k 1
k
(1)
For a thinwalled cylinder, k/1. The length and depth of an
axial surface crack in a cylinder are dened by 2c and a, respec
tively. The crack depth, a, is measured from the inner surface of the
cylinder for internal cracks and from the outer surface for external
cracks. The surface crack becomes an extended surface crack when
c/Nand a throughwall crack when a t.
The load types considered are internal pressure, p, throughwall
membrane stress, s
m
, and throughwall bending stress, s
b
. Their
limit values are denoted as p
L
, s
m
L
and s
b
L
, respectively. Solu
tions corresponding to other loading types, such as axial tension, N,
and axial global bending moment, M, are also reviewed.
The material considered is an elasticperfectly plastic type with
yield stress of s
y
. Therefore, all limit load solutions for closedend
cylinders subjected to internal pressure may be approximately
expressed as
p
L
gL
_
k;
a
t
;
a
c
; .
_
s
y
(2)
where L is a geometric function and g is the constraint factor, with
g
1 for Tresca yield condition
2
3
p for von Mises yield condition
_
_
_
(3)
for pressurised defectfree cylinders.
In this paper, the limit pressure is generally normalised by the
limit pressure of the crackfree cylinder based on the von Mises
yield criterion, p
0
, dened by
p
0
2
3
p s
y
ln k (4)
3. Review of the available solutions
In this section, limit load solutions for cylinders containing axial
surface/throughwall cracks under internal pressure, combined
Nomenclature
a Crack depth
A
df
Crack area
c Half crack length
c
1
, c
2
Equivalent lengths of crackfree cylinders for dening
local limit pressure
c
eq
Equivalent throughwall half crack length for surface
cracks
D Equivalent length of crackfree cylinder for dening
global limit pressure
f
pt
, f
ps
Crack face pressure factor for throughwall and
internal surface cracks, respectively
F
pt
Stress transfer factor
h
1
, h
2
Equivalent lengths of crackfree cylinders for dening
local limit pressure
k Ratio between the outer and inner radii of a cylinder,
k R
o
=R
i
L Function of geometry for limit load denitions
M Global axial bending moment
M
a1
wM
a3
Stress magnication factors due to the curvature of
cylinders (see denitions where they rst appear)
M
an
; M
axn
Stress magnication factors due to the curvature of
cylinders (see denitions where they rst appear)
M
ax1
wM
ax2
Stress magnication factors due to the curvature of
cylinders (see denitions where they rst appear)
M
Lp
Limit moment applied to the backwall of a cylinder
due to internal pressure
M
t
Folias factor
M
teq
Folias factor for equivalent throughwall cracks
M
t1
wM
t4
Stress magnication factors due to the curvature of
cylinders (see denitions where they rst appear)
M
tn
Stress magnication factor for throughwall cracks due
to the curvature of cylinders (see denition where it
rst appears)
N Axial tension force
N
L
Limit axial tension force
p Internal pressure
p
0
Limit pressure for a crackfree cylinder
p
I
Global limit pressure for a cylinder with an internal
surface crack under internal pressure
p
L
Limit pressure
p
Lf
Pressure corresponding to the frontwall failure of
a cylinder with a throughwall crack
Dp
L
Extra limit pressure due to the backwall effect in
cylinders
R
i
Inner radius of the cylinder
R
m
Average radius of the cylinder, R
m
R
o
R
i
=2
R
o
Outer radius of the cylinder
R
*
1
Equivalent radius considering the crack face pressure
for surface cracks dened by Carter
R
2
; R
*
2n
Equivalent radius considering the crack face pressure
for surface cracks (see denitions where they rst
appear)
R
t
; R
*
tn
Equivalent radius considering the crack face pressure
for throughwall cracks (see denitions where they
rst appear)
s
1
; s
2
Equivalent lengths of crackfree cylinders for dening
local limit pressure
t Cylinder thickness, t R
o
R
i
f Coefcient for dening the stress magnication factor
g 2=
3
p
for von Mises yield criterion and 1 for Tresca
yield criterion
l Load ratio
r
i
; r
m
; r
o
r factors dened by inner, mean and outer radii,
respectively
s
b
Throughwall bending stress
s
b
L
Limit throughwall bending stress
s
m
Throughwall membrane stress
s
m
L
Limit throughwall membrane stress
s
y
Yield stress.
Y. Lei / International Journal of Pressure Vessels and Piping 85 (2008) 825850 826
membrane stress and throughwall bending, and combined internal
pressure, tension and global bending are reviewed and discussed.
3.1. Solutions for extended axial surface cracks
under internal pressure
For an extended surface crack, the crack depth is a and the crack
length c/N(see Fig. 1). In this subsection, only internal pressure
is considered.
3.1.1. Internal defects
The geometry and dimensions of a cylinder with an internal
extended crack, a, under internal pressure are shown in Fig. 1(a).
Chell [8] proposed a limit pressure solution for an internal
extended crack in a thickwalled cylinder without crack face pres
sure, which can be expressed as
p
L
s
y
g
t a
R
i
a
g
k 1
_
1
a
t
_
1
a
t
k 1
(5)
where g takes the values dened in Eq. (3) for solutions based on
the von Mises and Tresca yield criteria. Note that Eq. (5) does not
reduce to the limit pressure for uncracked thickwalled cylinders
(see Eq. (4) for the case g 2=
3
p
) when a=t/0.
Carter [9] gave a limit pressure solution for thickwalled
cylinders, based on the Tresca yield criterion (g 1), in the
form
p
L
s
y
R
i
R
*
1
ln
_
R
o
R
i
a
_
t
R
*
1
k 1
ln
_
k
1
a
t
k 1
_
(6)
where
R
*
1
_
R
i
without crack face pressure
R
i
a with crack face pressure
(7)
Kim et al. [13] performed elasticperfectly plastic FE analyses,
using the von Mises yield criterion, for extended internal surface
cracks in cylinders under internal pressure and proposed a formula,
based on the FE results, which can be expressed as
p
L
s
y
3
p
t
R
m
_
1 0:356
a
t
1:6882
_
a
t
_
2
1:0442
_
a
t
_
3
_
3
p
2k 1
k 1
_
1 0:356
a
t
1:6882
_
a
t
_
2
1:0442
_
a
t
_
3
_ (8)
The cylinders considered were for t=R
m
0:2 and 0.05 with
50% of the internal pressure applied to the crack surface. Therefore,
Eq. (8) is valid for thinwalled cylinders with pressurised crack
faces.
Staat [10,11] applied the g factor to Carters solution (Eq. (6)) to
extend it to solutions based on both the von Mises and Tresca yield
criteria. He also applied a pressure magnifying factor, R
i
a=R
i
, to
the righthand side of Eq. (6) arguing that the pressure acts on the
inner surface of the cylinder rather than the surface at radius R
i
a
[10,11]. Staat and Vu [12] considered the resistance of the backwall
of the cracked part of the cylinder to the internal pressure and
further improved Staats solution [10,11]. The newsolution [12] is as
follows
p
L
s
y
g
R
i
R
*
2
R
i
a
R
i
ln
_
R
o
R
i
a
_
_
_
1
a
t
t
R
*
2
_
1
2
_
a
t
_
2
_
t
R
*
2
_
2
_
_
1
1
2
a
t
t
R
*
2
__
g
t
R
*
2
k 1
_
1
a
t
k 1
_
ln
_
k
1
a
t
k 1
_
_
1
a
t
t
R
*
2
_
1
2
_
a
t
_
2
_
t
R
*
2
_
2
_
_
1
1
2
a
t
t
R
*
2
__
9
where
R
*
2
R
i
without crack face pressure
R
i
a
2
with crack face pressure
_
(10)
The second termin the righthand side of Eq. (9) corresponds to
the extra pressure carried by the cylinder of inner radius R
i
and
thickness a with an extended penetrating crack. For a thinwalled
cylinder, this term is negligible. However, it can be signicant for
thickwalled cylinders. It is also seen from Eq. (10) that the crack
face pressure correction factor differs from Eq. (7) where the
internal pressure is assumed to be applied to a diameter 2R
i
a.
In contrast, in Eq. (10) the internal pressure is applied to 2R
i
a,
considering force equilibrium.
Internal crack
External crack
R
m
t
x
y
R
i
R
o
a
p
R
m
t
x
y
R
i
R
o
a
p
a
b
Fig. 1. Geometry and dimensions of axial extended surface cracks in thickwalled
cylinders under internal pressure.
Y. Lei / International Journal of Pressure Vessels and Piping 85 (2008) 825850 827
The solutions of Eqs. (5), (6), (8) and (9) (g 2=
3
p
for Eqs. (5)
and (9)) are compared with FE results based on the von Mises yield
criterion due to Staat and Vu [12]
,a
and Kim et al. [13] in Figs. 2
and 3 for various k. In the gures, the limit pressure is normalised
using p
0
dened by Eq. (4) and plotted against the normalised crack
depth, a=t.
Fig. 2 is for cases without crack face pressure. There is only one
set of available FE results, for k 2. From the gure, Chells solution
(Eq. (5)) is nonconservative for shallow cracks and Carters solu
tion (Eq. (6)) is conservative, partly due to its use of a factor g 1.
The solution due to Staat and Vu (Eq. (9)) provides the best
predictions and is conservative compared with the FE results.
Fig. 3 is for cases with crack face pressure. Fig. 3(a) and (b) are for
thinwalled cylinders with FE results from Kim et al. [13] and
Fig. 3(c) is for a thickwalled cylinder with FE results fromStaat and
Vu [12].
a
From Fig. 3, Carters solution (Eq. (6)) is conservative for
both thinwalled and thickwalled cylinders. The solution due to
Kim et al. (Eq. (8)) is very accurate for thinwalled cylinders
(Fig. 3(a) and (b)) because Eq. (8) was obtained by tting the FE data
shown in the two gures. It also gives good predictions for k 2
(Fig. 3(c)). The solution due to Staat and Vu (Eq. (9)) gives good
predictions for both thinwalled and thickwalled cases. However,
it may slightly overestimate the limit pressure for thinwalled
cylinders with very deep cracks.
3.1.2. External defects
The geometry and dimensions of a cylinder with an external
extended crack of depth, a, under internal pressure are shown in
Fig. 1(b).
Chell [8] proposed limit pressure solutions based on both the
von Mises and Tresca yield criteria for a cylinder with an extended
external crack under internal pressure, which can be expressed as
p
L
s
y
g
t a
R
i
gk 1
_
1
a
t
_
(11)
This equation does not reduce to the limit pressure for
uncracked thickwalled cylinders (see Eq. (4) for the case
g 2=
3
p
) when a=t/0.
Carters limit pressure solution [9] for thickwalled cylinders,
based on the Tresca yield condition (g 1), can be expressed as
p
L
s
y
ln
_
R
o
a
R
i
_
ln
_
k
a
t
k 1
_
(12)
The solutions due to Staat and Vu [12] based on both the von
Mises and Tresca yield criteria are as follows
p
L
s
y
gln
_
R
o
a
R
i
_
_
_
R
o
R
i
_
R
o
R
i
a
t
t
R
i
_
1
2
_
a
t
_
2
_
t
R
i
_
2
_
R
o
R
i
1
2
a
t
t
R
i
_
_
_
gln
_
k
a
t
k 1
_
_
k
_
k
a
t
k 1
_
1
2
_
a
t
_
2
k 1
2
_
_
k
1
2
a
t
k 1
_
_
13
FE, Staat & Vu
Prediction, Carter (eqn.(6)), (
Prediction, Kim et al. (eqn. (8))
Prediction, Staat & Vu (eqn. (9)), ( 3) = 2
3) = 2
k = 1.05
k = 1.22
k = 2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
p
L
/
p
0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
p
L
/
p
0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
p
L
/
p
0
FE, Kim et al.
Prediction, Carter (eqn.(6)), (
Prediction, Kim et al. (eqn.(8))
Prediction, Staat & Vu (eqn. (9)), ( 3) = 2
3) 2 =
0 0.2 0.4 0.6 0.8 1
a/t
0 0.2 0.4 0.6 0.8 1
a/t
0 0.2 0.4 0.6 0.8 1
a/t
FE, Kim et al.
Prediction, Carter (eqn. (6)), (
Prediction, Kim et al. (eqn. (8))
Prediction, Staat & Vu (eqn. (9)), ( 3) 2 =
3) = 2
a
b
c
Fig. 3. Comparison of normalised limit pressures between various solutions and FE
results due to Kim et al. [13] and Staat and Vu [12] for extended internal cracks under
internal pressure (with crack face pressure).
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
0 0.2 0.4 0.6 0.8 1
a/t
p
L
/
p
0
FE, Staat & Vu
Prediction, Chell (eqn. (5)),
Prediction, Carter (eqn. (6))
Prediction, Staat & Vu (eqn. (9)), ( 3) = 2
( 3)
= 2
Fig. 2. Comparison of normalised limit pressures between various solutions and FE
results due to Staat and Vu [12] for extended internal cracks under internal pressure.
a
All the FE data presented in 3D plots in Ref. [12] were provided by Prof.
Manfred Staat.
Y. Lei / International Journal of Pressure Vessels and Piping 85 (2008) 825850 828
The three solutions of Eq. (12) and Eqs. (11) and (13) with
g 2=
3
p
are compared in Fig. 4 with FE results based on the von
Mises yield criterion for k 2 obtained by Staat and Vu [12]. In the
gure, the normalised limit pressure is plotted against the nor
malised crack depth, a=t. From Fig. 4, Chells solution (Eq. (11)) is
nonconservative for a=t <0:6. Carters solution (Eq. (12)) is
conservative for all crack depths. The solution due to Staat and Vu
gives the best predictions and is conservative for this case.
3.2. Solutions for throughwall cracks under internal pressure
The geometry and dimensions of a cylinder with a throughwall
crack, a t, of length, 2c, under internal pressure are showninFig. 5.
For cylinders with throughwall cracks under internal pressure,
the solution due to Kiefner et al. [3] must be mentioned. Based on
pipe burst data, Kiefner et al. [3] found that the limit pressure of
a pipe with a throughwall crack can be approximately expressed as
p
L
s
y
t
R
m
M
t
2
M
t
k 1
k 1
(14)
where M
t
is the Folias factor [17] and can be approximately
expressed as (see Appendix A)
M
t
_
1 1:255 r
2
m
0:0135 r
4
m
_
0:5
(15)
where
r
m
c
R
m
t
p
2
k 1
k 1
_
t
c
(16)
Note that Eq. (14) is valid for thinwalled cylinders because
it was based on thinwalled pipe burst test data and the Folias
factor was dened for thinwalled cylinders [17]. Also, the Folias
factor was dened for r
m
< 4:4 [17] and there is no denition of M
t
beyond this limit. In addition, Eq. (14) cannot be used for the cases
with crack face pressure because the pipes used in the burst tests
carried out by Kiefner et al. were all sealed from the inside [3].
Another early solutionwas proposed by Erdogan [18], which can
be expressed as
p
L
s
y
t
R
m
M
t1
2
M
t1
k 1
k 1
(17)
where
M
t1
0:614 0:87542 r
m
0:386 exp 2:275r
m
(18)
Eq. (17) does not reduce to the solution for an uncracked thick
walled cylinder when the crack length becomes very small and,
therefore, can only be used for thinwalled cylinders.
Carter [9] also proposed a limit pressure solution similar to that
due to Kiefner et al. [3], but he used an older Folias factor [19], i.e.
p
L
s
y
t
R
i
M
t2
k 1
M
t2
(19)
where M
t2
is the Folias factor dened by
M
t2
_
1 1:61 r
2
i
_
0:5
(20)
Note that Eq. (20) is slightly different from that in Ref. [19]
where r
m
was used. However, in Eq. (20), r
i
is used instead, dened
by
r
i
c
R
i
t
_
k 1
p
t
c
(21)
Carter dened the M
t2
factor using the internal radius of the
cylinder, R
i
, rather than the mean radius, R
m
, to lead to a conser
vative prediction of the limit pressure. The use of Eq. (20) is further
discussed in Appendix A. Similar to Eqs. (14) and (17), Eq. (19) also
does not reduce to the solution for an uncracked thickwalled
cylinder when the crack length tends to zero and, therefore, is
unsuitable for use with thickwalled cylinders.
Kim et al. [13] proposed a limit pressure solution as follows,
based on their elasticperfectly plastic FE analyses, using the von
Mises yield criterion,
p
L
s
y
3
p
t
R
m
M
t3
3
p
2
M
t3
k 1
k 1
(22)
where M
t3
was obtained by tting the FE results for t=R
m
0:2 and
0.05 and can be expressed as
M
t3
_
1 0:34 r
m
1:34 r
2
m
_
0:5
(23)
Note that, intheir FEanalyses, Kimet al. applied50%of theinternal
pressure to the crack face. Therefore, Eqs. (22) and (23) are appro
priate for cases with crack face pressure. Eq. (22) is also restricted to
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
0 0.2 0.4 0.6 0.8
a/t
p
L
/
p
0
FE, Staat & Vu
Prediction, Chell (eqn. (11)), (
Prediction, Carter, (eqn. (12))
Prediction, Staat & Vu (eqn. (13)), ( 3) = 2
3) = 2
1
Fig. 4. Comparisonof normalisedlimit pressures betweenvarious solutions andFEresults
due to Staat and Vu [12] for extended external cracks under internal pressure (k 2).
2c
t
R
i
R
o
p
R
m
Fig. 5. Geometry and dimensions of an axial throughwall crack in a thickwalled
cylinder under internal pressure.
Y. Lei / International Journal of Pressure Vessels and Piping 85 (2008) 825850 829
thinwalledcylinders because of the thinwalledstyle of the equation
itself and the thinwalled FE data used in the calibration of M
t3
.
Staat and Vu [12] also proposed limit pressure
solutions based on both the von Mises and Tresca yield criteria
for thickwalled cylinders by considering the backwall effect,
that is
p
L
s
y
g
M
t4
ln
_
R
o
R
i
_
_
_
1
t
R
i
1
2
_
t
R
i
_
2
1
2
_
1
R
o
R
i
_
_
_
g
M
t4
ln k
_
1 k 1
1
2
k 1
2
_
1
2
1 k
_ (24)
The rst part in the righthand side of Eq. (24) is similar to Eq.
(17), but expressed in the thickwalled form. The second part
corresponds to the backwall correction, which has been found to
be signicant for thickwalled cylinders [12]. In Eq. (24), M
t4
is
dened by
M
t4
_
1 1:25 r
2
o
_
0:5
(25)
where
r
o
c
R
o
t
p
k 1
k
_
t
c
(26)
Eq. (25) was calibrated using FE data for k 1.1, 1.25, 1.5, 1.75 and
2 [12], and, therefore, can be used for both thinwalled and thick
walled cylinders. Examining Eq. (24), it can be seen that the second
termin the righthand side of the equation is a constant for given k.
This means that the limit pressure may be greater than that for the
uncracked cylinder when the crack length is very small and hence
M
t4
/1.
The ve solutions of Eqs. (14), (17), (19), (22) and (24) with g
2=
3
p
are compared in Figs. 6 and 7 with available FE results based
on the von Mises criterion due to Staat and Vu [12] (digitised from
Fig. 3 in Ref. [12]) and Kim et al. [13] for various k values. In the
gures, the normalised limit pressure is plotted against r
2
o
, where
r
o
is dened by Eq. (26).
The predictions for cases without crack face pressure are
compared with FE results based on the von Mises criterion due to
Staat and Vu [12] in Fig. 6. From the gure, Eq. (24) gives the best
predictions for all the ve k values but is slightly nonconservative
compared with the FE results. This is because the FE data presented
in Fig. 6 were used to calibrate Eq. (25). It is also seen from the
gure that Eq. (24) overestimates the limit pressure when r
o
/0,
especially for large k (Fig. 6(e)). The other two solutions, that is, Eq.
(14) due to Kiefner et al. and Eq. (17) due to Erdogan are conser
vative for all the ve k values considered. This is probably because
they are all based on the Tresca criterion. From Fig. 6(a) and (b),
they can all predict the FE results very well if a factor of 2=
3
p
is
applied. But, for thickwalled cylinders (Fig. 6(e)), they may
signicantly underestimate the limit pressure even if the factor
2=
3
p
is applied. The solution due to Carter, Eq. (19), is conservative
for k 1:25 (Fig. 6(a) and (b)). However, for thick cylinders
(Fig. 6(c)(e)), it overestimates the limit pressure for small r
o
,
although it is still conservative for large r
o
due to using the Tresca
criterion.
Fig. 7 compares predictions with FE results for cases with crack
face pressure due to Kim et al. [13] (Fig. 7(a)) and to Staat and Vu
[12] (Fig. 7(b)). Among the ve solutions, only Eq. (22) due to Kim
et al. can predict the limit pressure when the crack face pressure is
considered. From Fig. 7, Eq. (22) can predict the FE results for
k 1.22 very accurately (the FE data in Fig. 7(a) were used to
calibrate Eq. (23)), but it overestimates the FE results for k 2. This
indicates that Eq. (22) is nonconservative for thickwalled
cylinders.
3.3. Global solutions for axial surface cracks under internal pressure
The geometry and dimensions of a cylinder with an internal/
external surface crack of depth, a t, length 2c under internal
pressure are shown in Fig. 8.
Currently, the limit pressure solutions have been obtained by
simplifying the surface cracked cylinder to two coaxial cylinders,
Cylinder A with a throughwall crack and Cylinder B without
a crack (Fig. 9). The limit pressure is taken as the sum of the
limit pressures for the two cylinders. For the case of an internal
crack (Fig. 9(a)), for example, the inner cylinder, Cylinder A, with
an internal radius R
i
and thickness a contains a throughwall
crack of length 2c and the outer cylinder, Cylinder B, is crack
free with an internal radius R
i
a and thickness t a. Various
limit pressure solutions have been obtained because different
limit pressure solutions for throughwall and crackfree cylinders
were adopted.
Ewing [6] gave a limit pressure solution, based on the thin
walled limit pressure solutions for both cracked and crackfree
cylinders, which can be expressed as
p
L
s
y
t
R
m
_
1
a
t
a
t
1
M
a1
_
2k 1
k 1
_
1
a
t
a
t
1
M
a1
_
(27)
where
M
a1
_
1 1:61
c
2
R
m
a
_0:5
_
1 1:61
2
a
t
k 1
_
a
c
_
2
k 1
_
0:5
(28)
Eq. (27) can be used for cases of internal cracks without crack
face pressure and for external cracks.
3.3.1. Internal cracks
The geometry and dimensions of a cylinder with an internal
surface crack, a t, under internal pressure are shown in Fig. 8(a)
and the simplied model is shown in Fig. 9(a).
Carter [9] used his limit pressure solution for a cylinder with
a throughwall crack (Eq. (19)), which is for thinwalled cylinders,
and the limit pressure solution for a thickwalled uncracked
cylinder and obtained the following equation
p
L
s
y
a
R
i
1
M
a2
R
i
R
*
1
ln
_
R
o
R
i
a
_
a
t
k 1
M
a2
R
i
R
*
1
ln
_
k
1
a
t
k 1
_
(29)
where
M
a2
_
1 1:61
c
2
R
i
a
_0:5
_
1 1:61
a
t
k 1
_
a
c
_
2
_
0:5
(30)
The mismatch of thinwalled and thickwalled solutions for,
respectively, the cylinder with a throughwall crack and the crack
free one may cause problems when Eq. (29) is used for thickwalled
cylinders. It is also seen from Eq. (29) that the crack face pressure
term, R
i
=R
*
1
, is applied only to the crackfree cylinder. The correction
for the crack face pressure vanishes when a=t/1.
Staat and Vu [12] constructed limit pressure solutions based on
both the von Mises and Tresca yield criteria using their solution for
cylinders with throughwall cracks (Eq. (24)) and the limit pressure
expression for crackfree thickwalled cylinders, which can be
expressed as
Y. Lei / International Journal of Pressure Vessels and Piping 85 (2008) 825850 830
k = 1.1 k = 1.25
k = 1.5
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0 1 2 3 4 5
p
L
/
p
0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
p
L
/
p
0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
p
L
/
p
0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
p
L
/
p
0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
p
L
/
p
0
FE, Staat & Vu
Prediction, Kiefner et al. (eqn. (14))
Prediction, Erdogan (eqn. (17))
Prediction, Carter (eqn. (19))
Prediction, Staat & Vu (eqn. (24)), (
Prediction, present work (eqn. (56)), (
2
o
2
o
2
o
0 2 4 6 8 10
2
o
3) = 2
3) = 2
FE, Staat & Vu
Prediction, Kiefner et al. (eqn. (14))
Prediction, Erdogan (eqn. (17))
Prediction, Carter (eqn. (19))
Prediction, Staat & Vu (eqn. (24)), (
Prediction, present work (eqn. (56)), ( 3) = 2
3) = 2
0 10 15
FE, Staat & Vu
Prediction, Kiefner et al. (eqn. (14))
Prediction, Erdogan (eqn. (17))
Prediction, Carter (eqn. (19))
Prediction, Staat & Vu (eqn. (24)), (
Prediction, present work (eqn. (56)), ( 3) = 2
3) = 2
k = 1.75
k = 2
FE, Staat & Vu
Prediction, Kiefner et al. (eqn. (14))
Prediction, Erdogan (eqn. (17))
Prediction, Carter (eqn. (19))
Prediction, Staat & Vu (eqn. (24)), (
Prediction, present work (eqn. (56)), ( 3) = 2
3) = 2
FE, Staat & Vu
Prediction, Kiefner et al. (eqn. (14))
Prediction, Erdogan (eqn. (17))
Prediction, Carter (eqn. (19))
Prediction, Staat & Vu (eqn. (24)), (
Prediction, present work (eqn. (56)), ( 3) 2 =
3) 2 =
5 0 10 15 20 5
2
o
0 10 15 20 25 5
a b
c
e
d
Fig. 6. Comparison of normalised limit pressures between various solutions and FE results due to Staat and Vu [12] for cylinders with throughwall cracks under internal pressure
(without crack face pressure).
p
L
s
y
min
_
_
gln
_
R
o
R
i
_
; g
R
i
R
*
2
_
1
M
a3
ln
_
R
i
a
R
i
_
R
i
a
R
i
ln
_
R
o
R
i
a
__
_
1
a
t
t
R
*
2
1
2
_
a
t
_
2
_
t
R
*
2
_
2
_
_
1
1
2
a
t
t
R
*
2
__
_
_
min
_
_
gln k; g
R
i
R
*
2
_
1
M
a3
ln
_
1
a
t
k 1
_
_
1
a
t
k 1
_
ln
_
k
1
a
t
k 1
__
1
a
t
t
R
*
2
1
2
_
a
t
_
2
_
t
R
*
2
_
2
_
_
1
1
2
a
t
t
R
*
2
__
_
_
(31)
Y. Lei / International Journal of Pressure Vessels and Piping 85 (2008) 825850 831
where
M
a3
_
1 1:25
c
2
R
i
aa
_0:5
_
1 1:25
a
t
k 1
_
a
c
_
2
1
a
t
k 1
_
0:5
(32)
Eq. (31) may be nonconservative for short cracks for the
following two reasons. Firstly, the solution for cylinders with
throughwall cracks (Eq. (24)) is nonconservative for short cracks,
as discussed in Section 3.2 above. Secondly, a pressure magnica
tion factor, R
i
a=R
i
, is applied to the crackfree cylinder solution
assuming that the pressure applied on the outer cylinder is lower
than that applied on the inner one. However, for short cracks,
M
a3
/1 and the second formula in Eq. (31) will be greater than
gln k. Although a limit gln k is set in Eq. (31), it can still
potentially overestimate the limit pressure for short cracks.
Kim et al. [13] proposed a limit pressure solution, based on their
elasticperfectly plastic FE analyses with the von Mises yield
criterion, and expressed it as
p
L
s
y
3
p
t
R
m
_
1 A
1
a
t
A
2
_
a
t
_
2
_
3
p
2k 1
k 1
_
1 A
1
a
t
A
2
_
a
t
_
2
_
(33)
where
_
A
1
0:0462 0:0589 r
m
0:013 r
2
m
A
2
0:0395 0:3413 r
m
0:0652 r
2
m
(34)
and r
m
is dened by Eq. (16).
Eqs. (33) and (34) are based on FE data for
t=R
m
0:2; 0:1; 0:05and0:025, with 50% internal pressure applied
on the crack faces, and, therefore, are valid for thinwalled cylinders
with crack face pressure. Note that Eq. (33) is inconsistent with Eq.
(22) when a=t 1.
Fig. 10 compares the normalised limit pressures predicted
using Eqs. (27), (29) and (31) (g 2=
3
p
) with FE results based
on the von Mises yield criterion due to Staat and Vu [12] for cases
of k 2 without crack face pressure. From the gure, Eqs. (27)
and (29) due to Ewing and Carter, respectively, are conservative
for 0 a=t 1 and 0:2 a=c 1 probably because they are
based on the Tresca yield criterion. The predictions using Eq. (31)
due to Staat and Vu are very close to the FE results but are non
conservative for short and shallow cracks and throughwall
cracks.
Figs. 1113 compare the normalised limit pressures predicted
using Eqs. (29), (31) (g 2=
3
p
) and Eq. (33) with FE results based
on the von Mises yield criterion due to Staat and Vu [12] (Fig. 11 for
k 2), and due to Kim et al. [13] (Fig. 12 for k 1.05 and Fig. 13 for
k 1.22) for cases with crack face pressure. From Fig. 11 for k 2,
predictions using Eq. (33) due to Kim et al. are very close to the FE
results for shallow cracks but are nonconservative for deep cracks.
Carters solution (Eq. (29)) is conservative for long and shallow
cracks but overestimates the FE results for short and deep cracks. It
is also seen from the gure that the predictions using Eq. (31) due
to Staat and Vu are reasonably close to the FE results but are slightly
k = 1.22
k = 2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0 2 8 10
p
L
/
p
0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
p
L
/
p
0
FE, Kim et al .
Prediction, Kim et al. (eqn. (22))
Prediction, present work (eqn. (56)), ( 3) = 2
0 2 8 10 12 14 16
FE, Staat & Vu
Prediction, Kim et al. (eqn. (22))
Prediction, present work (eqn. (56)), ( 3) = 2
4 6
4 6
a
b
Fig. 7. Comparison of normalised limit pressures between various solutions and FE
results due to Staat and Vu [12] and Kim et al. [13] for cylinders with throughwall
cracks under internal pressure (with crack face pressure).
Internal surface crack
External surface crack
2c
a
t
R
i
R
o
R
m
p
2c
a
t
R
i
R
o
R
m
p
a
b
Fig. 8. Geometry and dimensions of axial surface cracks in cylinders under internal
pressure.
Y. Lei / International Journal of Pressure Vessels and Piping 85 (2008) 825850 832
nonconservative for shallowcracks and some throughwall cracks.
For thinwalled cylinders (Figs. 12 and 13), the solution due to Kim
et al. (Eq. (33)) gives accurate predictions of the FE results up to
a=t 0:8 but signicantly overestimates the limit pressure for
throughwall cracks. This is not surprising because Eq. (33) was
tted to the FE data presented in Figs. 12 and 13. From the gures,
Carters solution is conservative for all crack lengths and depths
considered by comparison with the FE data. The solution due to
Staat and Vu is reasonably close to the FE results and conservative,
but it slightly overestimates the limit pressures for throughwall
cracks.
3.3.2. External cracks
The geometry and dimensions of a cylinder with an external
surface crack, a t, under internal pressure are shown in Fig. 8(b)
and the simplied model is shown in Fig. 9(b).
Carters solution [9] for a cylinder with an external crack can be
expressed as
p
L
s
y
a
R
o
a
1
M
ax1
ln
_
R
o
a
R
i
_
a
t
k 1
k
a
t
k 1
1
M
ax1
ln
_
k
a
t
k 1
_
(35)
where
M
ax1
_
1 1:61
c
2
R
o
aa
_0:5
_
1 1:61
a
t
k 1
_
a
c
_
2
k
a
t
k 1
_
0:5
(36)
Eq. (35) was constructed using the limit pressure solution for
a cylinder with a throughwall crack (Eq. (19)), which is for thin
walled cylinders, and the limit pressure solution for crackfree
thickwalled cylinders. This mismatch may also cause problems
when Eq. (35) is used for thickwalled cylinders.
The limit pressure solutions based on both the von Mises and
Tresca yield criteria for external surface cracks due to Staat and Vu
[12] can be expressed as follows
p
L
s
y
g
_
1
M
ax2
ln
_
R
o
R
o
a
_
ln
_
R
o
a
R
i
__
_
R
o
R
i
_
R
o
R
i
a
t
t
R
i
_
1
2
_
a
t
_
2
_
t
R
i
_
2
_
R
o
R
i
1
2
a
t
t
R
i
_
_
_
g
_
1
M
ax2
ln
_
k
k
a
t
k 1
_
ln
_
k
a
t
k 1
_
_
_
k
_
k
a
t
k 1
_
1
2
_
a
t
_
2
k 1
2
_
_
k
1
2
a
t
k 1
_
_
(37)
where
M
ax2
_
1 1:25
c
2
R
o
a
_0:5
_
1 1:25
k 1
a
t
k
_
a
c
_
2
_
0:5
(38)
Eq. (37) may overestimate the limit pressure for short cracks
because the second term in the righthand side of Eq. (37) does not
depend on crack length.
Figs. 14 and 15 compare normalised limit pressures predicted
using Eqs. (27), (35) and (37) (g 2=
3
p
) with FE results
based on the von Mises yield criterion due to Staat and Vu
[12] for cases of k 2 and due to Zarrabi et al. [16] for k 1.57.
From the gures, Eq. (27) due to Ewing is conservative for all
crack lengths and depths considered. Carters solution (Eq. (35))
is also conservative, except for very short and deep cracks (see
Fig. 15(e)). It is also seen from Figs. 14 and 15 that the predictions
using Eq. (37) due to Staat and Vu are very close to the FE results
but slightly nonconservative for very short and throughwall
cracks.
3.4. Local solutions for axial surface defects under internal pressure
The limit pressure expression for a cylinder with a surface
crack under internal pressure given by Kiefner et al. [3] may be
expressed as
p
L
s
y
t
R
m
1
a
t
1
a
t
1
M
teq
2
k 1
k 1
1
a
t
1
a
t
1
M
teq
(39)
where the factor M
teq
should be evaluated using Eqs. (15) and (16).
The half crack length, c, in Eq. (16) should be replaced by the
equivalent half crack length, c
eq
, dened by
c
eq
A
df
2a
(40)
Internal crack
External crack
2c
a
t
R
i
R
o
Cylinder A
Cylinder B
p
2c
a t
R
i
R
o
Cylinder A
Cylinder B
p
a
b
Fig. 9. Mechanics models for determining the global limit pressures for cylinders with
surface cracks.
Y. Lei / International Journal of Pressure Vessels and Piping 85 (2008) 825850 833
where A
df
is the crack area and A
df
2ac for rectangular cracks.
Eq. (39) is an empirical formula obtained from burst experiments
on thinwalled pipes with internal or external defects [3]. It is,
therefore, a solution for thinwalled cylinders with internal/
external cracks. Note that the defective pipes used in the experi
ments were sealed from the inside of the pipes for the case of
internal defects. Hence, Eq. (39) applies to cases without crack face
pressure.
3.4.1. Internal cracks
Carter [9] dened the local limit pressure for a cylinder with
an internal surface crack under internal pressure as follows.
Firstly, the global limit pressure for a cylinder with an internal
surface crack under internal pressure (Eq. (29)) is alternatively
expressed as the average of the limit pressures of two crack
free cylinders of length D and a cylinder of length 2c with
an extended internal surface crack of depth a (see Fig. 16(a)),
that is
p
L
s
y
1
D c
_
D
p
L
for crackfree cylinder
s
y
c
p
L
for cylinder with extended crack
s
y
_
(41)
where D is an equivalent length of the crackfree cylinder,
which can be determined by equating Eq. (41) to Eq. (29). The
local limit pressure is then dened in a similar way to Eq. (41)
with a reduced equivalent length of the crackfree cylinder,
c
1
D, as
a/c = 0.2 a/c = 0.4
a/c = 0.6
FE, Staat & Vu
Prediction, Ewing (eqn. 27))
Prediction, Carter (eqn. (29))
Prediction, Staat & Vu (eqn. (31)), (
Prediction, present work (eqn.(62)), ( 3) 2 =
3) 2 =
FE, Staat & Vu
Prediction, Ewing (eqn. (27))
Prediction, Carter (eqn. (29))
Prediction, Staat & Vu (eqn. (31)), (
Prediction, present work (eqn. (62)), ( 3) = 2
3) = 2
FE, Staat & Vu
Prediction, Ewing (eqn. (27))
Prediction, Carter (eqn. (29))
Prediction, Staat & Vu (eqn. (31)), (
Prediction, present work (eqn. (62)), ( 3) 2 =
3) 2 =
a/c = 0.8
a/c = 1
FE, Staat & Vu
Prediction, Ewing (eqn. (27))
Prediction, Carter (eqn. (29))
Prediction, Staat & Vu (eqn. (31)), (
Prediction, present work (eqn. (62)), ( 3) = 2
3) = 2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0 0.2 0.4 0.6 0.8
a/t
p
L
/
p
0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
p
L
/
p
0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
p
L
/
p
0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
p
L
/
p
0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
p
L
/
p
0
FE, Staat & Vu
Prediction, Ewing (eqn. (27))
Prediction, Carter (eqn. (29))
Prediction, Staat & Vu (eqn. (31)), (
Prediction, present work (eqn. (62)), ( 3) 2 =
3) = 2
1
0 0.2 0.4 0.6 0.8
a/t
1 0 0.2 0.4 0.6 0.8
a/t
1
0 0.2 0.4 0.6 0.8
a/t
1 0 0.2
a b
c d
e
0.4 0.6 0.8
a/t
1
Fig. 10. Comparison of normalised limit pressures between various solutions and FE results due to Staat and Vu [12] for cylinders with internal surface cracks under internal
pressure (k 2, without crack face pressure).
Y. Lei / International Journal of Pressure Vessels and Piping 85 (2008) 825850 834
FE, Staat & Vu
Prediction, Carter (eqn. (29))
Prediction, Kim et al. (eqn. (33))
Prediction, Staat & Vu (eqn. (31)), (
Prediction, present work (eqn. (62)), ( 3) = 2
3) = 2
FE, Staat & Vu
Prediction, Carter (eqn. (29))
Prediction, Kim et al. (eqn. (33))
Prediction, Staat & Vu (eqn. (31)), (
Prediction, present work (eqn. (62)), ( 3) = 2
3) = 2
FE, Staat & Vu
Prediction, Carter (eqn. (29))
Prediction, Kim et al. (eqn. (33))
Prediction, Staat & Vu (eqn. 31)), (
Prediction, present work (eqn. (62)), ( 3) = 2
3) = 2
FE, Staat & Vu
Prediction, Carter (eqn. (29))
Prediction, Kim et al. (eqn. (33))
Prediction, Staat & Vu (eqn. (31)), (
Prediction,present work (eqn. (62)), ( 3) = 2
3) = 2
FE, Staat & Vu
Prediction, Carter (eqn. (29))
Prediction, Kim et al. (eqn. (33))
Prediction, Staat & Vu (eqn. (31)), (
Prediction, present work (eqn. (62)), ( 3) = 2
3) = 2
a/c = 0.2 a/c = 0.4
0.0
0.2
0.4
0.6
0.8
1.0
1.2
a b
c d
e
p
L
/
p
0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
p
L
/
p
0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
p
L
/
p
0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
p
L
/
p
0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
p
L
/
p
0
0 0.2 0.4 0.6 0.8
a/t
1 0 0.2 0.4 0.6 0.8
a/t
1
a/c = 0.6 a/c = 0.8
0 0.2 0.4 0.6 0.8
a/t
1
a/c = 1
0 0.2 0.4 0.6 0.8
a/t
1
0 0.2 0.4 0.6 0.8
a/t
1
Fig. 11. Comparison of normalised limit pressures between various solutions and FE results due to Staat and Vu [12] for cylinders with internal surface cracks under internal
pressure (k 2, with crack face pressure).
Y. Lei / International Journal of Pressure Vessels and Piping 85 (2008) 825850 835
t/c = 0.894 t/c = 0.447
t/c = 0.224
FE, Kim et al.
a b
c d
Prediction, Carter (eqn. (29))
Prediction, Kim et al. (eqn. (33))
Prediction, Staat & Vu (eqn. (31)), (
Prediction, present work (eqn. (62)), ( 3) 2 =
3) 2 =
FE, Kim et al .
Prediction, Carter (eqn. (29))
Prediction, Kim et al. (eqn. (33))
Prediction, Staat & Vu (eqn (31)), (
Prediction, present work(eqn. (62)), ( ) 3 2 =
) 3 2 =
FE, Kim et al.
Prediction, Carter (eqn. (29))
Prediction, Kim et al. (eqn. (33))
Prediction, Staat & Vu (eqn. (31)), (
Prediction, present work (eqn. (62)), (
3) 2 =
3) 2 =
t/c = 0.149
FE, Kim et al .
Prediction, Carter (eqn. (29))
Prediction, Kim et al. (eqn. (33))
Prediction, Staat & Vu (eqn. (31)), (
Prediction, present work (eqn. (62)), (
3) 2 =
3) 2 =
0.0
0.2
0.4
0.6
0.8
1.0
1.2
p
L
/
p
0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
p
L
/
p
0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
p
L
/
p
0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
p
L
/
p
0
0 0.2 0.4 0.6 0.8
a/t
1 0 0.2 0.4 0.6 0.8
a/t
1
0 0.2 0.4 0.6 0.8
a/t
1
0 0.2 0.4 0.6 0.8
a/t
1
Fig. 13. Comparison of normalised limit pressures between various solutions and FE results due to Kim et al. [13] for cylinders with internal surface cracks under internal pressure
(k 1.22, with crack face pressure).
t/c = 0.447 t/c = 0.224
t/c = 0.112
FE, Kim et al .
Prediction, Carter (eqn. (29))
Prediction, Kim et al. (eqn. (33))
Prediction, Staat & Vu (eqn. (31)), (
Prediction, present work(eqn. (62)), ( 3) = 2
3) = 2
FE, Kim et al.
Prediction, Carter (eqn. (29))
Prediction, Kim et al. (eqn. (33))
Prediction, Staat & Vu (eqn (31)), (
Prediction, present work (eqn. (62)), (
3) = 2
3) = 2
FE, Kim et al.
Prediction, Carter (eqn. (29))
Prediction, Kim et al. (eqn. (33))
Prediction, Staat & Vu (eqn. (31)), (
Prediction, present work (eqn. (62)), ( 3) 2 =
3) 2 =
t/c = 0.075
FE, Kim et al.
Prediction, Carter (eqn. (29))
Prediction, Kim et al. (eqn. (33))
Prediction, Staat & Vu (eqn. (31)), (
Prediction, present work (eqn. (62)), ( 3) 2 =
3) 2 =
0.0
0.2
0.4
0.6
0.8
1.0
1.2
a b
c d
p
L
/
p
0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
p
L
/
p
0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
p
L
/
p
0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
p
L
/
p
0
0 0.2 0.4 0.6 0.8
a/t
1
0 0.2 0.4 0.6 0.8
a/t
1 0 0.2 0.4 0.6 0.8
a/t
1
0 0.2 0.4 0.6 0.8
a/t
1
Fig. 12. Comparison of normalised limit pressures between various solutions and FE results due to Kim et al. [13] for cylinders with internal surface cracks under internal pressure
(k 1.05, with crack face pressure).
Y. Lei / International Journal of Pressure Vessels and Piping 85 (2008) 825850 836
a/c = 0.2 a/c = 0.4
a/c = 0.6
FE, Staat & Vu
Prediction, Ewing (eqn. (27))
Prediction, Carter (eqn. (35))
Prediction, Staat & Vu (eqn. (37)),
Prediction, present work (eqn. (65)), ( ) 3 2 =
( ) 3 2 =
FE, Staat & Vu
Prediction, Ewing (eqn. (27))
Prediction, Carter (eqn. (35))
Prediction, Staat & Vu (eqn. (37)),
Prediction, present work (eqn. (65)), ( ) 3 2 =
( ) 3 2 =
FE, Staat & Vu
Prediction, Ewing (eqn. (27))
Prediction, Carter (eqn. (35))
Prediction, Staat & Vu (eqn. (37)),
Prediction, present work (eqn. (65)), ( ) 3 2 =
( ) 3 2 =
FE, Staat & Vu
Prediction, Ewing (eqn. (27))
Prediction, Carter (eqn. (35))
Prediction, Staat & Vu (eqn. (37)),
Prediction, present work (eqn. (65)), ( ) 3 2 =
( ) 3 2 =
FE, Staat & Vu
Prediction, Ewing (eqn. (27))
Prediction, Carter (eqn. (35))
Prediction, Staat & Vu (eqn. (37)),
Prediction, present work (eqn. (65)), ( ) 3 2 =
( ) 3 2 =
a/c = 0.8
a/c = 1
0 0.2 0.4 0.6 0.8
a/t
1
0 0.2 0.4 0.6 0.8
a/t
1 0 0.2 0.4 0.6 0.8
a/t
1
0 0.2 0.4 0.6 0.8
a/t
1 0 0.2 0.4 0.6 0.8
a/t
1
0.0
0.2
0.4
0.6
0.8
1.0
1.2
p
L
/
p
0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
p
L
/
p
0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
p
L
/
p
0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
p
L
/
p
0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
p
L
/
p
0
a b
c d
e
Fig. 14. Comparison of normalised limit pressures between various solutions and FE results due to Staat and Vu [12] for cylinders with external surface cracks under internal
pressure (k 2).
Y. Lei / International Journal of Pressure Vessels and Piping 85 (2008) 825850 837
FE, Zarrabi et al.
Prediction, Ewing (eqn. (27))
Prediction, Carter (eqn. (35))
Prediction, Staat & Vu (eqn. (37)), (
Prediction, present work (eqn. (65)), ( ) 3 2 =
) 3 2 =
a/t = 0.9
0.0
0.2
0.4
0.6
0.8
1.0
1.2
e
d c
b a
0 2 3 6
a/c
p
L
/
p
0
4 5 1
a/t = 0.7
0 2 3 6
a/c
4 5 1
a/t = 0.5
0 2 3
a/c
4 1
a/t = 0.3
0 1 3
a/c
2
a/t = 0.1
0 2 3
a/c
1
0.0
0.2
0.4
0.6
0.8
1.0
1.2
p
L
/
p
0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
p
L
/
p
0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
p
L
/
p
0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
p
L
/
p
0
FE, Zarrabi et al.
Prediction, Ewing (eqn. (27))
Prediction, Carter (eqn. (35))
Prediction, Staat & Vu (eqn. (37)), (
Prediction, present work (eqn. (65)), ( ) 3 2 =
) 3 2 =
FE, Zarrabi et al.
Prediction, Ewing (eqn. (27))
Prediction, Carter (eqn. (35))
Prediction, Staat & Vu (eqn. (37)), (
Prediction, present work (eqn. (65)), ( ) 3 2 =
) 3 2 =
FE, Zarrabi et al.
Prediction, Ewing (eqn. (27))
Prediction, Carter (eqn. (35))
Prediction, Staat & Vu (eqn. (37)), (
Prediction, present work (eqn. (65)), ( ) 3 2 =
) 3 2 =
FE, Zarrabi et al.
Prediction, Ewing (eqn. (27))
Prediction, Carter (eqn. (35))
Prediction, present work (eqn. (65)), ( ) 3 2 =
Prediction, Staat & Vu (eqn. (37)), ( ) 3 2 =
Fig. 15. Comparison of normalised limit pressures between various solutions and FE results due to Zarrabi et al. [16] for cylinders with external surface cracks under internal
pressure (k 1.57).
Y. Lei / International Journal of Pressure Vessels and Piping 85 (2008) 825850 838
p
L
s
y
1
c
1
c
_
c
1
ln
_
R
o
R
i
_
c
R
i
R
*
1
ln
_
R
o
R
i
a
_
_
1
c
1
c
1
_
c
1
c
ln k
R
i
R
*
1
ln
_
k
1
a
t
k 1
_
_
(42)
where
c
1
c
a
_
1
a
t
_
M
a2
R
i
_
ln
_
R
o
R
i
_
R
i
R
*
1
ln
_
R
o
R
i
a
_
_
a
k 1
a
t
_
1
a
t
_
M
a2
_
ln k
R
i
R
*
1
ln
_
k
1
a
t
k 1
_
_
a
t
k 1
(43)
and
c
1
D
_
1
a
t
_
(44)
Staat and Vu [12] dened their local limit pressures based on
both the von Mises and Tresca yield criteria, using the methodology
employed by Carter [9] but their own limit pressure solutions for
a cylinder with a throughwall crack (Eq. (24)) and a cylinder with
an extended crack (Eq. (9)), as
p
L
s
y
gln
_
R
o
R
i
_
glnk for
p
I
gs
y
!lnk
g
s
1
c
_
s
1
ln
_
R
o
R
i
_
c
R
i
a
R
*
2
ln
_
R
o
R
i
a
_
_
g
s
1
c
1
_
s
1
c
lnk
R
i
R
*
2
_
1
a
t
k1
_
ln
_
k
1
a
t
k1
_
_
for
p
I
gs
y
<lnk
45
_
_
where
and
p
I
gs
y
R
i
R
*
2
_
1
M
a3
ln
_
R
i
a
R
i
_
R
i
a
R
i
ln
_
R
o
R
i
a
__
R
i
R
*
2
_
1
M
a3
ln
_
1
a
t
k 1
_
_
1
a
t
k 1
_
ln
_
k
1
a
t
k 1
__
(47)
Thetwolocal limit pressure solutions arenowcomparedwiththe
FE results. There is only one set of well documented FE results for
local limit pressureavailable, whichis theresults duetoJunet al. [15]
based on the von Mises yield criterion and the crack ligament
yielding. Eqs. (42) and (45) (g 2=
3
p
) due to Carter [9] and Staat
andVu[12], respectively, are comparedwiththeFEresults duetoJun
et al. [15] in Figs. 1719 for k 1.05, 1.11 and 1.22, respectively, for
internal surface cracks with crack face pressure. Eq. (39) is also
plotted in the gures for comparison, though it is for cases without
crack face pressure. From Figs. 1719, the predictions using Carters
solution (Eq. (42)) are reasonably close to the FE results and
conservative for all the three k values except for shallow cracks in
a very thin cylinder (see Fig. 17(a) and (b)). It is also seen from the
gures that the solution due to Staat and Vu [12] for g 2=
3
p
is
nonconservative for short and shallow cracks, especially for the
cylinder with a very thin wall (Figs. 17 and 18). The formula due to
Kiefner et al. (Eq. (39)) shows very good and conservative predic
tions for k 1.11 (Fig. 18) and 1.22 (Fig. 19). However, it may be non
conservative for short and shallowcracks for k 1.05 (see Fig. 17).
3.4.2. External cracks
Similar to the cases of internal cracks, Carters local limit pres
sure solution [9] for an external surface crack (see Fig. 16(b)) under
internal pressure is dened as follows
s
1
c
_
1
a
t
_
R
i
R
*
2
ln
_
R
i
a
R
i
_
M
a3
_
ln
_
R
o
R
i
_
R
i
R
*
2
_
1
M
a3
ln
_
R
i
a
R
i
_
R
i
a
R
i
ln
_
R
o
R
i
a
__
_
1
a
t
_
R
i
R
*
2
ln
_
1
a
t
k 1
_
M
a3
_
ln k
R
i
R
*
2
_
1
M
a3
ln
_
1
a
t
k 1
_
_
1
a
t
k 1
_
ln
_
k
1
a
t
k 1
__
_
(46)
Internal crack, D
t
a
c
1
= 1
External crack, c
2
D
t
a
= 1
D D
2c
c
1 c
1
Crack
a
t
R
i
R
o
p
D D
2c
c
2
c
2
Crack
a
t
R
i
R
o
p
a
b
Fig. 16. Alternative partitions to dene global and local limit pressures for cylinders
with surface cracks under internal pressure.
Y. Lei / International Journal of Pressure Vessels and Piping 85 (2008) 825850 839
p
L
s
y
1
c
2
c
_
c
2
ln
_
R
o
R
i
_
cln
_
R
o
a
R
i
__
1
c
2
c
1
_
c
2
c
ln k ln
_
k
a
t
k 1
__
(48)
where
c
2
c
a
_
1
a
t
_
M
ax1
R
o
aln
_
R
o
R
o
a
_
a
k 1
a
t
_
1
a
t
_
M
ax1
_
k
a
t
k 1
_
ln
_
k
k
a
t
k 1
_
a
t
k 1
(49)
The local limit pressure for external crack due to Staat and Vu
[12] can be expressed as
p
L
s
y
g
s
2
c
_
s
2
ln
_
R
o
R
i
_
cln
_
R
o
a
R
i
__
g
s
2
c
1
_
s
2
c
ln k ln
_
k
a
t
k 1
__
(50)
where
s
2
c
1
a
t
M
ax2
1
(51)
No relevant FE results have been found for local limit pressures
of cylinders with external surface cracks.
a/c = 0.33
a/c = 0.167
a/c = 0.083 a/c = 0.05
a/c = 0.033
FE, Jun et al.
Prediction, Kiefner et al. (eqn. (39))
Prediction, Carter (eqn. (42))
Prediction, Staat (eqn. (45)), (
Prediction, present work (eqn. (67))
3) 2 =
0.0
0.2
0.4
0.6
0.8
1.0
1.2
a b
c d
e
p
L
/
p
0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
p
L
/
p
0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
p
L
/
p
0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
p
L
/
p
0
0 0.2 0.4 0.6 0.8
a/t
1
0 0.2 0.4 0.6 0.8
a/t
1
0 0.2 0.4 0.6 0.8
a/t
1
0 0.2 0.4 0.6 0.8
a/t
1
0 0.2 0.4 0.6 0.8
a/t
1
0.0
0.2
0.4
0.6
0.8
1.0
1.2
p
L
/
p
0
FE, Jun et al.
Prediction, Kiefner et al. (eqn. (39))
Prediction, Carter (eqn. (42))
Prediction, present work (eqn. (67))
FE, Jun et al.
Prediction, Kiefner et al. (eqn. (39))
Prediction, Carter (eqn. (42))
Prediction, present work (eqn. (67))
FE, Jun et al.
Prediction, Kiefner et al. (eqn. (39))
Prediction, Carter (eqn. (42))
Prediction, present work (eqn. (67))
FE, Jun et al.
Prediction, Kiefner et al. (eqn. (39))
Prediction, Carter (eqn. (42))
Prediction, present work (eqn. (67))
Prediction, Staat & Vu (eqn. (45)), ( ) 3 2 =
Prediction, Staat & Vu (eqn. (45)), ( ) 3 2 =
Prediction, Staat & Vu (eqn. (45)), ( ) 3 2 =
Prediction, Staat & Vu (eqn. (45)), ( ) 3 2 =
Fig. 17. Comparison of normalised local limit pressures between various solutions and FE results due to Jun et al. [15] for cylinders with internal surface cracks under internal
pressure (k 1.05, with crack face pressure).
Y. Lei / International Journal of Pressure Vessels and Piping 85 (2008) 825850 840
3.5. Limit load solutions for axial cracks in cylinders subjected to
combined membrane and throughwall bending stresses
Limit load solutions for axially cracked cylinders subjected to
combined membrane and throughwall bending stresses
(Fig. 20) are generally obtained from solutions for cracked plates
under combined tension and bending [1,9,20,21], ignoring the
effect of curvature. In R6 [1], the limit load solution for a thin
walled cylinder with an internal axial surface crack under
combined membrane and throughwall bending stresses is
a local solution based on the plate solution due to Goodall and
Webster [22] and Lei [23,24]. Actually, this solution can be
extended to thickwalled cylinders with internal/external
surface cracks as long as the bending stress tends to open the
crack because the plate solution [2224] was derived for any
thickness of the plate.
3.6. Limit load solutions for axially cracked cylinders
under combined loading
A limit load solution for thinwalled cylinders with axial surface
cracks under combined internal pressure, axial tension and global
bending was proposed by Desquines et al. [25], followed Kitching
et al. [4]. However, the limit pressures predicted using this solution
are much lower than those predicted using the solution due to
Kiefner et al. [3] for the limiting case of a cylinder with a through
wall crack under internal pressure alone.
Kim et al. [14] performed an FE analysis for a cylinder of t=R
m
0:05 with a surface crack of a=t 0:2 and a=c 0:0224 under
combined internal pressure and global bending and concluded
that a bending load has only a slight effect on the limit pressure for
axial cracks. This might not be true for short cracks where the
limit pressure of the cylinder approaches the limit pressure of the
a/c = 0.33 a/c = 0.167
a/c = 0.083 a/c = 0.05
a/c = 0.033
FE, Jun et al.
Prediction, Kiefner et al. (eqn. (39))
Prediction, Carter (eqn. (42))
Prediction, Staat & Vu (eqn. (45)),
Prediction, present work (eqn. 67)
( ) 3 2 =
0.0
0.2
0.4
0.6
0.8
1.0
1.2
p
L
/
p
0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
p
L
/
p
0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
p
L
/
p
0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
p
L
/
p
0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
p
L
/
p
0
0 0.2 0.4 0.6 0.8
a/t
1 0 0.2 0.4 0.6 0.8
a/t
1
0 0.2 0.4 0.6 0.8
a/t
a/t
1 0 0.2 0.4 0.6 0.8
a/t
1
0 0.2 0.4 0.6 0.8 1
FE, Jun et al.
Prediction, Kiefner et al. (eqn. (39))
Prediction, Carter (eqn. (42))
Prediction, Staat & Vu (eqn. (45)),
Prediction, present work (eqn. 67)
( ) 3 2 =
FE, Jun et al.
Prediction, Kiefner et al. (eqn. (39))
Prediction, Carter (eqn. (42))
Prediction, Staat & Vu (eqn. (45)),
Prediction, present work (eqn. 67)
( ) 3 2 =
FE, Jun et al.
Prediction, Kiefner et al. (eqn. (39))
Prediction, Carter (eqn. (42))
Prediction, Staat & Vu (eqn. (45)),
Prediction, present work (eqn. 67)
( ) 3 2 =
FE, Jun et al.
Prediction, Kiefner et al. (eqn. (39))
Prediction, Carter (eqn. (42))
Prediction, Staat & Vu (eqn. (45)),
Prediction, present work (eqn. 67)
( ) 3 2 =
a b
c d
e
Fig. 18. Comparison of normalised local limit pressures between various solutions and FE results due to Jun et al. [15] for cylinders with internal surface cracks under internal
pressure (k 1.11, with crack face pressure).
Y. Lei / International Journal of Pressure Vessels and Piping 85 (2008) 825850 841
crackfree cylinder when internal pressure only is applied. Further
investigation is necessary for the limit load of axially cracked
cylinders under combined loading.
4. New limit load solutions for cylinders with
axial cracks under internal pressure
The results of the reviewof the limit loads for axially cracked cylin
ders under internal pressure in Section 3 can be summarised as follows.
(1) For extended internal/external surface cracks, solutions due to
Staat and Staat and Vu (Eqs. (9) and (13)) are for thickwalled
cylinders and give good predictions of the available FE results.
(2) For throughwall cracks, the solution due to Staat and Vu (Eq.
(24)) is for thickwalled cylinders and gives good predictions of
available FE results for both thinwalled and thickwalled
cylinders. However, Eq. (24) is nonconservative for short and
shallow cracks because the backwall correction in the equa
tion is incorrect and the stress magnication factor, M
t4
, needs
to be recalibrated.
(3) For the global limit pressure of internal surface cracks, the limit
pressure solution due to Staat and Vu (Eq. (31)) is for thick
walled cylinders and gives good predictions for available FE
results for both thinwalled and thickwalled cylinders.
However, it overestimates the FE results for short and shallow
cracks due to the probleminthe solutionfor throughwall cracks
described in (2) and the pressure magnifying factor, R
i
a=R
i
,
applied to the term corresponding to the crackfree cylinder.
(4) For the global limit pressure of external surface cracks, the limit
pressure solution due to Staat and Vu (Eq. (37)) is for thick
walled cylinders and gives good predictions for available FE
results for thickwalled cylinders. However, it overestimates
a/c = 0.33 a/c = 0.167
a/c = 0.083
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0
0.2
0.4
0.6
0.8
1.0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
0 0.2 0.4 0.6 0.8 1
a/t
p
L
/
p
0
p
L
/
p
0
FE, Jun et al.
Prediction, Kiefner et al. (eqn. (39))
Prediction, Carter (eqn. (42))
Prediction, Staat & Vu (eqn. (45)),
Prediction, present work (eqn. (67))
( =2
a/t
FE, Jun et al.
Prediction, Kiefner et al. (eqn. (39))
Prediction, Carter (eqn. (42))
Prediction, Staat & Vu (eqn. (45)),
Prediction, present work (eqn. (67))
( ) 3 2 =
0.0
0.2
0.4
0.6
0.8
1.0
1.2
a/t
p
L
/
p
0
FE, Jun et al.
Prediction, Kiefner et al. (eqn. (39))
Prediction, Carter (eqn. (42))
a/c = 0.05
a/c = 0.033
a/t
p
L
/
p
0
FE, Jun et al.
Prediction, Kiefner et al. (eqn. (39))
Prediction, Carter (eqn. (42))
Prediction, Staat & Vu (eqn. (45)),
Prediction, present work (eqn. (67))
( = 2
a/t
p
L
/
p
0
FE, Jun et al.
Prediction, Kiefner et al. (eqn. (39))
Prediction, Carter (eqn. (42))
Prediction, Staat & Vu (eqn. (45)), ( =2
Prediction, present work (eqn. (67))
3)
3)
3)
a
b
c
d
e
Prediction, Staat & Vu (eqn. (45)),
Prediction, present work (eqn. (67))
( = 2 3)
Fig. 19. Comparison of normalised local limit pressures between various solutions and FE results due to Jun et al. [15] for cylinders with internal surface cracks under internal
pressure (k 1.22, with crack face pressure).
Y. Lei / International Journal of Pressure Vessels and Piping 85 (2008) 825850 842
the FE results for short and shallow cracks due to the problem
in the solution for throughwall cracks described in (2).
(5) For the local limit pressures for internal/external surface
cracks, Carters solutions (Eqs. (42) and (48)) are for thick
walled cylinders and give reasonably good and conservative
predictions of FE results for thinwalled cylinders. However,
the expressions for the local limit pressure are based on the
relevant global solutions. Therefore, they need to be rederived
to maintain consistency with the global solutions.
New limit pressure solutions for axially cracked thickwalled
cylinders under internal pressure are derived in this section. They
can also be used for thinwalled cylinders.
4.1. Throughwall cracks under internal pressure
Newlimit load solutions based on both the von Mises and Tresca
yield criteria for a thickwalled cylinder with a throughwall crack
under internal pressure are obtained by summing the pressure
corresponding to the frontwall failure, p
0
=M
tn
, and the backwall
correction, Dp
L
(see Eq. (C1) in Appendix C). FromEq. (C1), the limit
pressure without considering the crack face pressure can be
expressed as
p
L
s
y
p
0
M
tn
s
y
Dp
L
s
y
(52)
where M
tn
is the stress magnication factor and is dened using the
outer radius of the cylinder, with the coefcient being recalibrated
using the FE data for k 2 (see Appendix C), that is,
M
tn
_
1 1:4 r
2
o
_
0:5
_
1 1:4
c
2
R
o
t
_0:5
_
1 1:4
k 1
k
_
t
c
_
2
_
0:5
fork 2 (53)
The crack face pressure can be considered, following Staat
and Vu [12], by applying a factor R
i
=R
*
t
for the pressure corre
sponding to the frontwall failure and Eq. (52) can be further
expressed as
p
L
s
y
R
i
R
*
t
p
0
M
tn
s
y
Dp
L
s
y
(54)
where R
*
t
is dened in Eq. (B6) (see Appendix B) and the second
term in the righthand side of Eq. (54) is given by Eq. (B7) in
Appendix B. Note that Eq. (54) leads to the limit pressure for
a defectfree cylinder R
i
=R
*
t
p
0
< p
0
when c/0 because the factor
R
i
=R
*
t
does not change with crack length, c, noting that the second
term in the righthand side of Eq. (54) tends to zero and M
tn
/1. In
order to avoid this, the R
*
t
in Eq. (54) may be replaced by R
*
tn
, which
is dened as
Internal crack
External crack
b
2c
a
t
R
i
R
o
2c
a
t
R
i
R
o
a
b
m
Fig. 20. Geometry and dimensions of axial surface cracks in thickwalled cylinders subjected to membrane stress and throughwall bending.
Y. Lei / International Journal of Pressure Vessels and Piping 85 (2008) 825850 843
R
*
tn
R
i
without crack face pressure
R
i
t
2
forc ! t
R
i
c
2
forc < t
with crack face pressure
_
_
_
_
(55)
It is seen from Eq. (55) that for long cracks (c ! t) R
*
tn
R
*
t
and
for short cracks (c < t) it is a linear interpolation between R
i
t=2
and R
i
. This allows the effect of the crack face pressure factor to
vanish when the crack length tends to zero and the limit pressure of
the crackfree cylinder to be accurately reproduced. Here, choosing
c < t as short cracks is for consistency with the cases of surface
cracks with c < a and is somewhat arbitrary. Using R
*
tn
, the limit
pressure for a thickwalled cylinder with a throughwall crack
under internal pressure can be expressed as
where f
pt
is the crack face pressure factor and can be expressed as,
from Eq. (55),
f
pt
R
i
R
*
tn
t
2
1
1
1
2
k 1
for
t
c
1
R
i
R
i
c
2
t
c
t
c
1
2
k 1
for
t
c
> 1
with crack face pressure
_
_
_
_
(57)
The new solution, Eq. (56) (g 2=
3
p
), is compared with
other existing solutions and the FE data due to Staat and Vu [12]
and Kim et al. [13] in Figs. 6 and 7. From Figs. 6 and 7, Eq. (56)
provides the best predictions of the FE results compared with
all other solutions. It is slightly conservative compared with the
FE data for cases without crack face pressure (Fig. 6) and accu
rate or slightly nonconservative for cases with crack pressure
(Fig. 7).
4.2. Surface cracks under internal pressure
4.2.1. Internal cracks (global)
New limit load solutions based on both the von Mises and
Tresca yield criteria for a thickwalled cylinder with an internal
surface crack under internal pressure are obtained by summing
the limit pressure corresponding to the cylinder of inner radius
R
i
and thickness a with a throughwall crack of length 2c
(Cylinder A in Fig. 9(a)) and that for the crackfree cylinder of
inner radius R
i
a and thickness t a (Cylinder B in Fig. 9(a)),
that is
p
L
s
y
p
L
Cylinder A
s
y
F
pt
R
i
R
*
2n
p
L
Cylinder B
s
y
(58)
where R
i
=R
*
2n
is the crack face pressure factor dened for Cylinder
A. The equivalent radius R
*
2n
is the R
*
tn
for Cylinder A and can be
obtained by applying Eq. (55) to Cylinder A, that is
R
*
2n
R
i
without crack face pressure
R
i
a
2
forc ! a
R
i
c
2
forc < a
with crack face pressure
_
_
_
_
(59)
In Eq. (58), F
pt
is the pressure transfer factor and is dened as
F
pt
1
a
R
i
_
1
1
M
an
_
(60)
where M
an
is the stress magnication factor for Cylinder A and can
be obtained by applying Eq. (53) to Cylinder A, that is
M
an
_
11:4
c
2
R
i
aa
_0:5
_
11:4
a
t
k1
_
a
c
_
2
1
a
t
k1
_
0:5
for
_
1
a
t
k1
_
2 (61)
The pressure transfer factor, F
pt
, is applied to the termin Eq. (58)
representing the limit pressure of the crackfree cylinder (Cylinder
B in Fig. 9(a)) to capture the behaviour of pressure transferring from
the inner surface of Cylinder A to the inner surface of Cylinder B
(Fig. 9(a)). For an extreme case c/N and hence M
an
/N, i.e. an
extended penetrating crack in Cylinder A in Fig. 9(a), F
pt
tends to
R
i
a=R
i
1a=R
i
because Cylinder A in Fig. 9(a) is almost
elastic and the pressure transfer is based on radial force equilib
rium. Another extreme case is c/0 and hence M
an
/1. In this case,
F
pt
tends to 1 because the fully yielded Cylinder A in Fig. 9(a) cannot
bear any more pressure difference and the pressure is transferred
constantly from the inner surface of Cylinder A (Fig. 9(a)) to the
inner surface of Cylinder B (Fig. 9(a)). For all other cases between
these two limits, the factor is estimated using linear interpolation
based on 1=M
an
.
Determining the limit pressure of Cylinder A in Fig. 9(a) by
applying Eq. (56) to a cylinder of inner radius R
i
and outer radius
R
i
a with a throughwall crack of length 2c and the limit pressure
for the defectfree cylinder of inner radius R
i
a and outer radius
R
o
(Cylinder B in Fig. 9(a)), the limit pressure of a thickwalled
cylinder with an internal surface crack can be obtained from Eq.
(58) and expressed as
p
L
s
y
R
i
R
*
tn
g
M
tn
ln
_
R
o
R
i
_
_
_
_
1
1
2
1
1
M
tn
t
R
*
tn
_
2
1
4
_
1
1
M
2
tn
__
t
R
*
tn
_
2
_
_
1
1
2
_
1
1
M
tn
_
t
R
*
tn
_
_
_
fpt
g
Mtn
ln k
_
_
_
1
1
2
1
1
M
tn
k 1f
pt
_
2
1
4
_
1
1
M
2
tn
_
_
k 1f
pt
_
2
_
_
1
1
2
_
1
1
M
tn
_
k 1f
pt
_
_
_
(56)
Y. Lei / International Journal of Pressure Vessels and Piping 85 (2008) 825850 844
where f
ps
is the crack face pressure factor and is dened, using
Eq. (59), as
f
ps
R
i
R
*
2n
1 without crackfacepressure
R
i
R
i
a
2
1
1
1
2
a
t
k1
for
a
c
1
R
i
R
i
c
2
a
c
a
c
1
2
a
t
k1
for
a
c
>1
withcrackfacepressure
_
_
_
_
(63)
Note that Eq. (62) is valid for p
L
=gs
y
1 because the pressure
transfer factor for long cracks is dened based on the assumption of
an elastic Cylinder A and yielding may take place in Cylinder A even
for the case of an extended surface crack when p
L
>gs
y
. This
condition is always satised for cylinders of k2:718 with any
crack size.
Eq. (62) reduces to Eq. (56) for throughwall cracks, when
a=t/1, and to Eq. (9) for internal extended cracks when a=c/0
and a=t > 0. It also reproduces the limit pressure for crackfree
thickwalled cylinders when a=t 0 or a=c/N.
The new solution, Eq. (62) with g 2=
3
p
, is compared with
other existing solutions and the FE data due to Staat and Vu [12] in
Figs. 10 and 11 and those due to Kimet al. [13] in Figs. 12 and 13. For
cases without crack face pressure (Fig. 10), Eq. (62) has largely
removed the nonconservatism of the solution due to Staat and Vu
[12] for short cracks. From the gure, the predictions using Eq. (62)
are close to the FE results and conservative. For cases with crack
face pressure (Figs. 1113), Eq. (62) has also improved the non
conservatism of the solution of Staat and Vu [12] for short
and shallow cracks for thickwalled cylinders (Fig. 11) and gives
reasonably good and conservative predictions for both thickwalled
(Fig. 11) and thinwalled (Figs. 12 and 13) cylinders.
4.2.2. External cracks (global)
New limit load solutions based on both the von Mises and
Tresca yield criteria for a thickwalled cylinder with an external
surface crack under internal pressure are obtained by directly
summing the limit pressure corresponding to the cylinder of inner
radius R
o
a and outer radius R
o
with a throughwall crack of
length 2c (Cylinder A in Fig. 9(b)) and that for the crackfree
cylinder of inner radius R
i
and outer radius R
o
a (Cylinder B in
Fig. 9(b)), that is
p
L
s
y
p
L
Cylinder A
s
y
p
L
Cylinder B
s
y
(64)
In Eq. (64), a simple addition for the limit pressures for the two
cylinders is used because Cylinder B in Fig. 9(b) is defectfree and
the pressure transfer factor from the inner surface of Cylinder B
(Fig. 9(b)) at R
i
to the inner surface of Cylinder A (Fig. 9(b)) at R
o
a
is unity (see Section 4.2.1 above).
Determining the limit pressure of Cylinder A of Fig. 9(b) by
applying Eq. (56) to a cylinder of inner radius R
o
a and outer
radius R
o
with a throughwall crack of length 2c and the limit
pressure for the defectfree cylinder of inner radius R
i
and outer
radius R
o
a (Cylinder B in Fig. 9(b)), the limit pressure of a thick
walled cylinder with an external surface crack can be obtained from
Eq. (64) and expressed as
pL
s
y
_
_
g
Maxn
ln
_
Ro
Roa
_
_
_
Ro
R
i
1
2
1
1
Maxn
a
t
t
R
i
_
2
1
4
_
1
1
M
2
axn
_
_
a
t
t
R
i
_
2
_
R
o
R
i
1
2
_
1
1
M
axn
_
a
t
t
R
i
_
_
_
_
_
gln
_
Roa
R
i
_
g
_
1
Maxn
ln
_
k
k
a
t
k1
_
ln
_
k
a
t
k 1
_
_
_
_
k
1
2
1
1
Maxn
a
t
k 1
_
2
1
4
_
1
1
M
2
axn
_
_
a
t
k 1
_
2
_
k
1
2
_
1
1
M
axn
_
a
t
k 1
_
_
_
(65)
pL
s
y
_
_
R
i
R
*
2n
g
M
an
ln
_
R
i
a
R
i
_
_
_
_
1
1
2
1
1
M
an
a
t
t
R
*
2n
_
2
1
4
_
1
1
M
2
an
__
a
t
t
R
*
2n
_
2
_
_
1
1
2
_
1
1
M
an
_
a
t
t
R
*
2n
_
_
_
_
_
1
a
R
i
_
1
1
M
an
__
R
i
R
*
2n
gln
_
R
o
R
i
a
_
gf
ps
_
1
M
an
ln
_
1
a
t
k 1
_
_
1
a
t
k 1
_
1
1
M
an
__
ln
_
k
1
a
t
k 1
__
_
_
_
1
1
2
1
1
M
an
a
t
f
ps
k 1
_
2
1
4
_
1
1
M
2
an
_
_
a
t
f
ps
k 1
_
2
_
_
1
1
2
_
1
1
M
an
_
a
t
f
ps
k 1
_
_
_
for k 2:718
(62)
Y. Lei / International Journal of Pressure Vessels and Piping 85 (2008) 825850 845
where the stress magnication factor, M
axn
, can be obtained by
applying Eq. (53) to Cylinder A in Fig. 9(b) and expressed as
M
axn
_
1 1:4
c
2
R
o
a
_0:5
_
1 1:4
k 1
a
t
k
_
a
c
_
2
_
0:5
for
k
k
a
t
k 1
2 (66)
Eq. (65) reduces to Eq. (56) for throughwall cracks when a=t/1
and to Eq. (13) for external extended cracks when a=c/0 and
a=t > 0. It also reproduces the limit pressure for crackfree thick
walled cylinders when a=t 0 or a=c/N.
The new solution, Eq. (65) with g 2=
3
p
, is compared with
other existing solutions and the FE data due to Staat and Vu [12] in
Fig. 14 and those due to Zarrabi [16] in Fig. 15. From the gures, Eq.
(65) has largely removed the nonconservatism of the solution due
to Staat and Vu [12] for deep and short cracks. It is also seen from
the gures that the predictions using Eq. (65) are very close to the
FE data and conservative for all cases shown in Figs. 14 and 15
except for the cases with very shallow cracks, where the FE results
are slightly overestimated by Eq. (65).
4.2.3. Internal cracks (local)
A new local limit pressure solution for a thickwalled cylinder
with an internal surface crack under internal pressure is obtained
from the methodology used by Carter [9] (see Section 3.4.1
above) based on the new limit load solutions for thickwalled
cylinders with internal surface cracks (Eq. (62)) and the limit
load solution for thickwalled cylinders with internal extended
cracks under internal pressure due to Staat and Vu [12] (Eq. (9)).
Following Carter [9], the local limit pressure for a thickwalled
cylinder with an internal surface crack of depth a and length 2c
can be expressed as the weighted sum of the limit pressures of
a cylinder of length 2c with an internal extended crack of depth
a and two crackfree cylinders of length h
1
(refer to Fig. 16(a)
with c
1
replaced by h
1
), that is
p
L
s
y
1
h
1
c
_
h
1
ln
_
R
o
R
i
_
c
R
i
R
*
2
R
i
a
R
i
ln
_
R
o
R
i
a
_
_
z
1
h
1
c
1
_
h
1
c
ln k f
ps
_
1
a
t
k 1
_
ln
_
k
1
a
t
k 1
__
(67)
where R
i
=R
*
2
f
ps
for c ! a (see Eq. (63)) and R
i
=R
*
2
zf
ps
for c < a
have been adopted. The normalised equivalent length of the crack
free cylinder, h
1
=c, can be obtained by following Eqs. (41)(44) but
using Eq. (62) as the global limit pressure for a thickwalled
cylinder with an internal surface crack and Eq. (9) as the limit
pressure for a thickwalled cylinder with an internal extended
crack. The result can be expressed as
Note that the backwall correction terms in Eqs. (9) and (62)
have been omitted as only local ligament yielding is considered.
The g factor is also set to unity because the comparison with
the FE data below shows the solution based on the von Mises
yield criterion may be nonconservative for short and shallow
cracks.
The new solution, Eq. (67), is compared with other existing
solutions and the FE data due to Jun et al. [15] in Figs. 1719 for
k 1.05, 1.11 and 1.22, respectively. From the gures, the limit
pressure obtained using Eq. (67) is very close to, but slightly
higher than that predicted using Carters solution. It is also
seen from the gures that the predictions using Eq. (67) are
reasonably close to and conservative compared with the FE
results for all cases shown in Figs. 1719. The conservatism of
Eq. (67) may increase with increase of k, noting the trends
shown in Figs. 1719.
4.2.4. External cracks (local)
A new local limit pressure solution for a thickwalled cylinder
with an external surface crack under internal pressure is obtained
from the methodology used by Carter [9] (see Section 3.4.2 above)
based on the new limit load solutions for thickwalled cylinders
with external surface cracks (Eq. (65)) and the limit load solution
for thickwalled cylinders with external extended cracks under
internal pressure due to Staat and Vu [12] (Eq. (13)). Following
Carter [9], the local limit pressure for a thickwalled cylinder with
an external surface crack of depth a and length 2c can be expressed
as the weighted sumof the limit pressures of a cylinder of length 2c
with an external extended crack of depth a and two crackfree
cylinders of length h
2
(refer to Fig. 16(b) with c
2
replaced by h
2
),
that is
p
L
s
y
1
h
2
c
_
h
2
ln
_
R
o
R
i
_
cln
_
R
o
a
R
i
__
1
h
2
c
1
_
h
2
c
ln k ln
_
k
a
t
k 1
_
_
(69)
The normalised equivalent length of the crackfree cylinder,
h
2
=c, can be obtained by following Eqs. (48)(50) but using Eq. (65)
as the global limit pressure for a thickwalled cylinder with an
external surface crack and Eq. (13) as the limit pressure for a thick
walled cylinder with an external extended crack. The result can be
expressed as
h
2
c
1
a
t
M
axn
1
(70)
Note that, again, the backwall correction terms in Eqs. (13) and
(65) have been omitted as only local ligament yielding is consid
ered. The g factor is also set to unity because of the same reason
given in Section 4.2.3 for internal cracks.
No relevant FE results have been found for local limit pressures
of cylinders with external surface cracks.
h
1
c
_
1
a
t
_
f
ps
_
ln
_
R
i
a
R
i
_
a
t
t
R
i
ln
_
R
o
R
i
a
__
M
an
_
ln
_
Ro
R
i
_
f
ps
_
1
Man
ln
_
R
i
a
R
i
_
_
1
a
R
i
_
1
1
Man
__
ln
_
Ro
R
i
a
___
1
a
t
_
f
ps
_
ln
_
1
a
t
k 1
_
a
t
k 1ln
_
k
1
a
t
k1
__
M
an
_
ln k f
ps
_
1
Man
ln
_
1
a
t
k 1
_
_
1
a
t
k 1
_
1
1
Man
__
ln
_
k
1
a
t
k1
___
(68)
Y. Lei / International Journal of Pressure Vessels and Piping 85 (2008) 825850 846
5. Conclusions
1. The limit load solutions for axially cracked cylinders have been
reviewed and compared with available FE results. The ndings
are as follows.
(1) For extended internal/external cracks under internal pres
sure, solutions due to Staat and Vu (Eqs. (9) and (13)) are for
thickwalled cylinders and give the best predictions of the
available FE results.
(2) For throughwall cracks under internal pressure, the solu
tion due to Staat and Vu (Eq. (24)) is for thickwalled
cylinders and gives the best predictions of available FE
results for both thinwalled and thickwalled cylinders.
However, it is nonconservative for short cracks because
the backwall correction in the equation is incorrect and
the stress magnication factor needs to be recalibrated.
(3) For the global limit pressure of internal surface cracks, the
solution due to Staat and Vu (Eq. (31)) is for thickwalled
cylinders and gives the best prediction of available FE results
for both thinwalled and thickwalled cylinders. However, it
overestimates the FE results for short and shallowcracks due
to the problems in the solution for throughwall cracks
addressedin(2) andthepressureamplifyingfactor, R
i
a=R
i
,
applied to the term corresponding to the crackfree cylinder.
(4) For the global limit pressure of external surface cracks, the
solution due to Staat and Vu (Eq. (37)) is for thickwalled
cylinders and gives the best prediction of available FE results
for thickwalled cylinders. However, it overestimates the FE
results for short and throughwall cracks due to the problem
in the solution for throughwall cracks addressed in (2).
(5) For the local limit pressures of internal/external surface
cracks, Carters solutions (Eqs. (42) and (48)) are for thick
walled cylinders and give reasonably good and conserva
tive predictions of available FE results for thinwalled
cylinders. However, the expressions for the local limit
pressure are based on the corresponding global solutions.
Therefore, they need to be rederived to maintain consis
tency with the global solutions. The solutions due to Staat
and Vu (Eqs. (45) and (50)) are for thickwalled cylinders.
However, the solution for internal cracks (Eq. (45)) with
g 2=
3
p
is nonconservative for short and shallow
cracks, especially for the cylinder with a very thin wall
compared with the available FE results.
(6) Little information for the effect of other load types, such as
axial tension and global bending moment, on the limit
pressure of a cylinder with an axial crack can be found.
Limit load solutions for axially cracked cylinders under
combined internal pressure, tension and global bending are
currently lacking.
2. New limit pressure solutions for thickwalled cylinders with
axial cracks under internal pressure have been developed to
overcome the problems addressed in Conclusion 1, above. The
new solutions are
(1) global solution for throughwall cracks,
(2) global solutions for internal/external surface cracks,
(3) local solutions for internal/external surface cracks.
3. The newly developed limit pressure solutions have been
compared with available FE data and the results show that the
predictions using the newsolutions are conservative and agree
well with the FE results.
Acknowledgements
The author wishes to acknowledge Dr. P.J. Budden of British
Energy Generation Ltd. for his comments on this paper and Prof.
Manfred Staat of Aachen University of Applied Sciences (Germany)
for providing FE data. This paper is published by permission of
British Energy Generation Ltd.
Appendix A. Folias factor
The Folias factor is a stress magnication factor due to the
curvature of shells and was rst reported by Folias [26] to address
the stress increase in the near crack tip area in a thinwalled
spherical vessel with a fully penetrating crack under internal
pressure. Folias [19] then derived the factor for a thinwalled
cylindrical vessel with a penetrating axial or circumferential crack
under internal pressure, based on elastic thinshell theory. At that
time, Folias [19] obtained a theoretical solution for the stress
magnication factor for axial cracks only for r
m
0:55 and
expressed it as
M
t
1 fr
2
m
_
(A1)
with
f 1:61 forr
m
0:55 (A2)
where
r
m
c
R
m
t
p (A3)
Later, Erdogan and Kibler [27] solved the problem numerically
and obtained the solution for axial cracks for r
m
4:4. The
results are tabulated in Table A1. Folias [28] found that the
numerical results could still be expressed in the form of Eq. (A1),
but the coefcient f 1:05 provided a good t for the data,
that is,
f 1:05 forr
m
4:4 (A4)
Kiefner et al. [3] found that the limit pressure data from burst
tests of pipes with throughwall defects could be well correlated
using a Folias factor. In their paper [3], Kiefner et al. tted the Folias
Table A1
Numerical solution of Folias factor [27,17].
r
m
M
t
0.110011 1.0096
0.220022 1.0371
0.330033 1.0795
0.440044 1.1344
0.550055 1.1993
0.660066 1.2723
0.770077 1.3519
0.880088 1.4367
0.990099 1.5256
1.10011 1.6177
1.210121 1.7122
1.320132 1.8085
1.430143 1.906
1.540154 2.0045
1.650165 2.1035
1.787679 2.2276
1.925193 2.3519
2.062706 2.4761
2.20022 2.5999
2.337734 2.7232
2.475248 2.8459
2.750275 3.0895
3.025303 3.3303
3.30033 3.5681
3.575358 3.8029
3.850385 4.0347
4.125413 4.2637
4.40044 4.4895
Y. Lei / International Journal of Pressure Vessels and Piping 85 (2008) 825850 847
factor data shown in Table A1 [27,17] and found the data could be
well represented by the following equation
M
t
1 1:255 r
2
m
0:0135 r
4
m
_
(A5)
Fig. A1 compares the three equations with the numerical data in
Table A1. From Fig. A1, Eqs. (A5) and (A1) with f 1:61 or 1.05 can
predict the numerical data very well in the region r
m
0:55. It is
also seen that Eq. (A1) with f 1:05 is a good representation and
Eq. (A5) is the best t of the data inthe regionr
m
4:4. However, Eq.
(A1) with f 1:61 is very conservative in the region 1 < r
m
4:4.
Several factors should be claried when using the Folias factor.
Firstly, the Folias factor was derived for elastic material properties.
It was used in the limit load solutions because Kiefner et al. found
that it could correlate their experimental data very well. The author
has not found any theoretical proof for elastic plastic materials.
Secondly, the Folias factor was obtained for thinwalled shells.
There is no solution for thickwalled shells. Finally, the theoretical
solution for the Folias factor is available only for r
m
4:4. Special
care should be made for problems beyond this limitation.
Appendix B. Backwall effect on the limit pressure of
a cylinder with an axial crack
For a cylinder with an axial defect under internal pressure, the
global limit load of the defective cylinder is the pressure corre
sponding to the plastic collapse of both the frontwall of the cylinder
containing the defect and the defectfree backwall. The frontwall is
weaker than the backwall due to the defect. Denoting the pressure
corresponding to the collapse of the frontwall, p
Lf
, the total global
limit pressure can be expressed as p
Lf
Dp
L
, where Dp
L
is the extra
pressure the backwall can bear after the onset of the frontwall
collapse. For thinwalled cylinders, Dp
L
is negligible. However, it
may become signicant for cylinders with very thick walls. In this
Appendix, Dp
L
for throughwall andsurface cracks will be estimated.
The backwall of a cracked cylinder can be treated as a plate
of thickness t subjected to combined tension force, N
L
, and bending
moment, M
Lp
, due to the internal pressure, p
Lf
Dp
L
. The limit load
of an uncracked plate with a thickness t and unit width under
combined tension and bending can be expressed as [29]
N
L
s
y
t
4l
2
1
_
2l (B1)
l
M
Lp
tN
L
(B2)
where l is the load ratio.
Cylinder with throughwall cracks
For a cylinder with an axial throughwall crack of length 2c
subjected to internal pressure, the tensile force, N
L
, and the moment,
M
Lp
, in the backwall due to the internal pressure, p
Lf
Dp
L
, are as
follows (see Fig. B1). The resultant force and moment in the back
wall can be obtained by taking the force equilibrium along the
direction normal to the crack face and moment equilibrium in the
backwall, assuming that the backwall only bears half of the force
due to p
Lf
but the full force due to Dp
L
, and expressed as
N
L
2R
*
t
Dp
L
R
*
t
p
Lf
R
*
t
_
2Dp
L
p
Lf
_
(B3)
M
Lp
_
N
L
R
*
t
p
Lf
_
_
R
*
t
t
2
_
2R
*
t
Dp
L
_
R
*
t
t
2
_
(B4)
The load ratio, l, following Eq. (B2), for this geometry is
l
M
Lp
N
L
t
2Dp
L
_
1
1
2
t
R
*
t
_
_
2Dp
L
p
Lf
_
t
R
*
t
(B5)
InEqs. (B3)(B5), R
*
t
is the equivalent radius to include the effect of
the crack face pressure and is dened, for long cracks (c ! t), as
R
*
t
R
i
without crack face pressure
R
i
t
2
with crack face pressure
_
(B6)
The normalised limit pressure increase due to the back
wall effect, Dp
L
=s
y
, can be obtained by inserting Eqs. (B3) and
(B5) into Eq. (B1) and solving for Dp
L
=s
y
. The result can be
expressed as
Dp
L
s
y
_
1
1
2
1
1
M
tn
t
R
*
t
_
2
1
4
_
1
1
M
2
tn
__
t
R
*
t
_
2
_
1
1
2
_
1
1
M
tn
_
t
R
*
t
_
B7
In Eq. (B7), the following assumption has been adopted
p
Lf
s
y
z
t
R
*
t
1
M
tn
(B8)
using Eq. (19), replacing M
t2
by M
tn
dened by Eq. (53).
0
1
2
3
4
5
6
0 1 2 3 4 5
m
M
t
Folias Factor, data
Kiefner equation (eqn. (A5))
Eqn. (A1) with = 1.61
Eqn. (A1) with = 1.05
Fig. A1. Comparison of Folias factor between numerical data [27,17] and three
equations.
R
i
t
p
L
p
Lf
+
Front wall with a
throughwall crack
Back wall
N
L
M
Lp
2R
*
t
Fig. B1. Backwall loads for a cylinder with a throughwall crack (R
*
t
shown for the case
of crack face pressure).
Y. Lei / International Journal of Pressure Vessels and Piping 85 (2008) 825850 848
Cylinder with internal surface crack
For a cylinder with an axial internal surface crack of length 2c
and depth a subjected to internal pressure, the backwall effect is
only from the cylinder of inner radius R
i
and thickness a with
a throughwall crack of length 2c (Fig. B2). The normalised
pressure increase due to the backwall effect, Dp
L
=s
y
, for this case
can be obtained directly from Eq. (B7) by replacing t, R
*
t
and M
tn
in Eq. (B7) by a, R
*
2
and M
an
, respectively. The result can be
expressed as
Dp
L
s
y
_
1
1
2
1
1
M
an
a
t
t
R
*
2
_
2
1
4
_
1
1
M
2
an
__
a
t
t
R
*
2
_
2
_
1
1
2
_
1
1
M
an
_
a
t
t
R
*
2
_
B9
where R
*
2
is dened in Eq. (10) and M
an
is dened in Eq. (61).
Cylinder with external surface cracks
For a cylinder with an axial external surface crack of length 2c
and depth a subjected to internal pressure, the backwall effect is
only from the cylinder of inner radius R
o
a and thickness a with
a throughwall crack of length 2c (Fig. B3). The normalised pressure
increase due to the backwall effect, Dp
L
=s
y
, for this case can be
obtained directly fromEq. (B7) by replacing t, R
*
t
and M
tn
in Eq. (B7)
by a, R
o
a and M
axn
, respectively, and then applying a factor
R
o
a=R
i
to the righthand side of Eq. (B7). The result can be
expressed as
Dp
L
s
y
_
k
1
2
1
1
M
axn
a
t
t
R
i
_
2
1
4
_
1
1
M
2
axn
_
_
a
t
t
R
i
_
2
_
k
1
2
_
1
1
M
axn
_
a
t
t
R
i
_
B10
where M
axn
is dened in Eq. (66).
Appendix C. Calibration of the stress magnication factor
for cylinders with throughwall cracks
The limit pressure, p
L
, for a cylinder with a throughwall crack
subjected to internal pressure may generally be expressed as
p
L
Dp
L
p
0
1
M
tn
(C1)
where p
0
is the limit pressure for crackfree cylinders, Dp
L
is the
pressure increase due to the backwall effect (see Appendix B) and
M
tn
is the stress magnication factor. Staat and Vu have shown that
M
tn
can be expressed by the following equation
M
tn
1 fr
2
o
_
(C2)
The factor r
o
is a function of crack length, c, cylinder outer radius,
R
o
, and cylinder wall thickness, t, and is expressed by Eq. (26). The
coefcient f may be calibrated from FE or experimental data.
Combining Eqs. (C1) and (C2), the relationship between
p
0
=p
L
Dp
L
and r
o
is as follows
_
p
0
p
L
Dp
L
_
2
1 fr
2
o
(C3)
This equation may be used to calibrate f. Fig. C1 shows the FE
limit pressure data for cylinders with throughwall cracks under
internal pressure (without crack face pressure) due to Staat and Vu
[12], plotted as p
0
=p
L
2
against r
2
o
. FromFig. C1, the data for various
k are widely scattered with increasing r
o
and the coefcient, f, may
depend on k. Moreover, the relationship between p
0
=p
L
2
and r
2
o
is
nonlinear for big k values. The FE data are then replotted in Fig. C2
considering the backwall effect, Dp
L
. From the gure, the FE data
for all k values considered tend to collapse to one line and can be
represented by a straight line with a slope f 1:4. Note that Dp
L
is
a function of M
tn
(see Eq. (B7) in Appendix B) and, therefore, f. The
result f 1:4 was obtained by increasing f gradually and checking
R
i
t
p
L
p
Lf
+
N
L
M
Lp
2R
*
2
Front wall
Back wall
a
Fig. B2. Backwall loads for a cylinder with an internal surface crack (R
*
2
shown for the
case of crack face pressure).
R
0
N
L
Front wall
Back wall
a
t
M
Lp
p
L
p
Lf
+
Fig. B3. Backwall loads for a cylinder with an external surface crack.
0
2
4
6
8
10
12
0 2 4 6 8 10 12 14
(
p
0
/
p
L
)
2
k = 1.1
k = 1.25
k = 1.5
k = 1.75
k = 2
Fig. C1. FE data [12] plotted in the form of Eq. (C3) for Dp
L
0.
Y. Lei / International Journal of Pressure Vessels and Piping 85 (2008) 825850 849
the agreement with the FE data to obtain an upperbound esti
mation of M
tn
for all the FE data.
References
[1] R6, Assessment of the Integrity of Structures Containing Defects, Revision 4,
British Energy Generation Ltd., 2001, with amendments to May 2007.
[2] Miller AG. Review of limit loads of structures containing defects. International
Journal of Pressure Vessels and Piping 1988;32:197327.
[3] Kiefner JF, Maxey WA, Eiber RJ, Duffy AR. Failure stress levels of aws in
pressurised cylinders. ASTM STP 536. Philadelphia, USA: American Society for
Testing and Materials; 1973. p. 461481.
[4] Kitching R, Davis JK, Gill SS. Limit pressures for cylindrical shells with unre
inforced openings of various shapes. Journal of Mechanical Engineering
Science 1970;12:31330.
[5] Kitching R, Zarrabi K. Limit and burst pressures for cylindrical shells with
partthrough slots. International Journal of Pressure Vessels and Piping 1982;
10:23570.
[6] Ewing DJF. On the plastic collapse of a thinwalled pressurized pipe with an
axial crack. CEGB Report TPRD/L/2566/N83; 1984.
[7] Chell GG. ADISC: a computer program for assessing defects in spheres and
cylinders. CEGB Report TPRD/L/MT0237/M84; 1984.
[8] Chell GG. Elasticplastic fracture mechanics. CEGB Report RD/L/R2007; 1979.
[9] Carter AJ. A library of limit loads for FRACTURE.TWO. Nuclear Electric Report
TD/SID/REP/0191. Berkeley; 1991.
[10] Staat M. Plastic collapse analysis of longitudinally awed pipes and vessels.
Nuclear Engineering and Design 2004;234:2543.
[11] Staat M. Local and global collapse pressure of longitudinally awed pipes and
cylindrical vessels. International Journal of Pressure Vessels and Piping
2005;82:21725.
[12] Staat M, Vu Duc Khoi. Limit analysis of aws in pressurized pipes and
cylindrical vessels. Part I: axial defects. Engineering Fracture Mechanics
2007;74:43150.
[13] Kim YunJae, Shim DoJun, Huh NamSu, Kim YoungJin. Plastic limit pressures
for cracked pipes using nite element limit analyses. International Journal of
Pressure Vessels and Piping 2002;79:32130.
[14] Kim YunJae, Shim DoJun, Nikbin Kamran, Kim YoungJin, Hwang SeongSik,
Kim JoungSoo. Finite element based plastic limit loads for cylinders with
partthrough surface cracks under combined loading. International Journal of
Pressure Vessels and Piping 2003;80:52740.
[15] Jun HK, Choi JB, Kim YJ, Park YW. The plastic collapse solutions based on nite
element analyses for axial surface cracks in pipelines under internal pressure.
ASME PVP 1998;373:5238.
[16] Zarrabi K, Zhang H, Nhim K. Plastic collapse pressure of cylindrical vessels
containing longitudinal surface cracks. Nuclear Engineering and Design
1997;168:3137.
[17] Folias ES. On the effect of initial curvature on cracked at sheets. International
Journal of Fracture Mechanics 1969;5:32746.
[18] Erdogan F. Ductile failure theories for pressurized pipes and containers.
International Journal of Pressure Vessels and Piping 1976;4:25383.
[19] Folias ES. An axial crack in a pressurized cylindrical shell. International Journal
of Fracture Mechanics 1965;1:10413.
[20] API Recommended Practice 579 Fitnessforservice. 1st ed. American Petro
leum Institute; 2000.
[21] Dillstro m P, Bergman M, Brickstad B, Zang W, SattariFar I, Sund G, et al. A
combined deterministic and probabilistic procedure for safety assessment of
components with cracks handbook. DET Norske Veritas; 2004. RSE R&D
Report 2004/01, Revision 41.
[22] Goodall IW, Webster GA. Theoretical determination of reference stress for
partially penetrating aws in plates. International Journal of Pressure Vessels
and Piping 2001;78:68795.
[23] Lei Y. Jintegral and limit load analysis of semielliptical surface cracks in
plates under bending. International Journal of Pressure Vessels and Piping
2004;81:3141.
[24] Lei Y. A global limit load solution for plates with semielliptical surface cracks
under combined tension and bending. ASME PVP 2004;475:12531.
[25] Desquines J, Poette C, Michel B, Wielgosz C, Martelet B. Limit load of an axially
cracked pipe under combined pressure bending and tension. In: Petit J, de
Fouquet J, Henaff G, Villechaise P, Dragon A, editors. Mechanisms and
mechanics of damage and failure, ECF 11 Proceedings; 1996. p. 21692174.
[26] Folias ES. A nite line crack in a pressurized spherical shell. International
Journal of Fracture Mechanics 1965;1:2046.
[27] Erdogan F, Kibler JJ. Cylindrical and spherical shells with cracks. International
Journal of Fracture Mechanics 1969;5:22937.
[28] Folias ES. On the fracture of nuclear reactor tubes. Paper C4/5, SMiRT III.
London; 1975.
[29] Lei Y. Jintegral and limit load analysis of semielliptical surface cracks in
plates under combined tension and bending. International Journal of Pressure
Vessels and Piping 2004;81:4356.
0
5
10
15
20
25
0 2 4 6 8 10 12 14 16
k = 1.1
k = 1.25
k = 1.5
k = 1.75
k = 2
= 1.4
1
1.4
(
p
0
/
(
p
L

p
L
)
)
2
Fig. C2. FE data shown in Fig. C1 replotted in the form of Eq. (C3) with considering the
backwall correction.
Y. Lei / International Journal of Pressure Vessels and Piping 85 (2008) 825850 850