International Journal of Pressure Vessels and Piping 88 (2011) 507e517

Contents lists available at ScienceDirect

International Journal of Pressure Vessels and Piping

journal homepage: www.elsevier.com/locate/ijpvp

Proposal of reference stress for a surface flaw on a cylindrical component from

a review-with-comparison of the local metal loss assessment rule between API
579-1 and the p-M diagram method
Kenji Oyamada a, *, Shinji Konosu b, Takashi Ohno c
a
The High Pressure Gas Safety Institute of Japan/Graduate School of Ibaraki University, 4-3-13, Toranomon, Minato-ku, Tokyo 105-8447, Japan
b
Ibaraki University, 4-12-1, Nakanarusawa, Hitachi, Ibaraki 316-8511, Japan
c
The High Pressure Gas Safety Institute of Japan, 2-16-4, Tadao, Machida-shi, Tokyo 194-0035, Japan

a r t i c l e i n f o a b s t r a c t

Article history: The Remaining Strength Factor (RSF) approach in Part 5 of API 579-1/ASME FFS-1 is an assessment
Received 17 December 2009 method for a cylindrical component with a local metal loss based on surface correction factors. Also,
Received in revised form reference stress solutions that are applied in the Failure Assessment Diagram (FAD) method for a cylin-
13 July 2011
drical component with a crack-like flaw are provided in Annex D using surface correction factors. In the
Accepted 25 July 2011
recently-developed p-M diagram method, the reference stress solution for local metal loss evaluation in
a cylindrical component is derived using bulging factors, which are similar but not identical to the
Keywords:
surface correction factors used in API 579-1/ASME FFS-1. This paper describes the results of a compar-
Reference stress solution
The p-M diagram method
ative study among the RSF approach, reference stress solutions for the FAD method, and the p-M diagram
API 579-1/ASME FFS-1 method, in terms of plastic collapse evaluation of a cylindrical component. These results were compared
Plastic collapse with the FEA and experimental results to confirm how these estimated stresses could be validated. This
Local metal loss study also involves recommended reference stress solutions for a cylindrical component with a crack-like
Crack-like flaw flaw or a local metal loss, which should be adopted as fitness-for-service rules, and a discussion on the
influence of the design margin of the construction code on allowable flaw depth.
Ó 2011 Elsevier Ltd. All rights reserved.

1. Introduction
The p-M diagram method is a fitness-for-service rule to evaluate
a local metal loss subjected to internal pressure and external
bending moment, which has been developed by Konosu, one of the
authors, and adopted as in the Japanese Ibaraki FFS rule [1]. The
p-M diagram method takes advantage of reference stress solutions
for crack-like flaws. The validity of the application of such reference
stresses to the assessment of a local metal loss is ascertained in
references [2,3].
Derivation of the reference stress solutions given in the p-M
diagram method is made clear and those solutions are considered
to be a lower bound to the plastic limit loads. The plastic limit load
indicates a load where appreciable plastic deformation (almost
equal to general yielding and not necessarily the physical collapse
load of the vessel) occurs at a local metal loss area, determined by
applying twice-elastic slope (TES) as recommended by ASME [4]. It
was clarified that there is good agreement between the predicted
* Corresponding author. Tel.: þ81 3 3436 1704; fax: þ81 3 3436 5704.
E-mail address: oyamada@khk.or.jp (K. Oyamada).
limit loads and finite element analysis (FEA) and experimental
results [3].
On the other hand, the RSF approach for a local metal loss in Part
5 of API 579-1/ASME FFS-1 [5] is based on surface correction factors
[6] and physical collapse loads in whole cylinders. So, the evalua-
tions count on margins from the plastic instability loads (physical
collapse loads of components and not from the plastic limit loads at
a local metal loss area) and consequently, the loads obtained by the
RSF approach results in far lower stress levels than the general
yielding at a local metal loss area in some cases due to taking a larger
safety factor into account. The plastic instability loads are strongly
dependent on not only flaw dimensions but also on cylinder sizes,
strain-hardening properties and so on. However, adequate levels of
integrity and safety would be assured when margins from the
plastic limit load (general yielding) at a local metal loss area are
used. Because keeping stress levels at the local metal loss area below
the plastic limit load (general yielding) can preclude secondary
damage such as ratcheting due to general yielding of the local metal
loss area, which is detailed in reference [7].
A comparative study between API 579-1/ASME FFS-1 and the
p-M method is introduced in this paper. This shows that the
RSF approach in the level 1 and 2 assessments of Part 5 of

0308-0161/$ e see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijpvp.2011.07.008
508 K. Oyamada et al. / International Journal of Pressure Vessels and Piping 88 (2011) 507e517

Nomenclature a Relative rectangular flaw depth equivalent to semi-

elliptical flaw depth
a Flaw Depth sb Circumferential bending stress component without
aeff Rectangular flaw depth equivalent to a half elliptical a flaw
flaw depth sf Flow stress of a material
cL A half of longitudinal flaw length sref Reference stress
cq A half of circumferential flaw length sref API Reference stress for a local metal loss in a cylindrical
C Factor to define a model for the cross section in API component in API 579-1/ASME FFS-1
579-1/ASME FFS-1 sref API Crack Reference stress for a crack-like flaw in a cylindrical
g Modified factor for bending effect component in API 579-1/ASME FFS-1
Lr Ratio of a reference stress against the yield strength of sref pM Reference stress for a flaw in a cylindrical component
a material in the p-M diagram method
LDC The limit or plastic collapse load of the damaged sref plate Reference stress for a crack-like flaw in a plate
component (component with flaws) sys Yield strength of a material
LUC The limit or plastic collapse load of the undamaged suts Ultimate tensile strength of a material
component sq Circumferential membrane stress without a flaw
MSKiefner Surface correction factor for a surface flaw l Shell parameter for a through-wall flaw
MSChell Bulging factor for a surface flaw la Shell parameter for a partly through-wall flaw
Mt Surface correction factor for a through-wall flaw lc Shell parameter for the circumferential extent of a local
Mt ðlÞ Bulging factor for a through-wall flaw metal loss for a partly through-wall flaw
Mt ðla Þ Bulging factor for a partly through-wall flaw
MSC Surface correction factor for the circumferential extent Acronyms
of a local metal loss API American Petroleum Institute
MtC Surface correction factor for the circumferential extent ASME American Society of Mechanical Engineers
of a through-wall flaw FEA Finite Element Analysis
Ri Inner radius of a cylindrical component FFS Fitness-for-service
t Wall thickness RSF Remaining Strength Factor
RSFa Allowable Remaining Strength Factor

API 579-1/ASME FFS-1 poses the possibility of giving either less longitudinal extent of a local metal loss on the surface of a cylin-
conservative or over-conservative results under certain conditions drical component.
from the viewpoint of the lower bound of the plastic collapse loads LDC
computed by the p-M method. The same conclusions will be RSF ¼ (1)
LUC
derived for the level 1 and 2 assessments of Part 4 in API 579-1/
ASME FFS-1 because the same philosophy applies as in Part 5. This Where,
is the first topic for discussion in this paper. LDC : The limit or plastic collapse load of the damaged compo-
In consideration of an assessment for a local metal loss and that nent (component with flaws).
for a crack-like flaw, stress concentration factors may vary with LUC : The limit or plastic collapse load of the undamaged
each geometric configuration of a flaw. However, the net section component.
stress state is almost the same where materials behave in a full Kiefner et al. [12] obtained experimentally a surface correction
plastic manner. Therefore, in terms of the plastic collapse of factor MSKiefner for high strength steels (API X52-X100) from the
a component assumed as an elastic perfectly-plastic material, the physical collapse loads by using the following equation. Note that
reference stress solution for a crack-like flaw is considered to be the strain-hardening properties of high strength steels differ
applicable sufficiently to that for a local metal loss, and vice-versa. greatly from those of the usual low strength carbon steels.



The plastic limit load is needed to assess the flawed component in A 1
terms of plastic collapse, which is usually evaluated by 1
A0 Mt
Lr ¼ ðsref =sys Þ factor which means the ratio of the reference stress MSKiefner ¼

A
against the yield strength of the material. Then, a comparative 1
A0
study among the reference stresses given in Annex D that are
or
applied to the FAD approach in Part 9 of API 579-1/ASME FFS-1,
a
1 
those in the p-M diagram method, and those in some of the other 1C
available references such as BS7910 [8], French RSE-M/RCC-MR t Mt
MSKiefner ¼ a (2)
codes [9], Staat and Vu [10], and Lei [11], is conducted. This is the 1C
second topic for discussion on this paper. t
Where,
2. Comparison of assessments for the longitudinal extent for A: Area of metal loss
a local metal loss or a longitudinal crack-like flaw of a cylin- A0 :Original metal area
drical component subjected to pressure between API 579-1/ C: Factor to define a model for the cross section in API 579-1/
ASME FFS-1 and the p-M diagram method ASME FFS-1 as C ¼ 0:85 to provide an optimum fit to the
experimental data and C ¼ 1:0 to correspond to a rectan-
In the level 1 and 2 assessments of Part 5 of API 579-1/ASME gular model
FFS-1, the Remaining Strength Factor (RSF) is defined to assess the a: Flaw Depth
 K. Oyamada et al. / International Journal of Pressure Vessels and Piping 88 (2011) 507e517 509

t: Wall thickness. aeff : Rectangular flaw depth equivalent to half the elliptical flaw
Mt : Surface correction factor for a through-wall flaw, which is depth.
given in the API 579-1/ASME FFS-1 [5] as follows: MSChell : Bulging factor for surface flaws proposed by Chell [16],
Mt ¼ 1:001  0:014195l þ 0:2909l  0:09642l þ 0:02089l 
2 3 4 which is given by
0:003054l þ 2:957ð104 Þl  1:8462ð105 Þl þ 7:1553ð107 Þl 
5 6 7 8
1
9 10 MSChell ¼
 (8)
1:5631ð108 Þl þ 1:4656ð1010 Þl for local metal loss assessment a a 1
1 þ
in Part 5. t t Mt ðla Þ
!0:5
1:02 þ 0:4411l þ 0:006124l
2 4
Mt ¼ for the reference Mt ðla Þ: Bulging factor that is computed Mt ðlÞ with l ¼ la .
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1:0 þ 0:02642l þ 1:533ð106 Þl
2 4
2
stress of crack-like flaws in Part 9. Mt ðlÞ ¼ 1 þ 0:317l : Bulging factor for a through-wall flaw
It is noted that those above two equations provide almost the originally defined by Folias [13].
same value of Mt forl  9. 1:818c
la ¼ pffiffiffiffiffiffiffi L : Shell parameter which is computed l with t ¼ a.
1:818c Ri a
l ¼ pffiffiffiffiffiffi L : Shell parameter defined by Folias [13]. The validity of the use of Eq. (6) is ascertained in reference [17].
Ri t
cL : A half of longitudinal flaw length. Also, the modification for the case in which a flaw depth to thick-
Ri : Inner radius of a cylindrical component. ness ratio is extremely large, e.g. a/t ¼ 0.95, is proposed in refer-
In API 579-1/ASME FFS-1, RSF in Eq. (1) is computed with ences [18,19]. The applications of p-M diagram method to
Kiefner overlapped or multiple flaws are respectively proposed in refer-
MS as the RSF approach in Eq. (3).
ences [19,20]. Eq. (6) is based on both the load equilibrium and the
1 moment equilibrium for a semi-elliptical crack-like flaw in a plate,
RSF ¼ (3) which is subjected to through-wall membrane and bending stress
MSKiefner
under rotation (bending) restraint condition, and modified to
The surface correction factor means an increase of the circum- incorporate the bulging effect due to the geometry of a cylindrical
ferential membrane stress of a cylindrical component due to the component [2]. The based reference stress for a semi-elliptical
existence of a local metal loss. Therefore, the RSF approach can be crack-like flaw in a plate is defined by Willoughby [15] and given
expressed as in the following equation. as in Eq. (9), which is identical to that provided in Annex D of API
579-1/ASME FFS-1.
scm ¼ MSKiefner sq (4) sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

2
1 1
Where, gs þ g sb þs2q ð1  aÞ2
scm : Circumferential membrane stress increased by a local metal 3 b 3
sref plate ¼ (9)
loss. ð1  aÞ2
sq : Circumferential membrane stress without a flaw.
Where,
Since the circumferential membrane stress is sufficient to
characterize the maximum principal stress and the minimum a
principal stress is assumed to be zero, the circumferential t acL
a ¼ ¼ (10)
membrane stress is considered to be equivalent to the reference t tðcL þ tÞ
1þ
stress under the Tresca theory. Therefore, Eq. (4) can be written as: cL

sref ¼ MSKiefner sq (5) The Eq. (10) of API 579-1/ASME FFS-1 is not identical to but is
API
almost equivalent to the Eq. (7) of the p-M diagram method. The
Where, difference is not significant and is disregarded in this paper.
sref API : Reference stress for a local metal loss in a cylindrical It is obvious by comparing Eqs. (6) And (9), that the reference
component in API 579-1/ASME FFS-1. stress solution of the p-M diagram method in Eq (6) can be obtained
On the other hand, in the p-M diagram method, the circumfer- by replacing sq with MSChell sq in Eq. (9).
ential membrane stress increased by a local metal loss is derived as Bending stress sb due to internal pressure is usually small and
the reference stress for a longitudinal semi-elliptical surface crack- can be considered to be negligible. Hence Eq. (6) can be rewritten
like flaw. The reference stress for a longitudinal semi-elliptical as:
surface crack-like flaw in the p-M diagram method is given as:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi MSChell

2  2 sref ¼ s (11)
1 1
pM
1a q
g sb þ g sb þ MSChell sq ð1  aÞ2
3 3 In the p-M diagram method, the bulging factor proposed by
sref pM ¼ (6)
ð1  aÞ2 Chell is adopted as mentioned above. If the surface correction factor
proposed by Kiefner in Eq. (2) were adopted for the p-M diagram
Where, method, Eq. (11) would be rewritten as Eq. (12) with parentheses,

0:75
a which denotes a supposed expression:
g ¼ 1  20 a3 : Modified factor for bending effect
2cL
proposed by Sattari-Far [14].   MSKiefner
sref ¼ s (12)
sb : Circumferential bending stress component without a flaw. pM
1a q
a: Relative rectangular flaw depth equivalent to semi-elliptical
Here, there might be some controversy that the surface
flaw depth proposed by Willoughby [15], which is given by
correction factor defined by Kiefner is the same as the bulging
aeff pacL factor defined by Chell. Each factor is associated with the bulging
a ¼ ¼ (7) factor and shell parameter for a through-wall crack-like flaw
t tðpcL þ 4tÞ
510 K. Oyamada et al. / International Journal of Pressure Vessels and Piping 88 (2011) 507e517

defined by Folias [5,13]. It is noted that for a through-wall flaw the
surface correction factor Mt in Eq. (2) is not identical to the bulging
factor Mt ðlÞ in Eq. (8), but it is disregarded in this paper because the
difference is not significant.
The empirical Eq. (2) of MSKiefner obviously does not coincide
with the surface correction factor Mt for a through-wall flaw when
a partly through-wall flaw approaches a through-wall configura-
tion. Therefore, the theoretical Eq. (8) of MSChell should be used as
stated in Annex D, which does not have such a discrepancy.
Further, Eq. (5) can be obtained from the following reference
stress for a longitudinal semi-elliptical crack-like flaw in a cylin-
drical component provided in Annex D of API 579-1/ASME FFS-1.
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

2  2
1 1
gs þ g sb þ MS sq ð1  aÞ2
3 b 3 reference stress is shown in Fig. 1. There is significant difference
sref API Crack ¼ (13)
ð1  aÞ2
Annex D gives an option to select MS from MSKiefner or MSChell. By
making sb ¼ 0 since it is considered negligible, then Eq. (13) can be
reduced to Eq. (5).However, by comparing Eq. (13) with Eq. (9), Eq.
(13) can be obtained by replacing sq in Eq. (9) with MS ð1  aÞsq ,
which is not consistent with derivation of either MSKiefner or MSChell.
Since Annex D of API 579-1/ASME FFS-1 recommends using
MSChell as MS in Eq. (13), the following equation can be obtained as
well as Eq. (5).
sref API ¼ MSChell sq (14)
It is described in reference [6] that Eq. (5) is better than Eq. (14)
in predicting the physical collapse load for a cylindrical shell,
although the MSKiefner does not approach the proper bound for
a through-wall flaw. To compare these reference stresses given by
Eqs. (5), (11), (12) and (14), the dependence of internal pressure on
among these equations. Also, Fig. 2, showing the dependence of
flaw depth ratioa=t, indicates that reference stress given by Eq. (14)
leads to a less accurate result. This is simply attributable to the
canceling of the denominator ð1  aÞ in Eq. (13).

Fig. 1. Effects of circumferential reference stress on being divided by (1a).

 K. Oyamada et al. / International Journal of Pressure Vessels and Piping 88 (2011) 507e517 511

Fig. 2. Example of unexpected calculation results by Eq. (14) with MSchell .

By using Eq. (11), RSF given by Eq. (1) can be obtained as follows;
LDC 1a
RSF ¼ ¼ Chell
LUC MS
Therefore, the value of Kiefner’s surface correction factorMSKiefner
given by Eq. (2), which is experimentally obtained for high strength
steels, should be similar to that of MSChell =ð1  aÞ as a logical
consequence.
For comparison purpose, reference stress given by Eq. (5) with
Eq. (2) of C ¼ 0:85 and C ¼ 1:0 are incorporated into Fig. 2. The
value of Eq. (5) with Eq. (2) of C ¼ 1:0 is close to that of theoretical
Eq. (11) but seems to be less accurate because of limited use for the
empirical relation of Eq. (2).
tc : Corroded wall thickness located away from the region of local
metal loss.
cq : A half of circumferential flaw length.
(15)
Here, MtC is derived from a correction factor of the stress
intensity factor of a cylindrical shell with a circumferential
through-wall crack subjected to membrane force by Erdogan et al.
[21] as an elastic solution.
On the other hand, Zahoor et al. [22] derived a correction factor
of a cylindrical shell with a circumferential through-wall crack
subjected to membrane force as a limit solution (for elastic
perfectly-plastic material).
a

3. Comparison of assessments for the circumferential extent

of a local metal loss or a circumferential crack-like flaw of a
cylindrical component subjected to external bending moment
between API 579-1/ASMEFFS-1 and the p-M diagram method

In the level 2 assessment of Part 5 of API 579-1/ASME FFS-1, the

longitudinal stress calculation Eq. (16) is given in order to assess
plastic collapse for the case where external bending moment such
as due to earthquakes, applies.

scl ¼ Msc sl (16)

Where, b
scl : Longitudinal stress increase due to local metal loss.
sl : Longitudinal stress without a flaw.
! !
1 d
1
MtC tc
MSC ¼ ! : Surface correction factor for the
d
1
tc
circumferential extent of a local metal loss (on the analogy of Eq.
(2)).

1:0 þ 0:1401ðlc Þ þ0:002046ðlc Þ

2 4
MtC ¼ (17)
1:0 þ 0:09556ðlc Þ þ0:0005024ðlc Þ
2 4

1:285ð2cq Þ Fig. 3. Comparison of increase of longitudinal membrane stress component between

lc ¼ pffiffiffiffiffiffiffi : Circumferential flaw length parameter. with plastic analysis and with elastic analysis.
Dtc
512 K. Oyamada et al. / International Journal of Pressure Vessels and Piping 88 (2011) 507e517
Fig. 4. Comparison of reference stresses computed by various references and experi-
mental results (Plastic collapse is assumed to have occurred at Lr ¼ 1).
Fig. 5. Experimental setup for a cylindrical component with surface flaw.
p
MtC Zahoor ¼ (18)
2arccosfsinðq=2Þg  q
Where,
2q: Angle (rad.) of a circumferential through-wall crack.
This equation is also adopted in Annex D of API 579-1/ASME
FFS-1 as for a cylinder with a circumferential through-wall crack-
like flaw.
Fig. 3 shows the difference between Erdogan’s elastic solution
(Eq. (17)) adopted in Part 5 for a local metal loss in API 579-1/ASME
FFS-1 and Zahoor’s limit solution (Eq. (18)). As shown in Fig. 3, some
geometric restriction in the case of using the elastic solution may be
necessary because of a lack of conservatism against plastic collapse
in a certain condition. As the reference stress of a cylindrical shell
with a circumferential partly through-wall flaw subjected to
membrane force as a limit solution (for elastic perfectly-plastic
material) is also available in Annex D, this limit solution should
be used instead of the elastic solution for assessing a cylindrical
component with a local metal loss. The assessment by using Mises
equivalent stress criterion based on unsure stress states and Kief-
ner’s relation mentioned in the preceding section of this paper may
lead to less accurate results.
4. Comparison of reference stresses computed by various
references with experimental and FEA results
To confirm the validity of the circumferential and longitudinal
reference stresses proposed in various references [3,5,8,9,10,11],
comparison of those with experimental data and FEA results for the
same condition of the experiment is shown in Fig. 4. The cylinder
size and flaw dimensions for experimentation are shown in Table 1
[3]. ABAQUS [23] finite element software was used as a solver.
Three-dimensional 8-node brick elements (C3D8R) are applied
(Figs. 7 and 9). Both material nonlinearity and geometrical
nonlinearity are considered. True stress-true strain curve shown in
Fig. 11 is assumed as that of a typical Japanese carbon steel (STPG
370). The experimental setup is shown in Fig. 5. Fig. 6 shows the
comparison of loadestrain relation obtained by experiment with
those by FEA. There is good agreement between them.
 K. Oyamada et al. / International Journal of Pressure Vessels and Piping 88 (2011) 507e517 513

Fig. 6. Results of experiments and comparison with FEA on the same models as the experiments.

Table 1
Summary of condition of test specimens and test results of pipes with local metal loss by pressurizing and external bending (v I P: Pressurizing, v II MT: External bending).

Cylinder no. Young’s modulus (GPa) smin

ys (MPa) smin
uts (MPa) smean
ys (MPa) smean
uts (MPa) Do (mm) Thickness t(mm) Flaw dimension (mm)

2CL 2Cq a
vIP 203 215 370 326 483 165.5 5.4 29.62 29.87 2.83
v II MT 165.5 5.5 29.9 29.8 3.52

Cylinder LTA Whole cylinder Failure

no. mode
p_pl.init M_pl.init p_TES M_TES p_pl.init M_pl.init p_TES M_TES p_max M_max
(MPa) (kNm) (MPa) (kNm) (MPa) (kNm) (MPa) (kNm) (MPa) (kNm)
vIP 6.6 12.2 16.8 20.8 28.9 Break
v II MT 24.1 30.1 37.3 39.2 45.6 Ovalization
514 K. Oyamada et al. / International Journal of Pressure Vessels and Piping 88 (2011) 507e517

Flaw
750[mm] 2cL = 29.62[mm]

Flaw
Do = 165.5[mm]

Whole Cylinder (t=5.4 [mm]) a = 2.83[mm] cθ = 14.94[mm]

Plate center

Fig. 7. FEA model for pure pressurizing case (ABAQUS_C3D8R).

Fig. 8. FEA results of pure pressurizing case.

Fig. 6 also shows that the difference of circumferential extent of the experimentally obtained TES loads and the limit loads predicted
a local metal loss does not significantly affect the TES value at the by these proposals as well as the proposal per the API 579-1/ASME
local metal loss area, since the TES loads computed by FEA for the FFS-1 Code as shown in Fig. 4.
models of 2cq ¼ 29.62 mm, 42.91 mm and for the other model of From Fig. 4, the p-M method can be used to estimate the plastic
2cq ¼ 14.9 mm are almost equivalent. Similar results for the model collapse load condition determined by experimentation and FEA
(Do ¼ 1626 mm, t ¼ 13 mm, a ¼ 11 mm, 2cL ¼ 300 mm, 2cq ¼ 70, better than other methods in both the cases where pure internal
300, 615, and 836 mm) are shown in reference [3]. pressure and pure external bending moment apply. As shown in
The shape of a local metal loss area such as a corroded pipe in Figs. 8 and 10, general yielding at a local metal loss area occurs at
a refinery plant varies widely. A rectangular configuration, the condition of around Lr ¼ 1 given by the p-M method.
however, gives the worst case evaluation slightly in terms of plastic
collapse. Experiments and FEA are conducted for a cylinder with
almost a rectangular local metal loss to obtain a lower bound. This
is partly because on-site measurement of the thickness profile
requires elaborate work.
Fig. 4 also shows that French RSE-M code method [9] computed
as sys ¼ 0:85sys , which is specified in the reference for carbon
steels, matches the TES loads computed with the Mises yield
criterion, which is denoted as (c) in Fig. 4(a). Recent proposals by
Staat and Vu [10] and Lei [11] resolved the underlying discrepancy
in Kiefner’s correction factor, which does not approach the proper
bound for a through-wall flaw. There are still differences between Fig. 9. FEA model for pure external bending case (ABAQUS_C3D8R).
 K. Oyamada et al. / International Journal of Pressure Vessels and Piping 88 (2011) 507e517 515

Fig. 10. FEA results of pure external bending case.

5. Discussion on a recommended criterion and margin as

a safety standard

As the RSF  RSFa ð ¼ sys =sf y0:9Þ criterion adopted in Part 5 of

API 579-1/ASME FFS-1 is not based on the criterion of Mises or
Tresca theory, the safety margin for general yielding at a metal loss
area may become less accurate especially for high metal tempera-
tures such as 300  C. Furthermore, the calculated stress by using
those current equations may be underestimated or overestimated
slightly or considerably than the actual stress state as studied so far,
especially, the shell parameter l is relatively small (less than around
1.0).
On the other hand, the p-M diagram method recommends
adopting a factor of 1.5 as a margin with regard to the specified
yield strength due to the fact that almost all codes adopt it. Fig. 12
Fig. 11. True stress-true strain curve for the FEA in Figs. 7 and 9.
gives a comparison of shell parameter dependence on the allowable
flaw depth between the RSFa criterion adopted in Part 5 of API 579-
1/ASME FFS-1 and the p-M diagram method. Keeping stress levels

Fig. 12. Comparison of allowable flaw depth for longitudinal extent of a flaw between the p-M diagram method and Part 5 of API 579-1/ASME FFS-1.
516 K. Oyamada et al. / International Journal of Pressure Vessels and Piping 88 (2011) 507e517
Fig. 13. Dependence of design margin Xs of construction code on allowable remaining thickness ratio where shell parameter l is relatively large.
at the local metal loss area below the plastic limit load (general
yielding) enables the avoidance of secondary damage such as
ratcheting due to general yielding of the local metal loss area,
which is detailed in reference [7].
Since RSF is applied to the MAWP (maximum allowable working
pressure) of a whole cylinder in API 579-1/ASME FFS-1, the safety
margin of RSF approach for high temperature relies on the allow-
able stress which is governed by ultimate tensile strength and so is
usually constant up to 300e350  C for normal carbon steel. But the
RSF approach becomes less accurate in the evaluation of general
yielding in a local metal loss area as shown in Fig. 12 where the
metal temperature is relatively high or shell parameter is less than
around 1.0, whereas the RSF approach becomes over-conservative
where the metal temperature is not high and, coincidently, shell
parameter is larger than around 1.0.
As shown in Fig. 12, the allowable flaw depths determined by
the p-M diagram method converge individually to constant values
depending on temperatures if l is relatively large. Eq. (11) can be
roughly rewritten as Eq. (19) for larger l.
 a 2
sref
pM ysq = 1
t - The condition of the specified minimum yield strength with
From the following equations,
sref ¼ sf (Plastic collapse condition) with
sf ¼ smin
ys =1:5 and sq ¼ suts =Xs
min
Eq. (21) can be derived.
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 u
a u1:5 smin
Rt inf pM ¼ 1  ¼ t uts
t Xs smin ys
The dependence of the design margin XS on the allowable
remaining thickness ratio Rt inf pM , given by Eq. (21) based on the
condition of the specified minimum yield strength with a safety
margin of 1.5 for a typical carbon steel ðsmin ys ¼ 215
MPa; smin
uts ¼ 370 MPaÞ, is shown in Fig. 13. The permissible flaw
depth, where l is large, can be allowed by around 20%
ða=t ¼ 1  Rt inf ¼ 0:2Þ of the minimum wall thickness for the
design margin XS ¼ 4, and around 10% ða=t ¼ 0:1Þ for XS ¼ 3,
which provides consistency with the requirements contained in
ASME Nuclear Code Case N-597-2-3622.4 [24] for equipment con-
structed by the design margin XS ¼ 3. The condition of the specified
minimum yield strength with a safety margin of 1.5 ðsf ¼ sminys =1:5Þ
can be applied to assess equipment possessing a metal loss area,
irrespective of the design margin of the construction code if the
same safety margin as Code Case N-597-2-3622.4 is required.
6. Conclusions
In concluding this comparative study, the following should be
recommended for adoption in the FFS rules.
- The reference stress for a longitudinal surface crack-like flaw or
the longitudinal extent of a local metal loss on a cylindrical com-
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 1 2  2  2
g sb þ g sb þ MSChell sq 1  a
ponent should be: sref ¼ 3 3
ð1  aÞ2
- The reference stress for a circumferential surface crack-like
flaw or the circumferential extent of a local metal loss should
be derived by incorporating the limit analysis solution
provided by Zahoor.
- The acceptable criteria should be based on the Mises or Tresca
theory.
(19)
a safety margin of 1.5 ðsf ¼ smin ys =1:5Þ should be applied to
assess equipment possessing a metal loss area.
(20) Acknowledgment
The authors would like to thank the specialized committee for
safety of Ibaraki Prefecture for kindly supplying the experimental
data used in this study.
(21)
References
[1] Ibaraki FFS Rule, 2006, Assessment Standard for Externally Corroded Pressure
Equipment, Ibaraki Prefecture (in Japanese).
[2] Konosu S, Mukaimachi N. Plastic collapse assessment procedure for vessel with
local thin area simultaneously subjected to internal pressure and external
bending Moment. ASME J. Pressure Vessel Technol. 2008;130:011207.
[3] Konosu S, Kano M, Mukaimachi N, Komura H, Takada H. Plastic collapse load
for vessel with external flaw simultaneously subjected to internal pressure
and external bending momenteexperimental and FEA results. ASME J. Pres-
sure Vessel Technol. 2009;131:021206.
[4] ASME, 2004, Boiler and Pressure Vessel Code Section VIII, Div.2, App.6, 6e153.
[5] API Standard 579e1/ASME FFS-1, Fitness-For-Service, 2007
[6] Janelle, JL, Osage, DA, Burkhart, SJ, 2005, An Overview and Validation of the
Fitness-For-Service Assessment Procedures for Local Thin Areas, WRC
Bulletin505, Welding Research Council, INC.
 K. Oyamada et al. / International Journal of Pressure Vessels and Piping 88 (2011) 507e517
[7] Oyamada K, Konosu S, Miyata H, Ohno T. Proposal of acceptance criterion for
plastic collapse assessment rule on local metal loss, Strength, Fracture and
Complexity, Vol. 6. IOS Press; 2010. 115e127.
[8] BS7910, 2005, Guide on methods for assessing the acceptability of flaws in
fusion welded structures.
[9] Marie S, Chapuliot S, Kayser Y, Lacire MH, Drubay B, Barthelet B, et al. French
RSE-M and RCC-MR code appendices for flaw analysis: presetation of the
fracture parameters calculation e Part IeV". Int J Pres Ves Piping 2007;84.
[10] Staat M, Vu DK. Limit analysis of flaws in pressurized pipes and cylindrical
vessels. Part I: axial defects. Eng Fracture Mechanics 2007;74:431e50.
[11] Lei Y. A review of limit load solutions for cylinders with axial cracks and
development of new solutions. Int J Pres Ves Piping 2008;85:825e50.
[12] Kiefner JF, Maxey WA, Eiber RJ, Duffy AR. Failure stress levels of flaws in
pressurized cylinders. Progress in flaw growth and fracture toughness
testing. ASTM STP 536. American Society for Testing and Materials; 1973.
461e481.
[13] Folias ES. Failure correlation between cylindrical pressurized vessels and flat
plates. Press Vessels Piping 1999;76:803e11.
[14] Sattari-Far. Finite element analysis of limit loads for surface cracks in plates.
Int J Pres Ves Piping 1994;57:237e43.
[15] Willoughby AA, Davey TG. Plastic collapse in part-wall flaws in plates, fracture
mechanics. In: Wei RP, Gangloff RP, editors. perspectives and directions
517
(twenties symposium), ASTM STP 1020. Philadelphia: American Society for
Testing and Materials; 1989. p. 390e409.
[16] Chell, GG, 1990, Application of the CEGB Failure Assessment Procedure, R6, to
Surface Flaws, ASTM STP1074, pp.525e544.
[17] Konosu S, Kano M, Mukaimachi N, Kanamaru S. Validity of assessment
procedure in p-M method for multiple volumetric flaws. ASME J Press Vessel
Technol 2009;132:006001.
[18] Mukaikachi N, Konosu S. Plastic collapse assessment procedure for vessels
with deep local thin area subjected to internal pressure. Nucl Eng Des 2009;
239:1171e9.
[19] Konosu S, Miyata H. Assessment of overlapped internal and external volu-
metric flaws in p-M diagrams. ASME J Press Vessel Technol 2011;133:031208.
[20] Konosu S. Assessment procedure for multiple volumetric flaws in p-M
diagram. ASME J Press Vessel Technol 2009;131:025903.
[21] Erdogan F, Ratwani M. A circumferential crack in a cylindrical shell in torsion.
Int J Frac Mech 1972;8:87e95.
[22] Zahoor A, Norris DM. Ductile fracture of circumferentially cracked Type-304
stainless steel pipes in Tension. Trans ASME 1984;106:399e404.
[23] ABAQUS, Finite Element Computer Program, Version 6.7.-5, Dassault Systems
Simulia Corp.
[24] ASME, 2003, Case of ASME B&PV Code N-597-2, Requirements for Analytical
Evaluation of Pipe Wall Thinning Section XI, Division1.