local bending stresses in pipes will normally be present at connections welded together
from pipes with different thickness such as at pipe supports and buckling arrestors. A
welded connection between pipes with different thickness is considered in the following
(see Fig. 9). The following derivation is based on an assumption of a centric neutral axis
going from one thickness to the next as shown in Fig. 9a. Then a shift in neutral axis is
considered at the end of this section with thickness transitions on the outer side (Fig. 9b) or
the inner side of the pipe. For derivation of an equation for stress concentration factor at
the circumferential weld the following requirements are dened at the welded connection:
The radial displacement of pipe no. 1 is equal to that of pipe no. 2.
The rotation of pipe no. 1 is the same as for no. 2.
The moments around the circumference in the two pipes are the same.
There is continuity in shear force.
Mathematically these requirements can be expressed as
w
1
w
2
(22)
qw
qx
_ _
1

qw
qx
_ _
2
(23)
ARTICLE IN PRESS
I. Lotsberg / Marine Structures 21 (2008) 138159 147
M
0

1
M
0

2
(24)
Q
0

1
Q
0

2
(25)
The axial force in the pipes due to internal pressure can be calculated as (end cap force
divided by pipe area)
N
pr
2
(26)
Then the particular part from Eq. (7) can be expressed as
w
part

pr
2
Et
1
n
2
_ _
(27)
From Eqs. (3), (22), (24), (25) and (27)
w
l
2
e1
2D
1
M
0
Q
0
l
e1

pr
2
Et
1
1
n
2
_ _

l
2
e2
2D
2
M
0
Q
0
l
e2

pr
2
Et
2
1
n
2
_ _
28
From Eqs. (8), (23)(25)
qw
qx

l
e1
2D
1
2M
0
Q
0
l
e1

l
e2
2D
2
2M
0
Q
0
l
e2
(29)
ARTICLE IN PRESS
a) Centric neutral axis
p
b) Shift in neutral axis
at thickness transition
p
c) Sloped thickness
transition
1
4
p
t
2 t
1
t
1
t
2
t
2
t
1
Support
Fig. 9. Thickness transition in pipes at support and buckling arrestors: (a) centric neutral axis; (b) shift in neutral
axis at thickness transition; and (c) sloped thickness transition.
I. Lotsberg / Marine Structures 21 (2008) 138159 148
Then from Eqs. (28) and (29)
M
0

prt
2
1
2 n
4

31 n
2

_
g
1
t
1

1
t
2
_ _
(30)
and
Q
0

2t
2:5
1
t
2:5
2

t
2:5
2
t
0:5
2
t
2
1
M
0
l
e1
(31)
where
g
2t
2:5
1
t
2:5
2

t
2:5
2
t
0:5
2
t
2
1
1
t
1
t
2
_ _
1:5
_ _

t
1
t
2
_ _
2
1 (32)
The bending stress at the connection in the thinnest pipe is obtained as
s
b

M
0
W

6M
0
t
2
1

pr
g
2 n
2

3
1 n
2

1
t
1

1
t
2
_ _
(33)
It is assumed that the radius rbt
1
. The stress due to the end cap pressure is calculated as
s
a

pr
2t
1
(34)
The total stress at the inner side and the outer side is calculated as
s
t
s
a
s
b
(35)
This equation can also be written as
s
t
s
a
1
s
b
s
a
_ _
s
a
SCF (36)
where the stress concentration factor is
SCF 1
s
b
s
a
(37)
For the inner side of the pipe
SCF 1
2 n
g

3
1 n
2
_
1
t
1
t
2
_ _
(38)
For the outer side of the pipe
SCF 1
2 n
g

3
1 n
2
_
1
t
1
t
2
_ _
(39)
The equation for moment in Eq. (30) is valid for all thickness relations. The derivation
of equations for stress concentration factors (Eqs. (38) and (39)) is based on calculation of
nominal stress in the pipe section with thickness t
1
. It should be noted that Eqs. (38) and
(39) are derived based on a centric neutral axis as shown in Fig. 9a. Calculated stress
concentration factors using these equations are presented in Fig. 10. The results from one
ARTICLE IN PRESS
I. Lotsberg / Marine Structures 21 (2008) 138159 149
geometry using nite element analysis using PREFEM [14] are included as check.
It is observed that there is some difference in the calculated stress concentration factors.
This may be explained by some difference in the stiffness properties of the nite element
model as compared with the classical analytical shell model. One 20-node isoparametric
element, representing a linear stress distribution, is used to represent one thickness in
pipe 1. However, 2 elements are used to model pipe 2 in order to connect the two pipes.
This gives a slightly more exible pipe 2 in the nite element model than in the analytical
approach.
It is observed from Fig. 10 that the largest stress is found at the inner side of the pipe.
An eccentric thickness transition as shown in Fig. 9b also introduces local bending over
the pipe wall when subjected to axial forces as explained in Section 4. The information
ARTICLE IN PRESS
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.3 1.5 2.0 2.3 2.5
S
t
r
e
s
s

c
o
n
c
e
n
t
r
a
t
i
o
n

f
a
c
t
o
r
Inner side
Outer side
FE inner side
FE outer side
Thickness ratio t
2
/t
1
1.0 1.8
Fig. 11. Stress concentration factor with step in thickness on outside.
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
1.3 1.5 1.8 2.0 2.3 2.5
SCF inner side
SCF outer side
FE inner side
FE outer side
S
t
r
e
s
s

c
o
n
c
e
n
t
r
a
t
i
o
n

f
a
c
t
o
r
Thickness ratio t
2
/t
1
1.0
Fig. 10. Stress concentration factor for longitudinal stress from internal pressure neglecting thickness transition
(or shift in neutral axis).
I. Lotsberg / Marine Structures 21 (2008) 138159 150
from Section 4 can be used to derive stress concentration factors for eccentric thickness
transitions by using the superposition principle.
For the inner side of the pipe with thickness transition on the outside
SCF 1
2 n
g

3
1 n
2
_
1
t
1
t
2
_ _

3t
2
t
1

t
1
1
1 t
2
=t
1

2:5
e
a
(40)
The parameter a is dened by Eq. (18).
For the outer side of the pipe with thickness transition on the outside
SCF 1
2 n
g

3
1 n
2
_
1
t
1
t
2
_ _

3t
2
t
1

t
1
1
1 t
2
=t
1

2:5
e
a
(41)
The calculated stress concentration factors using these equations are presented in Fig. 11
together with results from a nite element analysis. It is seen from Fig. 11 that a transition
in thickness on the outside of the pipe reduces the stress concentration factor at the inner
side as compared with that without shift in neutral axis shown in Fig. 10.
For the inner side of the pipe with thickness transition on the inner side
SCF 1
2 n
g

3
1 n
2
_
1
t
1
t
2
_ _

3t
2
t
1

t
1
1
1 t
2
=t
1

2:5
e
a
(42)
The parameter a is dened by Eq. (18).
For the outer side of the pipe with thickness transition on the inner side
SCF 1
2 n
g

3
1 n
2
_
1
t
1
t
2
_ _

3t
2
t
1

t
1
1
1 t
2
=t
1

2:5
e
a
(43)
It should be noted that these stress concentration factors applies only to the axial stress
that results from internal pressure. And the stress concentration factors are to be used
together with the nominal axial stress resulting from end cap force or internal pressure
only. For axial stress in the pipe wall resulting from external forces the stress concentration
factors from Eqs. (17)(19) should be used. Thus for design it is necessary to separate the
stress into these two stress components before the resulting stress range is calculated.
Thus the following superposition of stress conditions should be used for derivation of
hot spot stress at the circumferential weld:
Ds
hot spot
Ds
Internal axial pressure
SCF
Internal pressure
Ds
External axial
SCF
External axial
44
where Ds
Internal axial pressure
is the stress in the axial direction of the pipe due to internal
pressure, Ds
External axial
is the stress in the axial direction of the pipe due to external axial
force, SCF
Internal pressure
is the stress concentration factors from Eqs. (40) to (43) depending
on position of thickness transition and considered hot spot, SCF
External axial
is the stress
concentration factors from Eqs. (17)(19) depending on position of thickness transition
and considered hot spot.
In many practical cases a slope transition equal to 1:4 is made for transition from one
thickness to another as indicated in Fig. 9c. Finite element analyses of such geometries
shows that the effect of internal pressure on bending stress at the weld region reduces (refer
to Eqs. (38) and (39) representing effects from different pipe stiffness). Then the main
ARTICLE IN PRESS
I. Lotsberg / Marine Structures 21 (2008) 138159 151
contribution to local bending stress is from axial force at thickness transitions and
fabrication tolerances represented by Eqs. (17)(19).
6. Ring stiffeners and buckling arrestors in pipelines
A ringstiffener in a cylindrical shell as shown in Fig. 12 is considered. This applies also to
a short buckling arrestor in a pipeline and a bolted ange connection that is frequently
used in risers for oil and gas production (refer Fig. 13, where the most critical hot spot is
found at point B). The inward displacement of a ring subjected to a radial force (per unit
length) 2Q
0
on the outside can be calculated as
w
ring

2Q
0
r
2
EA
r
(45)
where A
r
is the area of ringstiffener. (For a anged connection A
r
is the area of the two
anges.)
At a ring stiffener the rotation at the shell is zero due to symmetry (see Fig. 12), and
from Eq. (8)
qw
qx
0 ) M
0

Q
0
l
e
2
(46)
The displacement of the ring is equal that of the shell at the connection which gives from
Eqs. (3), (27), and (45), for x 0.

2Q
0
r
2
EA
r

l
2
e
2D
M
0
Q
0
l
e

pr
2
Et
1
n
2
_ _
(47)
ARTICLE IN PRESS
2r
t
A
r
Deformed shape
p
Fig. 12. Ring stiffener on pipe with internal pressure.
B
p
Fig. 13. Section through bolted ange connection.
I. Lotsberg / Marine Structures 21 (2008) 138159 152
Combining Eqs. (46) and (47) and inserting for D from Eq. (2) the moment in the
cylindrical shell at a ring stiffener is obtained as
M
0

pr
2
t
2
A
r
l
e
2 n
8r
2
t
3
12A
r
l
3
e
1 n
2

(48)
By dividing by the section modulus the bending stress in the shell is obtained as
s
b

3
p
2 n

1 n
2
p
1
b
pr
2t
(49)
where b is dened as
b 1
2t

rt
p
A
r

31 n
2

4
_ (50)
And the stress concentration at a ring stiffener is obtained as
SCF 1

3
p
2 n

1 n
2
p
1
b
(51)
where the plus sign applies to the inner side and minus to the outer side. This stress concentration
includes effect from internal pressure and end cap pressure. It is to be used together with the
nominal stress acting in the axial direction of the pipe wall due to end cap pressure.
With n 0.3 for steel the expression for b becomes
b 1
1:56t

rt
p
A
r
(52)
and the stress concentration factor for the inner side becomes
SCF 1
3:087
b
(53)
and the stress concentration factor for the outer side becomes
SCF 1
3:087
b
(54)
Due to the notch of the weld itself the fatigue strength of the weld at the ring stiffener
itself becomes less than for the other shell side. As the stress is lesser on the outside than at
the inside it is thus recommended to place ring stiffeners on the outside of a shell structure
subjected to internal pressure. This is a different conclusion from that of ringstiffeners in
tubular members subjected to pure external axial force (see Ref. [4]).
From Eq. (51) it is seen that the stress concentration factor at the inside of the pipe at a
ringstiffener becomes signicant. Example pipe for CNG with outer diameter 42
00
and
thickness 31 mm. With a atbar ringstiffener 30 200 the SCF on inside becomes 2.53,
while for the outside it becomes 0.53.
The same equations as used for a ring stiffener can also be used for assessment of stress
concentration factors at an open bulkhead by calculating an effective area as
A
r

rt
b
1 n
(55)
where t
b
is thickness of the bulkhead.
ARTICLE IN PRESS
I. Lotsberg / Marine Structures 21 (2008) 138159 153
7. Stress concentrations at transitions from pipe to end spheres
The following derivation is based on an assumption of centric neutral axes going from
the pipe to a sphere. An eccentric transition from a pipe to a sphere can be included similar
to that derived for thickness transitions in pipes (see Section 5). The methodology to derive
stresses at the transition between the cylindrical pipe and the end spheres is similar to that
used to derive stress distributions at thickness transitions. This means compatibility in
displacements, rotations, bending moment and shear force (refer to Eqs. (22)(25)). By
putting a r and from Eqs. (3), (11), (14), (22), (24), and (25) the following equation is
derived:
w
l
2
e
2D
M
0
Q
0
l
e

pr
2
Et
c
1
n
2
_ _

2l
2
M
0
Et
s

2lrQ
0
Et
s

pr
2
2Et
s
1 n 56
From Eqs. (8), (12), (23), (24) and (25) the following equation is derived:
qw
qx

l
e1
2D
2M
0
Q
0
l
e

4l
3
M
0
Et
s
r

2l
2
Q
0
Et
s
(57)
From Eqs. (56) and (57) the bending moment at the junction is derived
M
0

pr
3

31 n
2

4
_
c
j
1 n
t
s

2 n
t
c
_ _
(58)
and the stress concentration factor for the inner side is obtained as
SCF 1

3
1 n
2
_
c
j
r
2
t
c
1 n
t
s

2 n
t
c
_ _
(59)
and for the outer side
SCF 1

3
1 n
2
_
c
j
r
2
t
c
1 n
t
s

2 n
t
c
_ _
(60)
where
j r
2
c
1
t
2
c

1
t
2
s
_ _
2r
1:5
1
t
1:5
c

1
t
1:5
s
_ _
(61)
and
c
t
2
c
t
2
s

t
2:5
c
t
2:5
s

t
s
t
c
r
_
(62)
where t
c
is the thickness of cylinder and t
s
is the thickness of end sphere.
The equation for moment in Eq. (58) is valid for all thickness relations. The derivation
of equations for stress concentration factors (Eqs. (59) and (60)) is based on calculation of
nominal stress in the pipe section without shift in neutral axis going from pipe to sphere.
The resulting stress concentration factors are presented graphically in Fig. 14.
ARTICLE IN PRESS
I. Lotsberg / Marine Structures 21 (2008) 138159 154
The shear force is obtained from Eqs. (56) and (57) as
Q
0

2

31 n
2

4
_
jr
M
0
(63)
Reference is made to Section 5 for considerations of eccentric neutral axes. A similar
term as the last terms in Eqs. (40) and (41) may be added to Eqs. (59) and (60) as an
approximation to include the effect of an eccentric thickness transition.
8. Longitudinal welds with bending stress over the pipe wall resulting from out-of-roundness
of fabricated pipes
The out-of-roundness of fabricated pipe elements results in increased stress due to a
bending moment over the wall thickness (see Fig. 15). The eccentricity due to out-of-
roundness is a function of tension in the hoop direction of the pipe, which is reduced as the
internal pressure is increased and the hoop tension is increased. Thus the bending stress
over the wall thickness is a nonlinear function of the internal pressure. Assuming that the
moment M results from an eccentricity d, where hoop tension is accounted for in the
analysis, the following derivation of a stress concentration factor is performed. It is
assumed that the out-of-roundness results in an eccentricity d
0
without any hoop tension
force from internal pressure. In order to simplify the calculation, the ring is transformed to
an equivalent beam model as shown in Fig. 15. The circumference of a pipe between an
inection point, where the bending moments are zero, and a symmetry point with
maximum moment is considered. Thus, the length l in the beam model becomes l pd/8.
The relation between bending curvature and bending moment on a beam can be expressed
according to classical beam theory as
q
2
y
qx
2

Mx
EI
(64)
where E is the Youngs modulus and I is the moment of inertia of pipe wall section
ARTICLE IN PRESS
0.0
0.5
1.0
1.5
2.0
2.5
1.00 1.25 1.50 1.75 2.00 2.25 2.50
S
t
r
e
s
s

c
o
n
c
e
n
t
r
a
t
i
o
n

f
a
c
t
o
r
SCF Inner side
SCF Outer side
Thickness ratio t
s
/t
c
Fig. 14. Stress concentration factor at weld to sphere as function of thickness ratio.
I. Lotsberg / Marine Structures 21 (2008) 138159 155
The eccentricity along the beam can be expressed as
dx
d
0
l
x (65)
and the moment along the pipe wall beam is
Mx dx yN (66)
By combining Eqs. (64)(66) this results in the following differential equation:
q
2
y
qx
2

N
EI
y
Nd
0
x
EIl
0 (67)
The general solution of Eq. (67) is
y C
1
e
lx
C
2
e
lx

d
0
x
l
(68)
where
l

N
EI
_
(69)
where N is the membrane stress in the circumferential direction times wall thickness for
unit length of the pipe.
The boundary conditions in the model shown in Fig. 15 are:
y 0 for x 0
qy
qx
0 for x l 70
These boundary conditions are used to determine the integration constants and the
results are
C
1

d
0
l
1
le
ll
e
ll

and C
2
C
1
(71)
Maximum eccentricity at x l is then obtained as
d
max
d
0
y
max

d
0
ll
tanhll (72)
ARTICLE IN PRESS
d
Inflection point

0
Maximum bending
moment
l = d/8

max
y
N
X
l
Fig. 15. Model for analysis of out-of-roundness.
I. Lotsberg / Marine Structures 21 (2008) 138159 156
The maximum stress at x l is obtained from
s
N
A

M
W

N
t

6Nd
max
t
2
(73)
This stress can also be presented as
s
N
t
SCF (74)
where the stress concentration factor is
SCF 1
6d
max
t
(75)
where d
max
is eccentricity as function of the axial force N as obtained from Eq. (72), and t is
thickness of the pipe. In terms of out of roundness the equation for stress concentration
factor can be derived as
SCF 1
1:5d
OOR
tll
tanh ll (76)
where the out of roundness is dened as d
OOR
d
max
d
min
, l pd/8 and with l from Eq.
(69), which is a function of the membrane hoop stress s
m
as follows:
l

12s
m
Et
2
_
(77)
Results by use of this equation are derived for a pipe with 42
00
diameter and wall
thickness 31.8 mm using tolerance requirements from DNV-OS-F101 [15], which is
referred to in the DNV rules for compressed gas, DNV [16]. For standard dimensional
requirements for line pipes the maximum allowable d
OOR
d
max
d
min
15 mm. For
enhanced dimensional requirements for line pipes the maximum allowable d
OOR
10 mm.
The results for this tolerance are shown in Fig. 16. The results have been compared with
nite element analysis using the Abaqus program [17] where nonlinear geometry is
accounted for. It is observed that the analytical approach for derivation of bending stress
in the pipe wall provides approximately the same results as derived from the Abaqus
analysis.
The results are also compared with the following equation from [18]:
s
b
s
m
1:5d
max
d
min

tf1 0:5p
m
1 n
2
=Ed t=t
3
g
(78)
where p
m
is the maximum pressure at the operating condition being assessed. In section for
remarks, it is stated that If under fatigue loading p
m
varies, use the mean value during the
time interval considered. s
m
is the membrane stress in the circumferential direction due to
internal pressure.
Assuming that the results from Abaqus analysis are the most accurate, it is observed that
the bending stress from BS 7910 is non-conservative using maximum pressure. However,
using mean pressure, as recommended in the BS 7910 document, Eq. (78) provides safe
results close to that from the Abaqus analysis.
ARTICLE IN PRESS
I. Lotsberg / Marine Structures 21 (2008) 138159 157
9. Closure
In this paper, analytical expressions for stress concentration factors in pipes subjected to
internal pressure and axial force are derived for a number of design cases based on classical
shell theory. The effect of fabrication tolerances in simple butt welds is assessed.
Analyses based on classical mechanics at circumferential welds are compared with
results from axisymmetric nite element analyses for verication of the presented
methodology. The results from these analyses are found to be in good agreement with the
proposed analytical equations for stress concentration factors that are derived based on
classical theory of elasticity.
It is shown that stress concentration factors due to eccentricity after fabrication at weld
roots of single sided symmetric welds are signicantly less than that for the weld toe.
Stress concentration factors are presented for circumferential butt welds in pipes
welded together from pipes with different thicknesses, welds at buckling arrestors, welds at
anged connections in pipelines, and welds at ring stiffeners on the inside and the outside
of the pipes.
These stress concentration factors can be used for fatigue assessment of pipelines when
the pipelines are subjected to variation of internal pressure such as at start and stops of
operation. The same stress concentration factors can also be used for fatigue assessment of
pipes for gas transportation, where each lling and emptying of the pipes gives a stress
cycle that contributes to fatigue damage accumulation. Also a stress concentration factor
for end closures in pipes for gas storage is presented.
Larger pipes are fabricated from plates with a longitudinal weld. This fabrication
process introduces out-of-roundness in the pipes. A pressure variation from internal
pressure in the pipe leads to variation of local bending in the pipe wall resulting from these
fabrication tolerances. A design methodology to account for the resulting stresses has been
presented. The analytical expressions for stress concentration factors can be used in fatigue
design of pipelines and cylinders used for transportation of gas.
ARTICLE IN PRESS
0
20
40
60
80
100
120
140
160
180
200
0 50 100 150 200 250 300
Pressure (bars)
S
t
r
e
s
s

(
M
P
a
)
Membran stress
Bending stress analytical
Bending stress ABAQUS
BS7910 mean p
BS7910 max p
Fig. 16. Example of membrane and bending stress in hoop direction of 42
00
diameter pipe.
I. Lotsberg / Marine Structures 21 (2008) 138159 158
The derived stress concentration factors can be used together with a hot spot stress SN
curve for calculation of fatigue damage.
Acknowledgements
The axisymmetric nite element analyses performed by Per Arvid Holth and Andrzej
Kludka are acknowledged. Also the nonlinear analysis of bending stress in the pipe due to
out-of-roundness performed by Rikard Trnkvist is appreciated.
References
[1] Maddox SJ. Fitness for purpose assessment of misalignment in transverse butt welds subjected to fatigue
loading. London: International Institute of Welding, IIW Document XIII-1180, 1985.
[2] Connely LM, Zettlemoyer N. Stress concentration at girth welds of tubulars with axial wall misalignment. In:
Proceedings of the international conference on tubular structures. London: E & F N Spon; 1993.
[3] Lotsberg I, Rove H. Stress concentration factors for butt welds in stiffened plates. OMAE New Orleans,
ASME, 2000.
[4] Lotsberg I. Stress concentration factors at circumferential welds in tubulars. J Mar Struct 1998;11:20330.
[5] Torseletti E, Marchesani F, Bruschi R, Vitali L. Buckle propagation and its arrest: buckle arrestor design
versus numerical analyses and experiments. OMAE2003-37220. In: Proceedings of the 22nd international
conference on offshore mechanics and arctic engineering, Mexico, 2003.
[6] IIW. Fatigue design of welded joints and components. In: Hobbacher A, editor. Recommendations of IIW
joint working group, XIII-1539-96/XV-845-96. Abington Publishing, The International Institute of Welding,
1996.
[7] DNV-RP-C203 Fatigue strength analysis of offshore steel structures, August 2005.
[8] Timoshenko SP, Woinowsky-Krieger S. Theory of plates and shells. 2nd ed. McGraw-Hill Book Company,
Inc.; 1959.
[9] Flugge W. Stresses in shells. 2nd ed. Berlin, Germany: Springer; 1973.
[10] Spence J, Tooth AS, editors. Pressure vessel design concepts and principles. 1st ed. London: E & FN Spon,
an inprint of Chapman & Hall 2-6 Boundary Row; 1994.
[11] Lotsberg I, Holth PA. Stress concentration factors at welds in tubular sections and pipelines. In: OMAE,
2007, OMAE paper no. 2007-29571.
[12] Radaj D, Sonsino CM, Fricke W. Fatigue assessment of welded joints by local approaches. 2nd ed.
Cambridge, England: Woodhead Publishing Limited; 2006.
[13] Lotsberg I, Sigurdsson G. Hot spot SN curve for fatigue analysis of plated structures. In: OMAE-FPSO04-
0014, international conference, Houston, 2004. Also in J Offshore Arctic Eng 2006;128:3306.
[14] PREFEM, Preprocessor for generation of nite element models, users manual. DNV Sesam Report No. 95-
7014/Rev.0, August 1, 1995.
[15] DNV-OS-F101 Submarine pipeline systems, January 2000.
[16] DNV Rules for Classication of Ships. Compressed natural gas carriers, Part 5, January 2005 [Chapter 15].
[17] Hibbit, Karlson & Sorensen Inc. ABAQUS/standard users manual, vol. 1, version 6.4, 2003.
[18] BS 7910:1999. Guidance on methods for assessing the acceptability of aws in metallic structures. BSI, 1999.
ISBN:0580330818.
ARTICLE IN PRESS
I. Lotsberg / Marine Structures 21 (2008) 138159 159

Potrebbero piacerti anche