Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
www.elsevier.com/locate/ijpvp
Tepco Systems Corporation, Tokyo Bijyutsu Club Building, 6-19-15 Shimbashi, Minato, Tokyo 105-0004, Japan
b
Institute of Environmental Studies, The University of Tokyo, 7-3-1 Hongo, Bunkyo, Tokyo 113-8656, Japan
Received 7 August 2001; revised 11 January 2002; accepted 12 January 2002
Abstract
In conventional probabilistic fracture mechanics (PFM) analyses, seismic loading is considered as a large deterministic value, although
there exists the variation of the seismic load as well as response of building and components. On the other hand, such stochastic behaviours
have already been taken into account in the eld of seismic probabilistic safety assessment. This paper proposes a new PFM model for
nuclear piping that takes into account the variation of seismic loading. The distribution in ground acceleration is modelled with the seismic
hazard curve. The distribution in piping response during a seismic event is modelled with a lognormal distribution. Since the seismic load
has large variation, when not adopting an upper limit to the distribution in seismic stress, the break probability calculated from the present
PFM analysis becomes equal to the probability that the seismic stress exceeds the collapse stress of a sound pipe. This implies that the
existence of a crack has no effect in these PFM analyses, and this result does not satisfy the purpose of PFM analysis to evaluate the failure
probability per crack. Therefore, the seismic stress was limited to the collapse stress of a sound pipe in the present PFM analysis to evaluate
the conditional break probability per crack. q 2002 Elsevier Science Ltd. All rights reserved.
Keywords: Probabilistic fracture mechanics; Earthquake; Seismic load; Variation; Break; Probability
1. Introduction
Design and structural integrity evaluation of components
in nuclear power plants are usually performed using a deterministic method. Here, the results obtained usually involve
excessive margin, because a certain safety margin is taken
into account in every evaluation process. Such a situation
inevitably causes an increase in plant construction cost. A
probabilistic method is one of the candidates to reduce such
excessive margin. In a probabilistic structural integrity
evaluation, the failure probability is calculated using mathematical models, which include the dominant factors
concerning failure behaviours. Here structural integrity is
assessed by the failure probability. As the safety margin is
considered only once in the nal stage of failure probability,
the probabilistic evaluation is regarded to give a rational
estimation compared to the conventional deterministic
evaluation. Fracture mechanics considering probabilistic
issues is called probabilistic fracture mechanics (PFM).
The study on PFM started in the middle of the 1970s to
assess structural reliability of an aircraft and a pressurized
* Corresponding author. Tel.: 181-3-4586-6723; fax: 181-3-4586-1190.
E-mail address: machida-hideo@tepsys.co.jp (H. Machida).
0308-0161/02/$ - see front matter q 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0308- 016 1( 02) 00011- X
194
H. Machida, S. Yoshimura / International Journal of Pressure Vessels and Piping 79 (2002) 193202
H. Machida, S. Yoshimura / International Journal of Pressure Vessels and Piping 79 (2002) 193202
195
196
H. Machida, S. Yoshimura / International Journal of Pressure Vessels and Piping 79 (2002) 193202
H. Machida, S. Yoshimura / International Journal of Pressure Vessels and Piping 79 (2002) 193202
#
1
1 ln SE =m 2
p exp 2
f SE
;
2
s
sS E 2p
SD
S
D
RF
3:92
s 0:95;
(2)
a
S
aS2 DS2
197
Table 1
Analysed pipe dimensions
Pipe
4B
16B
26B
Diameter (mm)
Thickness (mm)
Radius/thickness ratio
114.3
11.1
5.1
406.4
26.2
7.8
660.4
33.3
9.9
122.6 MPa
183.4 MPa
404 MPa
293.7 MPa
178.5 GPa
0.3
198
H. Machida, S. Yoshimura / International Journal of Pressure Vessels and Piping 79 (2002) 193202
Table 3
Loading conditions for crack growth analysis
Load no.
1
2
3
4
5
6
Frequency (cycles/year)
7
18
320
8
16
330
Min.
Max.
Min.
Max.
0
49.0
92.0
0
0
0
122.6
183.9
122.6
0
0
0
0
0
0
2122.6
261.3
212.3
0
0
0
122.6
61.3
12.3
s m 0:5Sm
p
DK , 13:2 MPa m
p
DK $ 13:2 MPa m
3Sm
C2
s bg
c
b .1
a
da
m=cycle
dn
8
< 1:738 10213 DK 5:95 ;
where B1 and B2 are the stress indices [12], Pin and Mip are
the internal pressure and the bending moment due to the
mechanical load, and Do, t and Zi are the outer diameter of
the pipe, pipe thickness, and the secondary section moment,
respectively. The rst term of the left hand side of Eq. (9)
denotes the stress caused by internal pressure, which is
assumed to be 0.5Sm as mentioned above. Therefore, the
stress caused by all the mechanical loads including the seismic stress is limited to 2.5Sm 3Sm 2 0:5Sm : Since the
moment caused by dead weight usually has only a small
inuence, the global bending stress of the pipe caused by
an earthquake is given by the following equation with the
stress index, B2:
sb
2:5Sm
B2
10
Table 4
Loads applied for crack stability assessment except seismic load
Pipe size
Pm (MPa)
Pbg (MPa)
Qm (MPa)
Qbg (MPa)
4B
16B
26B
61.3
61.3
61.3
0
0
0
0
0
0
151.0
109.8
91.1
H. Machida, S. Yoshimura / International Journal of Pressure Vessels and Piping 79 (2002) 193202
199
Table 5
Seismic load and frequency which were evaluated by the deterministic technique
Pipe size
Frequency
(per year)
Pm (MPa)
Pbg (MPa)
Qm (MPa)
Qbg (MPa)
4B
16B
26B
0.0125
0.0125
0.0125
61.3
61.3
61.3
125.5
91.2
75.9
0
0
0
151.0
109.8
91.1
G1
9
>
=
G max
180
>
>
Q ;
: 0:692 2 0:0115DB 1 0:188 1 0:0104DB log
p
8
>
<1
6 # G B
(11)
where M0 is the critical moment calculated using the netsection criterion [14], DB is the nominal size of pipe in inch,
and Q is the crack angle in radian.
Pre-service inspection (PSI) is considered in the analyses.
The non-detection probability, PND, is given by [8]
P ND
(
1
2 ZV
1 2 1 1 2 p
2
p 0
lnA=Ap
)
e
2t2
dt 1 1
12
200
H. Machida, S. Yoshimura / International Journal of Pressure Vessels and Piping 79 (2002) 193202
Fig. 8. Conditional break probabilities with deterministic and probabilistic seismic loads.
where
8 p
>
< 2 ac;
A
>
: p aD ;
B
4
1 0:005;
2c # DB
2c . DB
Ap
DB 25:4 mm;
p p
a DB ;
4
V 1:33;
ap 6:35 mm
b
4.2. Reference analysis
Based on the conditions described in Section 4.1, a PFM
analysis considering the variation of an earthquake was
carried out. Fig. 8 shows the break probabilities of all
cases. Since the ratio of the areas of the crack and the
ligament is large, the break probability of the smaller
diameter pipe is large. When considering the deterministic
seismic load, the break probability is almost identical to that
without considering the seismic load. On the other hand,
when considering the probabilistic seismic load, the break
probability increases notably especially for larger diameter
pipes. For smaller diameter pipes, the inuence of the earthquake tends to be much less since the failure probability due
to the loads, except the earthquake, are relatively larger.
4.3. Upper limit of seismic stress
The inuence of variability (expressed by Eq. (2)) of
the seismic stress on the break probability of the pipe
is assessed. The conditions for the calculation are as
follows:
Ground acceleration: 4 m/s 2.
1
a
P
p2 u2 mp
2
t
sf
M0
4r 2 ts f
4
s f 373 MPa
2
p
Z
pr t
14
15
H. Machida, S. Yoshimura / International Journal of Pressure Vessels and Piping 79 (2002) 193202
201
Table 6
Probabilities that seismic stress exceeds collapse stress
Pipe size
4B
16B
26B
6.16 10 25
2.21 10 25
1.17 10 25
Fig. 10. Conditional break probabilities with and without upper limit to seismic stress.
202
H. Machida, S. Yoshimura / International Journal of Pressure Vessels and Piping 79 (2002) 193202
when ground acceleration exceeds the level of an S2 earthquake. For the piping system, it is expected that a support
structure loses its function before spoiling the soundness of
the piping itself. Therefore, a proportional relation between
the actual stress in a pipe and the ground acceleration acting
on a building is not guaranteed, and then Eq. (2) is not
applicable. In the PFM analysis considering an earthquake,
it is difcult to determine the distribution in piping stress,
although this has a very large inuence on the results.
Therefore, more precise studies on the probability density
function of the seismic stress and its upper limit are necessary to improve calculation accuracy.
5. Conclusions
This paper has described a new PFM analysis considering
two kinds of distribution in the seismic stress acting on
nuclear piping. One is the distribution in ground acceleration, which is modelled with the seismic hazard curve. The
other is the distribution in piping response during a seismic
event, which is modelled with a lognormal distribution.
The function shape is determined by referring to the
response factor method used in a seismic PSA. It is found
from sensitivity analyses that the inuence of the latter
distribution dominates break probabilities. When not adopting an upper limit to the distribution in seismic stress, the
break probability calculated from the present PFM analysis
becomes equal to the probability that the seismic stress
exceeds the collapse stress of a sound pipe. This implies
that the existence of a crack has no effect in these PFM
analyses. This contradiction is caused by the simple adoption of the lognormal distribution to model the distribution
in seismic stress, although this is often adopted in seismic
PSAs. Therefore, the seismic stress was limited to the
collapse stress of a sound pipe in the present PFM analysis
to evaluate the conditional break probability per crack. The
results clearly show that effects of seismic stress are more
signicant in larger diameter pipes. More precise studies on
the probability density function of seismic stress and its
upper limit will be necessary to improve calculation
accuracy.
Acknowledgements
This study was performed as a part of the activities of the
PFM research subcommittee organized within the Japan
Welding Engineering Society sponsored by the Japan
Atomic Energy Research Institute.
References
[1] Becher PE, Pedersen A. Application of statistical linear elastic fracture mechanics to pressure vessel reliability analysis. Nucl Engng Des
1974;27:41325.
[2] W. Marshall, et al. An assessment of the integrity of PWR pressure
vessels. H.M. Stationery Ofce, London, 1976.
[3] Sinozuka M. Development of reliability-based aircraft safety criteria.
AFFDL-TR-76-31 1976.
[4] Harris DO, Lim EY, Dedhia DD. Probability of pipe fracture in the
primary coolant loop of a PWR plant. NUREG/CR-2189 1981.
[5] Wells JE, George LL, Cummings GE. Seismic safety margins
research program phase I nal report, systems analysis (project
VII). NUREG/CR-2015, UCRL-53021 1981;8.
[6] Ebisawa K, Abe K, Muramatsu K, Itoh M, Kohno K, Tanaka T.
Evaluation of response factors for seismic probabilistic safety assessment of nuclear power plants. Nucl Engng Des 1994;147:197210.
[7] Machida H. Effect of variation of seismic load on integrity of nuclear
power plant piping. SMiRT-16, MK03/2 2001.
[8] Lo TY, Mensing RW, Woo HH, Holman GS. Probability of pipe
failure in the reactor coolant loops of combustion engineering PWR
plants. Pipe failure induced by crack growth. NUREG/CR-3663
1984;2.
[9] Asada Y, Takumi K, Hata H, Yamamoto Y. Development of criteria
for protection against pipe breaks in LWR plants. Int J Pressure
Vessels Piping 1990;43:95111.
[10] Raju IS, Newman Jr JC. An empirical stressintensity factor equation
for the surface crack. Engng Fract Mech 1981;15:18592.
[11] ASME Boiler & Pressure Vessel Code, Section III, 1973.
[12] Technical standards for construction of nuclear power plant components, MITI Notication No. 501, 1994.
[13] Technical guidelines for aseismic design of nuclear power plants,
JEAG4601-1991.
[14] Kanninen MF, Broek D, Marschal CW, Rybicki EF, Sampath CG,
Simonen FA, Wilkowski GM. Mechanical fracture predictions for
sensitized stainless steel piping with circumferential crack. EPRI
NP-192 1976.