ARTICLE IN PRESS

International Journal of Pressure Vessels and Piping 83 (2006) 663–671

www.elsevier.com/locate/ijpvp

Average shear stress yield criterion and its application to

plastic collapse analysis of pipelines
Xian-Kui Zhu, Brian N. Leis
Battelle Memorial Institute, 505 King Avenue, Columbus, OH 43201, USA
Received 25 January 2006; received in revised form 8 June 2006; accepted 13 June 2006

Abstract

It is known that available analytical and empirical solutions for the burst pressure of defect-free line pipes cannot broadly fit
experimental data for different materials. Usually the Tresca prediction provides a lower bound to the burst pressure, and the von Mises
prediction provides an upper bound to the burst pressure. A new multiaxial yield criterion, referred to as the average shear stress yield
(ASSY) criterion for isotropic hardening materials, is developed in this paper based on the traditional Tresca and von Mises yield criteria
so that the burst pressure of a pipeline at plastic collapse can be accurately predicted.
As an application of the proposed criterion to the plastic collapse analysis of pipelines, an ASSY-based solution for defect-free line
pipes is obtained and formulated as a function of the pipe geometry, the strain hardening exponent and the ultimate tensile stress.
Extensive experimental results are then adopted to validate the proposed solutions. Comparisons indicate that (1) the ASSY criterion can
well fit the classical experimental data of different ductile metals for both initial and subsequent plastic yielding of a material; and (2) the
ASSY-based solution of pipeline burst pressure closely matches the average experimental data for the burst pressure of defect-free pipes
for various pipeline steels.
r 2006 Elsevier Ltd. All rights reserved.

Keywords: Plastic collapse; Burst pressure; Pipeline; Material strength; Yield criterion; Tresca criterion; Mises criterion

1. Introduction remaining life, there is now a significant premium on

accuracy. Therefore, historically accepted errors in pres-
Accurate prediction of burst pressure for line pipes is sure prediction are no longer acceptable, which leads to the
crucial in the engineering design and integrity assessment present focus on improved consideration of multiaxial
of gas and oil transmission pipelines. The burst pressure is stress state effects on yield criteria so as to develop an
usually defined as the limit load or failure pressure of a pipe accurate prediction of burst pressure for line pipes.
at plastic collapse, representing the maximum load-bearing Considerable theoretical, numerical and experimental
capacity of the pipe. Understanding of burst pressure is investigations have been devoted for many years to the
very important to determine a realistic safety factor and an formulation of burst pressure predictions for pipelines, and
economic or reasonable operating pressure for pipelines. a number of analytical and empirical prediction equations
When dealing with the failure pressure prediction, the were proposed for internally pressurized defect-free pipes.
desire might be a conservative or lower bound prediction. A comparative review of these burst pressure formulae was
In contrast, predictions for integrity management now recently compiled by Law et al. [1] for thin-wall pipes, and
focus on remaining life or defect acceptance. For such by Christopher et al. [2] for thick-wall pressure vessels.
cases, prediction schemes that are conservative for pressure They concluded that there was no one prediction method
may produce nonconservative results for critical defect size. that was accurate and broadly accepted. This is not a
As small pressure errors can lead to large errors in surprise because most of them were developed for one or
several specific materials. In pipeline design codes, the
Corresponding author. burst pressure of a line pipe is simply defined when the
E-mail address: [email protected] (X.-K. Zhu). hoop stress reaches the flow stress or yield stress of the pipe

0308-0161/$ - see front matter r 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijpvp.2006.06.001
 ARTICLE IN PRESS
664 X.-K. Zhu, B.N. Leis / International Journal of Pressure Vessels and Piping 83 (2006) 663–671

steel. It is noted that the yield stress is used for working well match the average experimental data. These findings
stress design, and the flow stress is used for plastic design. imply that a more rational multiaxial yield criterion for
Recent experiments have indicated that the code criterion ductile materials may be needed for developing a reliable
may be too conservative for modern high strength pipeline prediction of burst pressure for line pipes. As pointed out
steels. Nevertheless, our previous results [3] showed that by Yu [13], more experimental results of strength of
the flow stress-based prediction by code does not guarantee materials are obtained under complex stress states, and
safety in application to modern high strength pipeline more accurate choices of strength theory are demanded.
steels. Motivated by the information above, the present paper
As early as in the 1950s, Cooper [4] and Svensson [5] proposes a new multiaxial yield criterion for isotropic
developed a theoretical solution for predicting the burst hardening materials, which is referred to as the average
pressure based on the von Mises yield criterion and the shear stress yield (ASSY) criterion. This new criterion is
plastic instability theory. Hiller [6] did similar work for thin- then used to predict and compare with classical experi-
wall tubes. However, the discussion by Clark and Wood- mental data for extensive metals for both initial and
burn attached to Cooper’s paper [4] first showed that the subsequent plastic yielding of a material. Based on the
experimental data of hoop stress for aluminium pipes at proposed criterion, a new model to predict burst pressure
the maximum pressure were appreciably lower than the of defect-free pipes is obtained as a function of the pipe
theoretical values, although they had similar trends. Kiefner geometry, the strain hardening exponent and the ultimate
et al. [7] reported a number of full-scale experimental results tensile stress (UTS). To validate the proposed model,
for extensive pipeline steels ranging from Grade B to X65 in extensive experimental data of burst pressure for various
the 1970s, including comparisons with Cooper’s predictions pipeline steels are analysed and compared with the
[4]. They found that Cooper’s prediction overestimates the theoretical solutions. Based on the proposed solution, the
ultimate strength for the thin-wall pipes they considered. influence of material hardening behaviour on failure
Likewise, Rajan et al. [8] showed that the prediction of pressure, the equivalent stress and the hoop stress of pipes
Svensson [5] overestimates their experimental data for thin- at plastic collapse are discussed.
wall tubes for structural steel AISI 4130, with the
discrepancy as high as 11% on average. 2. Multiaxial yield criteria
Cronin and Pick [9] recently examined the influence of
elastic deformation on the plastic instability of line pipes. 2.1. Three classical yield criteria
They adopted the Ramberg-Osgood material model, rather
than a pure power-law material model. Using the von Mises The Tresca criterion is the first classical yield criterion in
yield criterion, these authors obtained a theoretical solution the strength theory for isotropic ductile materials, often
in an implicit form, and found that their prediction referred to as the maximum shear stress criterion. In
generally over-predicts experimental data for instability principal stress space (s1 ; s2 ; s3 ), the Tresca criterion can be
pressure of defect-free pipes. In particular, their experi- expressed as
mental data are approximately 0.86 times their predictions

js1
s2 j js2
s3 j js1
s3 j s0
for X46 pipeline steel. Similarly based on the von Mises tmax ¼ max ; ; ¼ , (1)
criterion, Updike and Kalnins [10] developed a general 2 2 2 2
mathematical model for limit load at tensile plastic where tmax is the maximum shear stress and s0 is the yield
instability for axisymmetric thin-wall pressure vessels. They stress in tension. For convenience, the Tresca equivalent
showed that their calculated instability pressure was an stress, sT , is defined as
upper bound to the burst pressure that was achieved by test.
sT ¼ maxðjs1
s2 j; js2
s3 j; js3
s1 jÞ (2)
Therefore, it is evident from the analyses in Refs. [4–10] that
the von Mises yield criterion may only determine an upper so that the Tresca criterion in Eq. (1) can be simply written
bound to the burst pressure for a line pipe. as sT ¼ s0 .
The Tresca criterion is an alternative to the von Mises The von Mises criterion is the second classical yield
criterion to account for multiaxial effects on the plastic criterion in strength theory, often referred to as the
yielding and inelastic responses of a material. Based on octahedral shear stress criterion. It can be expressed by
these two yield criteria and the plastic instability theory, the principal stresses in the form:
Steward and Klever [11] obtained two different theoretical rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1  s0
solutions of burst pressure for defect-free pipes. They tM ¼ ðs1
s2 Þ2 þ ðs2
s3 Þ2 þ ðs3
s1 Þ2 ¼ pffiffiffi ,
found that the experimental data of burst pressure for 6 3
different ductile steels lie between predictions for the two (3)
criteria, with the von Mises prediction as an upper bound, where tM is the von Mises effective shear stress. Similarly,
and the Tresca prediction as a lower bound to the burst the von Mises equivalent stress, sM , is defined as
pressure. In fact, these results are consistent with those for rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
perfectly plastic materials (Miller [12]). Furthermore, these 1 
sM ¼ ðs1
s2 Þ2 þ ðs2
s3 Þ2 þ ðs3
s1 Þ2 (4)
authors found that the average of the two predictions can 2
 ARTICLE IN PRESS
X.-K. Zhu, B.N. Leis / International Journal of Pressure Vessels and Piping 83 (2006) 663–671 665

and the von Mises criterion in Eq. (3) is simply written as of the intermediate principal stress on the yield state, in
sM ¼ s0 . principal stress space with s1 Xs2 Xs3 , Lode introduced a
In addition to the Tresca and von Mises criteria, an stress parameter:
equivalent shear stress yield criterion proposed by Drucker 2s2
s1
s3
[14] is another but not often used yield criterion, which can m¼ . (8)
s1
s3
be expressed as
The stress difference s1
s3 can be expressed as a
1 1=2
teq ¼ ðs1
s2 Þ2 þ ðs2
s3 Þ2 þ ðs3
s1 Þ2 function of the Lode stress parameter m. Accordingly, the
3 four yield criteria introduced above can be rewritten as
" #1=6
2½ð2s1
s2
s3 Þð2s2
s3
s1 Þð2s3
s1
s2 Þ2
 1
 3
3 ðs1
s2 Þ2 þ ðs2
s3 Þ2 þ ðs3
s1 Þ2  Tresca criterion:
pffiffiffi
1=6 s1
s3
2 2 ¼ 1. (9)
¼ s0 . ð5Þ s0
3 3
Compared to Eqs. (1) and (3), Eq. (5) for the equivalent
shear stress criterion is quite complex. Therefore, it has
 von Mises criterion:
seen limited use in practical applications. s1
s3 2
¼ pffiffiffiffiffiffiffiffiffiffiffiffiffi . (10)
s0 3 þ m2
2.2. ASSY criterion
 Drucker criterion:
To construct an intermediate yield criterion between the
Tresca and von Mises criteria, an average shear stress (tA ) s1
s3 2
s0
¼ 1=6 . (11)
is defined as the average value of the maximum shear stress 3
ð3 þ m 2 Þ3
1m2 ð9
m2 Þ2
2 2
(tmax ) and the von Mises effective shear stress (tM ). Note
that the von Mises shear stress
ffi is related to the octahedral
pffiffiffiffiffiffiffi
shear stress by tM ¼ 3=2toct . As a result, the average
 ASSY criterion:
pffiffiffi
shear stress is a weighted average of the maximum shear s1
s3 2þ 3
stress and the octahedral shear stress. It is assumed that ¼ pffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffi . (12)
s0 3 þ 3 þ m2
yielding occurs if the average shear stress reaches a critical
value, namely:
" rffiffiffi # pffiffiffi Fig. 1 compares the four yield criteria, as described in
1 3 2þ 3 Eqs. (9)–(12) using the Lode stress parameter, with well-
tA ¼ tmax þ toct ¼ pffiffiffi s0 . (6)
2 2 4 3 known classical yield experimental results for various
metals developed by different researchers, including Lode
This yield criterion is referred to as the ASSY criterion [15] for steel, copper, and nickel, Ros and Eichinger [16] for
hereafter. In the uniaxial tension test, the maximum shear structural steel, Lessels and MacGregor [17] for Ni–Cr–Mo
ffiffiffi the octahedral shear stress at yielding are s0 =2
stresspand steel, Davis [18] for AISI 1023 steel, Marin et al. [19] for
and 2s0 =3, respectively,
 pand
ffiffiffi therefore
pffiffiffi the average shear 75s-T6 aluminium alloy, Marin and Hu [20] for mild steel,
stress at yielding is 2 þ 3 s0 =4 3 as evident in Eq. (6). Naghdi et al. [21] for 2024-T4 aluminium, and Maxey [22]
Using the principal stresses with the assumption of for X52 and X60 pipeline steels. The same experimental
s1 Xs2 Xs3 , from Eqs. (2), (4) and (6), the ASSY data used in Fig. 1 are plotted in the principal stress s1 s2
equivalent stress, sA , is defined as plane (where s3 ¼ 0) in Fig. 2. From these two figures, it
1 hpffiffiffi i can be seen that (1) the ASSY criterion is very close to the
sA ¼ pffiffiffi 3sT þ 2sM . (7) Drucker criterion, and both criteria lie between the Tresca
2þ 3
and von Mises criteria; (2) the von Mises criterion provides
Eq. (7) indicates that the ASSY effective stress is a an upper bound for all experimental data considered, while
weighted average of the Tresca effective stress and the von the Tresca criterion provides a conservative lower bound;
Mises effective stress. For simple tension, the ASSY yield (3) the ASSY criterion provides an average result or a
criterion in Eq. (6) can be simplified as sA ¼ s0 . reasonable lower bound for these experimental data. At the
stress state with the intermediate principal stress being the
2.3. Experimental validation for initial yielding average of the maximum and minimum stresses or m ¼ 0,
such as for an end-capped pressurized pipe, the ASSY and
The simplest specimens used for validating multiaxial Drucker criteria can equally represent the average of
yield criteria are thin-wall tubes, as adopted by Lode [15], experimental data. Comparing Eqs. (5) and (6), or (11) and
who tested tube specimens for steel, copper, and nickel (12), the ASSY criterion is much simpler than the Drucker
under various combined loadings of longitudinal tension criterion, but both criteria can provide similar results.
and internal hydrostatic pressure. To examine the influence Therefore, the complex Drucker criterion will not be
 ARTICLE IN PRESS
666 X.-K. Zhu, B.N. Leis / International Journal of Pressure Vessels and Piping 83 (2006) 663–671

1.3

1.2 Drucker criterion ASSY criterion

Mises criterion

1.1
(σ1-σ3) /σ0

Tresca criterion
1

Steel, Lode [15] AISI 1023 steel, Davis [18]

Copper, Lode [15] Al-755-T6 alloy, Martin [19]
0.9
Nickel, Lode [15] Mild steel, Marin and Hu [20]
Steel, Ros and Eichinger [16] 2024-T4 Al, Naghdi [21]
Ni-Cr-Mo steel, Lessels [17] X52 and X60, Maxey [22]
0.8
-1 -0.5 0 0.5 1
Lode stress parameter, µ

Fig. 1. Comparison of the four yield criteria and classical yield experimental data in reference to Lode stress parameter.

1.5

σ2

1
steel, Lode [15]
ASSY yield locus
copper, Lode [15]
0.5 nickel, Lode [15]

steel, Ros & Eichinger [16]

NiCrMo steel, Lessels [17]

0
-1.5 -1 -0.5 0 0.5 1 AISI 1023 steel, Davis [18]
σ1 1.5
Al-75s-T6 alloy, Martin [19]

-0.5 Mild steel, Marin & Hu [20]

2024-T4 Al, Naghdi [21]

Tresca hexagon yield locus
X52 and X60, Maxey [22]
-1
Mises ellipse yield locus

-1.5

Fig. 2. Comparison of the three yield loci and yield experimental data in principal stress s1 s2 plane.

discussed further. Instead, we focus on the ASSY criterion equivalent strain:

and its applications to plastic collapse analysis of pipelines.

1 1 1 1

A ¼ ð
T þ
M Þ ¼ g þ pffiffiffi goct , (13)
2.4. Experimental validation for subsequent yielding 2 2 2 max 2
where the maximum shear strain and the octahedral shear
To analyse the subsequent plastic yielding of a multiaxial strain are expressed by
stress state for isotropic hardening materials, an equivalent
strain must be defined in accordance with its corresponding gmax ¼
1

3 (14)
equivalent stress for a yield criterion. As evident from and
Fig. 2, the yield surface determined by the ASSY criterion qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2
meets the condition of convexity as required by the theory goct ¼ ð
1

2 Þ2 þ ð
2

3 Þ2 þ ð
3

1 Þ2 . (15)
of plasticity [23]. According to the definition of the average 3
shear stress, an ASSY equivalent strain is defined as the Osgood [24] conducted a set of excellent experimental
average of the Tresca equivalent strain and the von Mises tests, and obtained biaxial stress–strain curves using five
 ARTICLE IN PRESS
X.-K. Zhu, B.N. Leis / International Journal of Pressure Vessels and Piping 83 (2006) 663–671 667

24s-T aluminium alloy tubes with closed ends for different improved correlations, and (3) the equivalent average shear
stress ratios. All test tubes with a wall thickness of 0.05 in stress–strain curve from the ASSY criterion provides the
(1.27 mm) and internal diameter of 1.75 in (44.45 mm) were best correlations for all five tests. It is noted that Fig. 3 only
loaded by combined internal pressure and axial tension addresses aluminium. Recently, a similar comparison for
with five nominal ratios of hoop stress to axial stress, mild steel has been done by the present authors [25], and
namely, p ¼ 0.0, 0.5, 1.0, 2.0, and N (infinite). The results the same results were drawn. Therefore, we can conclude
of the five tests were then presented graphically in two that the ASSY criterion can well characterize initial and
forms of stress–strain curves. The first form was plotted as subsequent plastic yielding for multiaxial stress states. An
the true maximum shear stress tmax against the maximum application of the ASSY criterion to plastic collapse of
natural shear strain gmax , as shown in Fig. 3(a); the second pipelines is considered in the following sections.
form was plotted as the true octahedral shear stress toct
against the natural octahedral shear strain goct , as shown in 3. Application of ASSY criterion to plastic collapse of
Fig. 3(b). Using the test data in Figs. 3(a) and (b), the true pipelines
ASSY shear stress tave against the natural ASSY equivalent
strain
ave are determined from Eqs. (6) and (13), and 3.1. Stress–strain law
plotted in Fig. 3(c) for the five tests. In these figures, the
lower scale along the abscissas refers to the lower curves, At plastic collapse, a ductile pipeline usually experiences
and the upper scale refers to the upper curves. large plastic deformation. Therefore, the stress–strain
Fig. 3 shows that (1) the maximum shear stress–strain relation of materials can be reasonably characterized by a
curve from the Tresca criterion can roughly correlate pure power-law curve:
biaxial stress–strain relations, (2) the octahedral shear  e n
stress–strain curve from the von Mises criterion provides s ¼ K
n ; K ¼ s0uts , (16)
n

40000 40000
Octahedral shear stress, τtoc, psi
Maximum shear stress, τmax, psi

30000 30000

20000 20000
p=0 p=0
p = 0.5 p = 0.5
p=1 p=1
10000 p = 2.06 10000 p = 2.06
p = infinite p = infinite

0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18
0 0
0.000 0.002 0.004 0.006 0.008 0.010 0.012 0.014 0.016 0.018 0.000 0.002 0.004 0.006 0.008 0.010 0.012 0.014 0.016 0.018

(a) Maximum shear strain, γmax (b) Octahedral shear strain, γtoc
Equivalent average shear stress, τave, psi

40000

30000

p=0
20000
p = 0.5
p=1
p = 2.06
10000
p = infinite

0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 0.20
0
0.000 0.002 0.004 0.006 0.008 0.010 0.012 0.014 0.016 0.018 0.020
(c) Equivalent average strain, 2εave

Fig. 3. Experimental results of biaxial plastic stress–strain relations for five loading cases: (a) maximum shear stress versus maximum shear strain, (b)
octahedral shear stress versus octahedral shear strain, (c) equivalent average shear stress versus equivalent average strain.
 ARTICLE IN PRESS
668 X.-K. Zhu, B.N. Leis / International Journal of Pressure Vessels and Piping 83 (2006) 663–671

where s and e are the uniaxial true stress and the true strain value by
in simple tension, respectively. e ¼ 2:7183, n is the strain
D D0 ðð4pffiffi3=2þpffiffi3Þ
A Þ
hardening exponent, and s0uts is the engineering UTS. It is ¼ e . (23)
t t0
noted that the true UTS is related to the engineering UTS
in the form of suts ¼ en s0uts , and the true strain at the From Eqs. (16)–(23), the applied pressure is obtained as
ultimate stress is equal to n. In most available material a function of the pipe geometry, the material property and
databases, the yield stress (Y) and the UTS (T) are the ASSY equivalent strain:
available, but the strain hardening exponent n is not often

1 1 2t0
ð4pffiffi3=ð2þpffiffi3Þ
A Þ  e n 0 n
reported. In such cases, the yield-to-tensile stress ratio, i.e., P¼ þ pffiffiffi e suts
A . (24)
Y/T is determined, and the hardening exponent n can be 2 3 D0 n
estimated using the following approximate expression [3] Application of the plastic instability condition of
for pipeline steels: qP=q
A ¼ 0 to Eq. p (24) yields the limit ASSY equivalent
ffiffiffi p ffiffiffi

0:596 strain as
A ¼ ð2 þ 3=4 3Þn. Therefore, the ASSY-based
1
n ¼ 0:239
1 , (17) bust pressure of the line pipe at plastic collapse is
Y =T determined as a function of the original pipe geometry,
where the yield stress is defined at 0.5% total strain. the UTS and the material hardening exponent:

pffiffiffinþ1
2þ 3 4t0 0
3.2. ASSY-based solution PA ¼ p ffiffiffi s . (25)
b
4 3 D0 uts
It is herein focused on a long thin-wall defect-free pipe
with capped ends subjected to internal pressure. According
to stress equilibrium, the three principal stresses in the pipe 3.3. General form of three theoretical solutions
are obtained as
PD PD Follow similar procedures to that above, the Tresca and
s1 ¼ syy ¼ ; s2 ¼ saa ¼ ; s3 ¼ srr  0, (18) von Mises based solutions for the burst pressure of
2t 4t
pipelines can be determined, as detailed in Refs. [26,27].
where the coordinates (y, r, a) denotes the hoop, radial and
These theoretical solutions for burst pressure Pb for the
axial directions, respectively. D is the instant average
three yield criteria can be simply expressed in the following
diameter, and t is the instant wall thickness of the pipe. For
general form:
a long earth-anchored line pipe, the axial strain is usually
very small, and can be neglected. For a finite strain
nþ1
k 4t0 0
analysis, three principal strains can be expressed as Pb ¼ s , (26)
2 D0 uts
D t

1 ¼
yy ¼ ln ;
2 ¼
aa  0;
3 ¼
rr ¼ ln , (19) where k is a yield criterion dependent constant:
D0 t0 8
> 1 for Tresca criterion;
where Do is the original average diameter, and to is the < pffiffiffi
original wall thickness of the pipe. For large plastic k ¼ 2= 3 for Mises criterion; (27)
: 1=2 þ 1=pffiffi3ffi for ASSC criterion:
>
deformation, the volume incompressibility requires
1 þ

2 þ
3  0. From Eq. (19), one has

 Fig. 4 shows the variation of burst pressure determined
D t0 from Eq. (26) with the strain hardening exponent n for the

1

3 ¼ 2
1 ¼ ln . (20)
t D0 Tresca, von Mises and ASSY criteria, respectively. In this
figure, the burst pressure is normalized by 2t0 s0uts =D0 , a
From Eqs. (2), (4), (7) and (18), the ASSY equivalent
nominal critical pressure that can result in the nominal
stress in the pipe is determined as
pffiffiffi hoop stress equal to s0uts . It is seen from this figure that the
1 pffiffiffi 2 3 normalized burst pressures for the three yield criteria have
sA ¼ pffiffiffi ð 3sT þ 2sM Þ ¼ pffiffiffi s1 , (21) similar trends with n, and all burst pressures increase as n
2þ 3 2þ 3
decreases. However, significant differences are observed
From Eqs. (13), (19) and (20), the ASSY equivalent among these solutions—most notably, the burst pressure
strain is obtained as based on the ASSY criterion lies between the von Mises
pffiffiffi and Tresca solutions.
1 2þ 3

A ¼ ð
T þ
M Þ ¼ pffiffiffi
1 . (22) At the burst pressure, the corresponding equivalent
2 2 3
stress, hoop stress, equivalent strain and hoop strain for the
From Eqs. (21) and (22), one has sA
A ¼ s1
1 ¼ s
, three yield criteria were obtained as follows:
which satisfies Hill’s plastic work assumption [23].
n
Eqs. (20) and (22) lead to the ratio of current average k
ðse Þb ¼ suts , (28)
diameter to wall thickness with respect to its original 2
 ARTICLE IN PRESS
X.-K. Zhu, B.N. Leis / International Journal of Pressure Vessels and Piping 83 (2006) 663–671 669

nþ1
k Comparisons of Eq. (32) with Eqs. (28) and (29) indicate
ðsyy Þb ¼ 2 suts , (29) that at the limit state of plastic collapse, the ratio of the
2
nominal equivalent (or hoop) stress to the engineering UTS
nk is identical to that based on true stress values. Fig. 5 shows
ð
e Þ b ¼ , (30) the variations of true effective stress and true hoop stress,
2
normalized by the true UTS, with the strain hardening
n exponent n for the three yield criteria. In this figure,
ð
yy Þb ¼ , (31) syy =suts was used as the first (left) title of the y-axis, and
2
se =suts was used as the second (right) title of the y-axis. It is
where suts ¼ en s0uts is used. It is interesting that for all three
evident from this figure that the yield criterion has a
yield criteria, the hoop strain of the pipe at plastic collapse
significant effect on the hoop stresses, and a relatively
has the same value of n/2.
smaller effect on the equivalent stresses. The ASSY-based
For convenience, nominal stresses are often used in
solutions for the equivalent stress and the hoop stress are in
engineering practice. The nominal equivalent stress and the
the middle of those from the Tresca and von Mises criteria.
nominal hoop stress at the burst pressure for the three
All normalized stresses in Fig. 5 decrease with increased n.
criteria can be expressed by
Due to the observed dependence on n, therefore, the

n
1 k nominal hoop stress based criterion of ðsyy Þ0 ¼ s0uts is
ðse Þ0b ¼ ðsyy Þ0b ¼ s0uts . (32) inaccurate generally for determination of burst pressure for
k 2
pipeline steels although it may be correct for some cases.

1.2 4. Full-size experimental validation

Mises criterion

1.1 ASSY criterion

Tresca criterion
Actual failure pressures determined in full-size tests of
pipelines are used in this section to validate the proposed
1.0
Pb/(2t0σ'uts/D0)

theoretical models. In the burst pressure tests, short pipes

with capped ends were usually used as the full-size test
0.9
specimens. Experiments indicated that the axial strain for
0.8 short pipes were very small when the pipe length was larger
than twice the pipe diameter, and thus it can be ignored.
0.7 Therefore, the theoretical solutions based on the long pipe
assumption with zero axial strain can be applied to the
0.6 short test pipes. Fig. 6 compares the Tresca, von Mises and
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 ASSY-based solutions with various full-size experimental
n
data of burst pressure for defect-free pipes. In this figure,
Fig. 4. Variation of burst pressure with strain hardening exponent for the 103 full-size tests were involved for various pipeline
Tresca, von Mises, and ASSY criteria. steels, including 41 tests by Kiefner et al. [7] for pipeline

1.5 1.0
Mises criterion
ASSY criterion
Tresca criterion
1.3 0.9

σe / σuts
σθθ / σuts

σe / σuts

1.1 0.8

0.9 σθθ / σuts 0.7

0.7 0.6
0.0 0.1 0.2 0.3 0.4
n

Fig. 5. Variation of equivalent stress and hoop stress at burst pressure with strain hardening exponent for Tresca, von Mises, and ASSY criteria.
 ARTICLE IN PRESS
670 X.-K. Zhu, B.N. Leis / International Journal of Pressure Vessels and Piping 83 (2006) 663–671

1.2

1.1

1.0

0.9
Pb/(2t0σ'uts/D0)

0.8 Mises based solution Maxey ([28], X70)

ASSY based solution Mok et al. ([29], X60)
0.7
Tresca based solution Chouchaoui ([30], X46)
Kiefner et al. ([7], Gr.B-X65) Amano et al. ([31], X65,X70)
0.6
Papka et al. ([35], X120) Hillenbrand et al. ([32], X100)

0.5 Law et al ([36], X65-X80) Liessem et al. ([33], X60-X80)

Paslay et al. ([37],H40-Q125) Okaguchi et al. ([34], X80, X120)
0.4
0.00 0.05 0.10 0.15 0.20
n

Fig. 6. Comparisons of full-size experimental data and Tresca, von Mises and ASSY-based solutions to burst pressure for various pipeline steels.

steel grades ranging from Grade B to X65, three tests by plastic collapse analysis of pipelines was performed in this
Maxey [28] for X70, two tests by Mok et al. [29] for X60, work. A theoretical solution to predict burst pressure of
six tests by Chouchaoui [30] for X46, 12 tests by Amano et defect-free pipes was then developed using the ASSY yield
al. [31] for X65 and X70, two tests by Hillenbrand et al. [32] criterion, and discussed by comparing with the solutions
for X100, eight tests by Liessem et al. [33] for X60, X65, determined by the Tresca and von Mises criteria. Numer-
X70, and X80, two tests by Okaguchi et al. [34] for X80 and ical and experimental results were used to validate the
X120, nine tests by Papka et al. [35] for X120, five tests proposed models. The primary results include:
from Law et al. [36] for X60 and X80, and 13 tests by
Paslay et al. [37] for H40-Q125 casting pipe steels. For (1) The proposed ASSY criterion is similar to the Drucker
most of the tests, the strain hardening exponent n is equivalent shear stress criterion, but it is much simpler
estimated from Eq. (17) using the Y/T ratio. and easier to be used. Extensive comparisons indicated
Fig. 6 demonstrates that the ASSY-based solution can that the ASSY criterion can well fit classical experi-
well correlate the test data and closely match the average mental data for initial and subsequent plastic yielding
value of experimental burst pressure for all pipeline steels for various ductile metals.
considered, including traditional low strength and modern (2) Three theoretical solutions for the burst pressure, the
high strength pipeline steels. Fig. 6 also shows the von equivalent stress and the hoop stress for defect-free
Mises-based solution serves as an upper bound to the pipes based on the Tresca, von Mises and ASSY criteria
experimental data, while the Tresca-based solution serves have similar trends with the strain hardening exponent.
as a lower bound. In this figure, several test points are However, significant differences between these theories
outside the upper and lower bounds. This may be caused can be observed.
by inaccuracy of material property and experimental (3) The experimental results confirmed that the von Mises
measurement. Therefore, one can conclude that the criterion generally overestimates burst pressure, and
proposed ASSY criterion is an effective yield criterion for provides an upper bound to the burst pressure; while
the plastic collapse analysis of defect-free pipes, and can be the Tresca criterion underestimates burst pressure, and
used as a reliable tool to predict burst pressure or to determines a lower bound to the burst pressure.
perform plastic collapse analysis for defect-free line pipes (4) At plastic collapse, the nominal equivalent (or hoop)
in the engineering design and integrity assessment of stress normalized by the engineering UTS is identical to
pipelines. its true value for all three yield criteria, and increases
with decreasing n. In general, both the nominal
5. Summary and conclusion equivalent and hoop stresses do not equal the
engineering UTS, and therefore these nominal stress-
A new multiaxial yield criterion, i.e. ASSY, based on the based criteria are inaccurate in prediction of burst
average shear stress was proposed, and its application to pressure of defect-free pipes.
 ARTICLE IN PRESS
X.-K. Zhu, B.N. Leis / International Journal of Pressure Vessels and Piping 83 (2006) 663–671 671

Acknowledgements [19] Martin J, Ulrich BH, Hughes WP. Plastic stress–strain relations for
75s-T6 aluminum alloy subjected to biaxial tensile stresses, NACA
technical note 2425; August 1951.
The support of the US Department of Transportation
[20] Marin J, Hu LW. Biaxial plastic stress–strain relations of a mild steel
through Broad-Agency Announcement funding is grate- for variable stress ratios. J Appl Mech, Trans ASME 1956;78:
fully acknowledged, as is the support of Battelle’s Pipeline 499–509.
Technology Center. Special thanks go to the anonymous [21] Naghdi PM, Essenburg F, Koff W. An experimental study of initial
reviewers for their useful comments and suggestions for and subsequent yield surface in plasticity. J Appl Mech 1958;25:
this paper. 201–9.
[22] Maxey WA. Measurement of yield strength in the mill expander. In:
Proceedings of the fifth symposium on line pipe research, Houston,
References TX, November 20–22, 1974; p. N1–N32.
[23] Hill R. The mathematical theory of plasticity. Ely House, London:
[1] Law M, Bowie G, Fletcher L. Burst pressure and failure strain in Oxford University Press; 1950.
pipeline, Part 2: comparison of burst pressure and failure-strain [24] Osgood WR. Combined-stress tests on 24s-t aluminum-alloy tubes.
formulas. J Pipeline Integrity 2004;3:102–6. J Appl Mech, Trans ASME 1947;69:A147–53.
[2] Christopher T, Rama Sarma BSV, Potti PKG, Rao BN, Sankarnar- [25] Zhu XK, Leis BN. Theoretical and numerical predictions of burst
ayanasamy K. A comparative study on failure pressure estimations of pressure of pipelines. In: Proceedings of 2006 ASME PVP conference,
unflawed cylindrical vessels. Int J Pressure Vessels Piping 2002;79: Vancouver, BC, Canada; July 23–27, 2006.
53–66. [26] Zhu XK, Leis BN. Analytic prediction of plastic collapse failure
[3] Zhu XK, Leis BN. Influence of yield-to-tensile strength ratio on pressure of line pipes. In: Proceedings of 2005 ASME PVP
failure assessment of corroded pipelines. J Pressure Vessel Technol conference, Denver, CO, USA; July 17–21, 2005.
2005;127:436–42. [27] Zhu XK, Leis BN. Strength criteria and analytic predictions of failure
[4] Cooper WE. The significance of the tensile test to pressure vessel pressures in line pipes. Int J Offshore Polar Eng 2004;14:125–31.
design. Weld J—Weld Res Suppl, January 1957, p. 49s–56s. [28] Maxey WA. Y/T significance in line pipe. In: Proceedings of the
[5] Svensson NL. Bursting pressure of cylindrical and spherical vessels. seventh symposium on line pipe research. Houston, TX, October
J Appl Mech 1958;25:89–96. 14–16, 1986; p. 91–918.
[6] Hiller MJ. Tensile plastic instability of thin tubes. J Mech Eng Sci [29] Mok DRB, Pick RJ, Glover AG, Hoff R. Bursting of line pipe with
1962;4:251–6. long external corrosion. Int J Pressure Vessels Piping 1991;46:
[7] Kiefner JF, Maxey WA, Duffy AR. The significance of the yield-to- 1216–952.
ultimate strength ratio of line pipe materials. Summary report to [30] Chouchaoui BA. Evaluating the remaining strength of corroded
Pipeline Research Committee, American Gas Association, March pipelines. PhD thesis, Department of Mechanical Engineering,
1971. University of Waterloo, Canada; 1993.
[8] Rajan KM, Deshpande PU, Narasimhan K. Experimental studies on [31] Amano K, Matsuoka M, Ishihara T, Tanaka K, Inoue T, Kawaguchi
bursting pressure of thin-walled flow formed pressure vessels. J Mater Y, Tsukamoto M. Significance of yield ratio limitation to plastic
Process Technol 2002;125–126:228–34. deformation of pipeline in high pressure proof test. In: Proceedings of
[9] Cronin DS, Pick RJ. Prediction of the failure pressure for complex the seventh symposium on line pipe research, Houston, TX, USA,
corrosion defects. Int J Pressure Vessels Piping 2002;79:279–87. October 14–16, 1986; p. 81–821.
[10] Updike DP, Kalnins A. Tensile plastic instability of axisymmetric [32] Hillenbrand HG, Liessem A, Knauf G, Niederhoff K, Bauer J.
pressure vessels. J Pressure Vessel Technol 1998;120:6–11. Development of large-diameter pipe in grade X100. In: Proceedings
[11] Stewart G, Klever FJ. An analytical model to predict the burst of the third international conference of pipeline technology, Brugge,
capacity of pipelines. In: Proceedings of international conference of Belgium, 2000; p. 1263–81.
offshore mechanics and arctic engineering, vol. V, Pipeline technol- [33] Liessem A, Grdef, MK, Knauf G, Marewski, U. Influence of thermal
ogy, 1994; p. 177–188. treatment on mechanical properties on UOE linepipe. In: Proceedings
[12] Miller AG. Review of limit loads of structures containing defects. Int of the fourth international conference of pipeline technology, Ostend,
J Pressure Vessels Piping 1988;32:197–327. Belgium, 2004; p. 1263–81.
[13] Yu MH. Advances in strength theories for materials under complex [34] Okaguchi S, Makino H, Hamada M, Yamamoto A, et al.
stress state in the 20th century. Appl Mech Rev 2002;55:169–218. Development and mechanical properties of X120 linepipe. Int
[14] Drucker DC. Relation of experiments to mathematical theories of J Offshore Polar Eng 2004;14:28–35.
plasticity. J Appl Mech, Trans ASME 1949;71:349–57. [35] Papka SD, Stevens JH, Macia DP, Fairchild DP, Petersen CW. Full-
[15] Lode W. Versuche ueber den einfluss der mittleren Hauptspannung size testing and analysis of X120 linepipe. Int J Offshore Polar Eng
auf das fliessen der metalle eisen, kupfer, und nickel. Zeits Physik 2004;14:42–51.
1926;36:913–39. [36] Law M, Bowie G, Fletcher L. Pipeline behaviour, the hydrostatic
[16] Ros M, Eichinger A. Versuche zur klaerung der frage der bruchefahr strength test and failure strain. In: Proceedings of the 15th conference
III, mettalle, eidgenoss. Material pruf. Und Versuchsantalt Industriell of PRCI and EPRG pipeline conference, Orlando, FL, May 16–20,
Bauwerk und Geerbe, Diskussionsbericht 1929;34:3–59. 2005.
[17] Lessels JM, MacGregor CW. Combined stress experiments on a [37] Paslay P, Cernocky E, Wink R. Burst pressure prediction on thin-
nickel–chrome–molybdenum steel. J Franklin Inst 1940;230:163–80. walled, ductile tubulars subjected to axial load. In: Proceedings of
[18] Davis EA. Yielding and fracture of medium-carbon steels under applied technology workshop on risk based design of well cashing
combined stresses. J Appl Mech, Trans ASME 1945;67:A13–24. and tubing, Woodlands, TX, USA; 1998.