International Journal of Pressure Vessels and Piping 87 (2010) 66–73

Contents lists available at ScienceDirect

International Journal of Pressure Vessels and Piping

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

Reference stress method for evaluation of failure assessment curve

of cracked pipes in nuclear power plants
Masayuki Kamaya a, *, Hideo Machida b
a
Institute of Nuclear Safety System, Inc., 64 Sata, Mihama-cho, Mikata-gun, Fukui 919-1205, Japan
b
TEPCO Systems Corp., Shibusawa City Place Eitai, 2-37-28 Eitai, Koto-ku, Tokyo 135-0034, Japan

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

Article history: In order to obtain a precise failure assessment curve (FAC) in the R6 defect assessment procedure, it is
Received 24 October 2008 necessary to evaluate the J-value of cracked components. The reference stress method can be used for
Accepted 9 February 2009 estimating J-values. However, the accuracy of estimation depends on the limit load used for evaluating
the reference stress. In this study, the applicability of several limit load solutions was investigated
Keywords: through comparison with the results of elastic-plastic finite element analyses (FEA). A pipe containing
Reference stress
a circumferential surface crack was analyzed under pure bending load. Six materials used in nuclear
J-integral
power plants were assumed. It was shown that the reference stress method is valid for FAC evaluation.
Failure assessment curve
Two-parameter method The maximum non-conservativeness caused by using the reference stress method is less than 20%
Structural integrity compared to the results obtained by FEA.
Ó 2009 Elsevier Ltd. All rights reserved.

1. Introduction reference stress method is valid for estimating J-values under

various conditions [6–9]. However, the accuracy of the estima-
The CEGB’s defect assessment procedure (R6) [1] is used tion is dependent on the limit load used in evaluating the
extensively for assessing the integrity of cracked components in reference stress [10,11] and there is no theoretical solution that
nuclear power plants [2,3]. Two kinds of curves are employed for gives a precise J-value. Several attempts have been made to show
the assessment. One represents the driving force of crack relevant limit load solutions [12]. Moreover, modification was
extension and is referred to as the failure assessment curve made to a limit load solution in order to optimize the J-value
(FAC). The other relates to the resistance of unstable ductile crack estimation [13,14]. However, it is not yet clear which solution is
extension and is derived from J-R curves and the stress intensity available for the R6 procedure. In the fitness-for-service code for
factor of the cracked component. The R6 procedure provides nuclear power plant components of the Japan Society of
three options for the evaluation of FAC. Option 3 enables accurate Mechanical Engineers (JSME Code) [3], the failure assessment
assessment for unstable ductile crack extension. However, it is procedure based on the R6 procedure is prescribed. However, in
necessary to evaluate the J-integral value (J-value) of the cracked JSME Code, Option 2 is not included and the J-value has to be
component. Since the J-value depends not only on the geometry evaluated in order to obtain FAC. Then, in order to apply the
of the crack and component but also on the stress-strain rela- reference stress method to JSME Code, available limit load
tionship of the material, only limited solutions are available for solutions for materials used in the nuclear power plant compo-
materials of which stress-strain relationship is approximated by nents were investigated.
the Ramberg–Osgood law [4,5]. Therefore, a numerical calcula- In this study, the reference stress method was applied to the
tion such as finite element analysis (FEA) is required in Option 3. evaluation of FAC of pipes with circumferential surface crack under
On the other hand, in Option 2, FAC is derived from the stress- bending stress. For assessment of nuclear power plant components,
strain relationship of the material. The basis of this option is the FAC for various combinations of crack and pipe geometry was
estimation of the J-value using the reference stress method. The assumed for six kinds of materials used in nuclear power plants.
Four limit load solutions were used for the reference stress method
and the accuracy was quantified through comparison with the
* Corresponding author. Tel.: þ81 770 379114; fax: þ81 770 372009. results obtained by elastic-plastic FEA. Finally, the validity of the
E-mail address: [email protected] (M. Kamaya). limit load solutions were discussed.

0308-0161/$ – see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijpvp.2009.11.002
 M. Kamaya, H. Machida / International Journal of Pressure Vessels and Piping 87 (2010) 66–73 67

Fig. 1. Geometry of pipe containing circumferential crack under bending load.

2. FAC by reference stress method  

K 2 1  n2
Je ¼ ; (4)
2.1. Procedure for FAC evaluation E
where K, E and n denote the stress intensity factor (SIF), Young’s
A pipe containing a circumferential surface crack under global modulus and Poisson’s ratio, respectively.
bending load was treated as shown in Fig. 1. FAC is obtained by In the reference stress method, J-value can be estimated by the
connecting points (Sr, Kr), and is defined as follows: following equation:
P
Sr ¼ (1)
PC E3ref
J ¼ Je : (5)
sffiffiffiffi sref
Je
Kr ¼ (2)
J
sref is the reference stress evaluated by:
where P is the bending stress. PC is the collapse stress and is given
by the following equation in the JSME Code [3]:  
P
sref ¼ sy : (6)
  PL
4sy ba a
PC ¼ cos  sin b ; (3)
p 2t 2t The reference strain 3ref is the strain corresponding to the reference
stress in the stress-strain relationship of the material. PL is the limit load
sy is the 0.2% proof (yield) strength of the material. FAC has an discussed in this study. By using Eq. (5), Kr can be expressed as follows:
upper value of Sr ¼ sf/sy (cut-off value), where sf is flow stress
defined as the mean of sy and the ultimate strength. J in Eq. (2) !0:5
E3ref
denotes the J-value and Je is the elastic J-value evaluated under Kr ¼ : (7)
elastic calculation:
sref

Fig. 2. Examples of finite element mesh.

68 M. Kamaya, H. Machida / International Journal of Pressure Vessels and Piping 87 (2010) 66–73

Table 1
Tensile properties and parameters for Ramberg–Osgood law.

Temperature Yield Ultimate Young’s Parameters Ref.

(degree C) strength Strength modulus for R–O’s
(MPa) (MPa) (GPa) law

a n
SCS14A 325 276 726 174 0.965 4.34 [17]
SCS14 325 380 551 174 1.186 9.17 [17]
AW
SUS304W 288 353 446 176 2.15 12.97 [19]
SFVQ1A 325 413 577 180 1.47 7.91 [20]
STPA24 310 245 482 181 1.64 5.98 [20]
SUS316 RT 285 589 195 – – [18]

Therefore, we can obtain FAC without using the J-value. In the R6

procedure, the following equation is used by incorporating a plas-
ticity correction at the crack tip region:
!0:5
E3ref s3ref Fig. 4. Stress intensity factors for cracked pipe.
Kr ¼ þ : (8)
sref 2s2y E3ref

g ¼ g1 ða=tÞ2 þg2 ða=tÞ þ 1:04 (10b)

2.2. Limit loads
2
g1 ¼ 4:26ðb=pÞ 1:35ðb=pÞ þ 0:8 (10c)
In order to estimate the J-value by the reference stress method,
the limit load must be defined in Eq. (6). Several limit load solutions
2
have been proposed. In this study, the following four equations g2 ¼ 2:30ðb=pÞ þ1:57ðb=pÞ  0:77 (10d)
were used for the reference stress method:

(1) Global limit load (same as the collapse stress for Sr)
(3) Local limit load [15]
 
PL 4 ba a
¼ cos  sin b (9)
sy p 2t 2t !
PL pR2m t 1  at
¼ (11a)
sy Zp 1  at M1
o

(2) Optimized limit load [14] qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

2 3
Mo ¼ 1 þ 0:26ðb=pÞ þ 47ðb=pÞ 59ðb=pÞ (11b)
 
PL 4R2 t ba a
¼ g m cos  sin b (10a)
sy Zp 2t 2t

Fig. 3. Stress-strain relation used in finite element analyses. Fig. 5. Effect of mesh size on FAC (SCS14A).
 M. Kamaya, H. Machida / International Journal of Pressure Vessels and Piping 87 (2010) 66–73 69

Fig. 7. Comparison of FAC obtained by finite element analyses and reference stress
method (for small crack). (a) SCS14A. (b) SUS316.
Fig. 6. FAC obtained by finite element analyses. (a) For various geometries (SCS14A).
(b) For various materials.

model, which contains 7794 20-node solid elements and 40192

(4) Limit load from R6 procedure [1] nodes. The geometry of the analyzed model is shown in Fig. 1,
where the length of pipe was set to L ¼ 5(Ri þ t) in order to suppress
  2
the edge effect.
PL 4R2m t 1 t a
¼ 1þ sin r  fc sin b (12a)
sy Zp 12 Rm 2t
   4. Geometrical and material conditions
p t a ba
r¼ 1 1 þ (12b)
2 2Rm 2Rm pt Geometrical conditions were possible combinations of the
following:
t þ a 3t 2  6at þ 4a2
fc ¼ 1  þ (12c) Ri/t: 5, 10
Rm 12R2m
a/t: 0.3, 0.5, 0.7
a/c: 0.2, 0.5
3. Finite element analysis
Pipe diameter was set to a constant value of 14B, because the
In order to evaluate the accuracy of estimated J-values obtained same FAC was obtained by FEA for various pipe diameters from 4B–
by the reference stress method, elastic-plastic FEA was carried out 36B under the same value of Ri/t.
using the general-purpose finite element program ABAQUS, Materials assumed in this study were cast stainless steel
Version 6.5 [16]. Fig. 2 shows an example of the finite element (SCS14A) [17], cast stainless steel weld metal (SCS14AW) [17],
70 M. Kamaya, H. Machida / International Journal of Pressure Vessels and Piping 87 (2010) 66–73

stainless steel (SUS316) [18], stainless steel weld metal 5. Results and discussion
(SUS304 W) [19] and low-alloy steel for piping (SFVQ1A and
STPA24) [20]. Properties of the six materials are summarized in 5.1. Validity of FEA
Table 1. Except for SUS316, the stress-strain relationship was
approximated by the Ramberg–Osgood law, and is expressed by Fig. 4 shows the comparison between SIF obtained by the elastic
the following equation: analysis using the current finite element model and reference data
 n (Ref. [21] and [22]). The analysis was carried out under conditions
3E s s of Ri/t ¼ 10, a/t ¼ 0.4 and a/c ¼ 0.25. The agreement of the results
¼ þa : (13)
sy sy sy implies that the finite element model used in this study is valid for
precise analysis of the stress field near the crack tip. Furthermore,
The constants a and n are also shown in Table 1. In FEA, the stress-
FAC for SCS14A was evaluated for different mesh size for the
strain relationships shown in Fig. 4 were used. These curves consist
conditions of Ri/t ¼ 10, a/t ¼ 0.5 and a/c ¼ 0.2 Fig. 5. The number of
of lines connecting the points obtained by Eq. (13), although the
elements and nodes for the current model are 17910 and 92156,
strain at the yield strength is sy/E and a linear relationship is
whereas 45090 and 214446 for precise analysis, respectively.
assumed below this strain.
Almost identical FAC shows the well converged FAC was obtained
The magnitude of applied load varied from zero up to the cut-off
with the current mesh size.
value. FEA was performed at every 5% of the cut-off load. Namely,
FEA was performed 20 times for each geometrical and material
condition. 5.2. FAC by finite element analyses

Fig. 6 shows the FAC using the J-value obtained by FEA. The
elastic J-value, Je, was derived using stress intensity factors
obtained by elastic FEA of the same model. The shape of FAC was
dependent on the crack and pipe geometry. Kr seems to be larger as
the crack size increases when Sr is more than 1.0. In the range of
such large Sr, the material effect was significant and Kr decreased as
the constant n increased.

5.3. FAC by reference stress method

FAC obtained by the reference stress method is shown in

Figs. 7 and 8 together with that by FEA. Since FAC obtained using
Eq. (12) was almost the same as that obtained using Eq. (9) (global
limit load), the results from Eq. (12) were omitted in this study.
The reference stress method gave reasonable agreement with FAC
obtained by FEA in most cases. In the R6 procedure, the distance
between the point (Sr, Kr) and the origin of the diagram relates to
failure load due to unstable ductile crack extension. Smaller
distance to the origin implies greater conservativeness. Therefore,
based on FAC by FEA, the reference stress method gives conser-
vative results in the cases of Fig. 7(a) and Fig. 8(a) regardless of

Fig. 8. Comparison of FAC obtained by finite element analyses and reference stress Fig. 9. FAC obtained by reference stress method using poly-linear stress-strain curve
method (for large crack). (a) SCS14A. (b) SUS316. (SCS14A).
 M. Kamaya, H. Machida / International Journal of Pressure Vessels and Piping 87 (2010) 66–73 71

the limit solution. As for the results for SUS316, assessment using solution was relatively small. On the other hand, the global limit
the global and optimized limit load solutions possibly derives load solution exhibited the best estimation.
non-conservative evaluation depending on the geometrical The position along the FAC corresponds to the failure mode
conditions of the crack and pipe. The local limit load was (unstable ductile crack extension or plastic collapse). For ductile
conservative in all cases (geometry and material), although it material, such as stainless steel, failure is expected to be evaluated
seemed excessively conservative in some cases as shown in at a position of large Sr or over the cut-off value in FAC. Therefore,
Fig. 8(a). the maximum Q is not necessarily always less than unity for all
The non-conservativeness typically occurred between the ranges of Sr, since the materials assumed in this study are ductile
position of Sr ¼ 0.7 and 1.0. In particular, in the case of SUS316, FAC and have relatively large fracture toughness. When actually using
showed discontinuous change near Sr ¼ 1.0 and the reference stress the assessment procedure, the validity of the limit load should be
method caused larger Kr compared to FEA. This was brought about considered at large Sr.
by the stress-strain relationship, in which strain increases abruptly
after the yield strength. It should be noted that, except for SUS316,
Eq. (13) was used for the evaluation of reference stress and it shows
continuous change near the yield strength. For comparison
purposes, FAC obtained using the poly-linear relationship of Fig. 3 is a 1.5
shown in Fig. 9 for SCS14A material. At the yield strength, the strain Maximum Global LL
by Eq. (13) is (1 þ a)sy/E, which is larger than sy/E assumed in Fig. 3.
1.4 Optimized LL
Although the difference in strain is minor, less than 0.5%, the J-value Local LL
near the yield point is sensitive. Larger strain results in greater J-
value and smaller Kr. In Eq. (8), the effect of yielding at the crack tip 1.3

K r(re f) / K r(FEM )
is compensated. FAC obtained using Eq. (7) instead of Eq. (8) also is
shown in Fig. 9. The plasticity correction makes Kr small and seems 1.2
to contribute to accurate evaluation near Sr ¼ 1.0. It is expected that
the use of Eq. (13) brings about a similar effect to the plasticity 1.1
correction.
In order to quantitatively evaluate the conservativeness or non- 1.0
conservativeness of the reference stress method, the ratio
Q ¼ Kr(ref)/Kr(FEM) was evaluated (Fig. 10). The ratio of failure stress
0.9
evaluated by the reference stress method to that by FEA is
proportional to the parameter Q, when the critical points from the
failure assessment were on the same line which includes the origin
0.8

SFVQ1A

STPA24
SCS14A

SUS316
SCS14AW

SUS304W
of the diagram. Therefore, based on this assumption, the reference
stress method is conservative compared to FEA when Q < 1. The
maximum and mean value of Q among all geometrical and loading
conditions is shown in Fig. 11 for each material. As mentioned, FAC
obtained using the local limit load solution was always Q < 1 for all
material. The maximum Q is relatively large for SUS316. Unity of the Maximum value in all cases
mean of Q implies that the J-value estimation by the reference
stress method is reasonable. Due to excessive conservativeness of b 1.3
the local limit load solution, the mean value by the local limit load Average Global LL
Optimized LL
1.2 Local LL

1.1
K r(re f) / K r(FEM )

1.0

0.9

0.8

0.7
SFVQ1A
SCS14A

STPA24

SUS316
SCS14AW

SUS304W

Averaged value of all cases

Fig. 11. Summary of difference between Kr value obtained by finite element analyses
Fig. 10. A schematic drawing of failure assessment curve. (a) Maximum value in all and the reference stress method. (a) Maximum value in all cases. (b) Averaged value of
cases. all cases.
72 M. Kamaya, H. Machida / International Journal of Pressure Vessels and Piping 87 (2010) 66–73

The position in the FAC can be represented by the parameter

SC ¼ Kr(FEM)/Sr(FEM), which corresponds to the inclination of the
line connecting the origin and point (Sr(FEM), Kr(FEM)) on FAC as
shown in Fig. 10. Fig. 12 shows the same figure as Fig. 11 obtained
using data satisfying SC < 0.8. When we use the reference stress
method only for the condition of SC < 0.8, the non-conservative-
ness evaluated in Fig. 11 is reduced. Moreover, Q obtained using
Eq. (7) was almost the same with that by Eq. (8) as shown in
Fig. 13. This implies that it is possible to use Eq. (7) instead of
Eq. (8) when the critical point on the FAC in the failure assess-
ment is under the condition of SC < 0.8.
In all cases, the maximum Q is less than 1.2. Therefore, we can
conclude that the reference stress method is valid for failure
assessment of nuclear power plant components. All limit load

a 1.5
Maximum Global LL
1.4 Optimized LL Fig. 13. Difference between Kr value obtained by finite element analyses and the
Local LL reference stress method (for Kr(FEM)/Sr(FEM) < 0.8).

1.3
K r(re f) / K r(FEM )

solutions shown in this study can be used for reference stress

1.2 evaluation from an engineering point of view.

1.1 6. Conclusions

1.0 The reference stress method was applied to evaluate FAC in the
R6 assessment procedure for cracked pipe under pure bending.
0.9 The validity of four limit solutions was investigated by comparing
the results of elastic-plastic FEA assuming various materials used in
0.8 nuclear power plants. The conclusions obtained in this study are as
SFVQ 1A
SCS14A

SC S14AW

STPA24

SU S316
SU S304W

follows:

(1) The reference stress method can be applied to the R6 failure

assessment procedure for cracked pipe in nuclear power
plants.
Maximum value in all cases (2) The plasticity correction considered in Eq. (8) is effective for
b 1.3
J-value estimation under small-scale yielding conditions.
Approximation of stress-strain relationship by the Ramberg–
Average Global LL Osgood law can be used as an alternative for the correction.
Optimized LL
1.2 Local LL (3) Plasticity correction is not necessary in an assessment of
unstable ductile crack extension and plastic collapse.
(4) The maximum non-conservativeness caused by the reference
1.1
K r(re f) / K r(FEM )

stress method is less than 20% compared to the results

obtained by FEA, when the limit load solutions shown in this
1.0 study are used.

0.9 References

[1] Assessment of the integrity of structures containing defects. R6 revision 4:

0.8 British Energy Generation Ltd; 2007.
[2] ASME. Boiler & pressure vessel code section XI 2002 addenda. ASME 2002.
[3] The rules on fitness-for-service for nuclear power plants. JSME S NA 1–2004:
Maruzen 2004.
0.7
[4] Goldman NL, Hutchinson JW. Fully plastic crack problems: the center-cracked
SFVQ1A
S C S 14A

S C S 14AW

STPA24

S U S 316
SUS304W

strip under plane strain. International Journal of Solids and Structures

1975;11:575–91.
[5] Zahoor A. Ductile fracture handbook. EPRI; 1991. NP-6301-D: EPRI.
[6] Roche RL. Models of failure - primary and secondary stresses. Transaction of
ASME Journal of Pressure Vessel Technology 1988;110(3):234–9.
[7] Takamatsu S, Shimakawa T. Development of a simplified J-estimation scheme
Averaged value of all cases based on the reference stress method. Journal of Society of Material Science,
Japan 1994;43(493):1284–9.
Fig. 12. Summary of difference between Kr value obtained by finite element analyses [8] Shimakawa T, Miura N, Nakayama Y, Takahashi Y. Verification of simplified J-
and the reference stress method (for Kr(FEM)/Sr(FEM) < 0.8). (a) Maximum value in all integral evaluation method for flaw evaluation at high temperature. Journal of
cases. (b) Averaged value of all cases. Society of Material Science, Japan 2000;49(8):851–6.
 M. Kamaya, H. Machida / International Journal of Pressure Vessels and Piping 87 (2010) 66–73 73

[9] Miura N, Shimakawa T, Nakayama Y, Takahashi Y. Systematization of [15] Miler AG. Review of limit loads of structures containing defects. International
simplified J-integral evaluation method for flaw evaluation at high Journal of Pressure Vessel and Piping 1988;32:197–327.
temperature. Journal of Society of Material Science, Japan 2000;49(8): [16] ABAQUS/Standard user’s manual version 6.5. USA: ABAQUS Inc.; 2005.
845–50. [17] Koyama K, Hojo K, Muroya I and Kawaguchi S. Z-factors for aged cast duplex
[10] Ainsworth RA. The assessment of defects in structures of strain hardening stainless steel pipes and welds. 7th International Conference on nuclear
material. Engineering Fracture Mechanics 1984;19(4):633–42. engineering 1999: ICONE-7477.
[11] Asano M, Fukakura J, Kashiwaya H, Saito M. Application of the r6-rev. 3 [18] Matsuoka S. Relationship between 0.2% proof stress and Vickers hardness of
approach to ductile fracture analysis of carbon steel pipe with a circumferen- work-hardened low carbon austenitic stainless steel 316SS. Transactions of the
tial through-wall crack. Transactions of the Japan Society of Mechanical Japan Society of Mechanical Engineers 2004;70A:1535–41.
Engineers, Series A 1989;55(519):2299–306. [19] Asano M, Fukakura J, Kikuchi M. Failure assessment curves for austenitic
[12] Kim YJ, Shim DJ. Relevance of plastic limit loads to reference stress approach stainless steel pipes with a circumferential crack at a welded joint. Trans-
for surface cracked cylinder problems. International Journal of Pressure actions of the Japan Society of Mechanical Engineers 2001;67A:1194–200.
Vessels and Piping 2005;82:687–99. [20] Koyama K, Muroya I, Tanaka T, Nakamura T. Low alloy steel piping test for fracture
[13] Kim YJ, Oh CS and Song TK. Net-section limit pressure and engineering j criteria of leak before break. Nuclear Engineering and Design 1999;191:147–56.
estimates for axial part-through surface cracked pipes. 2007 ASME [21] Bergman M. Stress intensity factors for circumferential surface cracks in pipes.
pressure vessels and piping division Conference. PVP2007–26220; 2007. Fatigue Fract. Engng Mater.struct 1995;18(10):155–1172.
[14] Kim YJ, Kim JS, Lee YZ, Kim YJ. Non-linear fracture mechanics analyses of part [22] Chapuliot S. Formulaire de KI Pour les Tubes Comportant un Default de
circumferential surface cracked pipes. International Journal of Fracture Surface Semi-elliptique Longitudinal ou Circonferentiel. Interne Ou Externe
2002;116:347–75. 2000. Paoort CEA-R-5900.