Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

A

Proceedings of the ASME 2023

tte
Pressure Vessels & Piping Conference

nd
ee
PVP2023
July 16-21, 2023, Atlanta, Georgia

R
ea
d-
O
nl
PVP2023-106288

y
C
op
y
EVALUATION OF CRACK INITIATION PROBABILITY BY FERMAT CODE IN CONFORMITY
WITH THE PROCEDURE BY A JAPANESE INDUSTRIAL PFM GUIDELINE

Satoshi Miyashiro1, Takayuki Sakai1, Masaki Nagai1, Masato Yamamoto1


1
Central Research Institute of Electric Power Industry, Japan

ABSTRACT PI Probability of crack initiation


The implementation of probabilistic fracture mechanics PTS Pressurized thermal shock
(PFM) on integrity assessment standards of reactor pressure 𝛷𝐾𝐼𝑐 Cumulative probability
vessels (RPVs) is not yet actualized in Japan. The most recent RPV Reactor pressure vessel
information on how PFM will be used in the near future is RTNDT Reference temperature for nil-ductile
provided by the guideline JEAG 4640-2018, which describes a transition
standard approach for evaluating the failure frequency of RPVs SBLOCA Small break loss of coolant accident
based on PFM. Central Research Institute of Electric Power SOV Stuck open valve
Industry is creating a new PFM analysis code FERMAT TWCF Through-wall cracking frequency
(Fracture mechanics Evaluation of RPV MATerials) for future
application of PFM in Japan. Our code is based on JEAG 1. INTRODUCTION
4640-2018 and has a minimum design for practical usage in Integrity assessment for reactor pressure vessels (RPVs) of
RPV structural integrity assessments. With a graphical user light water reactors is specified mainly by JEAC 4206-2016 [1]
interface, all processes can be handled in a single piece of based on deterministic fracture mechanics in Japan. The major
code. In the present study, calculation results by FERMAT were focus of JEAC 4206-2015 is the evaluation of stress intensity
compared with another Japanese PFM code PASCAL4 with a factor and fracture toughness for a crack with a certain under-
model case targeting Japanese RPV. The distribution of clad crack under a pressurized thermal shock event. Fracture
temperature and stress evaluated by FERMAT and PASCAL4 toughness is assessed with reference temperature for nil-ductile
were very similar. The stress intensity factor calculated by transition (RTNDT) considering shift caused by radiation
FERMAT and PASCAL4 was well corresponding with each embrittlement evaluated based on Japanese code JEAC 4201-
other. Additionally, sampled values for reference temperature 2007 (2013 addenda) [2] (hereinafter referred to as JEAC4201)
for nil-ductile transition were confirmed. The probability and with fixed parameters. On the other side, the Japanese integrity
the frequency of crack initiation were compared between assessment of RPVs is attempting to incorporate probabilistic
FERMAT and PASCAL4. The outcomes assessed using those fracture mechanics (PFM). Guideline JEAG 4640-2018 [3]
codes were closely related to one another. (hereinafter referred to as JEAG4640) was established as the
Keywords: Probability/Probabilistic, Structural Integrity, first step of standardization of PFM in Japan. The basic
Pressure vessels, Reactor/Nuclear framework for analyzing the failure frequency of RPVs based
on PFM is provided in JEAG4640.
NOMENCLATURE FAVOR [4] developed in the United States is famous as
CPI Conditional probability of crack initiation the PFM analysis code for RPV. However, FAVOR is difficult
EFPY Effective full power year to utilize for evaluating the integrity of Japanese RPV since it
FCI Frequency of crack initiation does not incorporate Japanese evaluation techniques like the
KI Stress intensity factor JEAC4201 method. PASCAL [5] has already been created by
KIc Fracture toughness Japan Atomic Energy Agency (JAEA) for PFM analysis of
LBLOCA Large break loss of coolant accident Japanese RPVs. As of January 2022, PASCAL version 4
MBLOCA Medium break loss of coolant accident (PASCAL4) has been made available. FAVOR code was used
MSLB Main steam line break to benchmark PASCAL4 [6]. PASCAL is confirmed
PFM Probabilistic fracture mechanics continuously by a working group consisting of universities,

1 Copyright © 2023 by ASME


manufacturers, and the Central Research Institute of Electric make up the FERMAT evaluation process for crack initiation:
Power Industry (CRIEPI) [7]. A wide range of evaluation determining the stress intensity factor for each crack (described
methods and parameter settings give PASCAL4 a great degree in Section 2.1); calculating the conditional probability of crack
of versatility. This feature was made possible by PASCAL4, initiation (CPI) using the fracture toughness equation
which was crucial in the development of JEAG4640. On the (described in Section 2.2); and determining FCI (described in
other hand, in exchange of their flexibility, PFM analysis codes Section 2.3). Below is a description of each step's specific
tend to be complex to operate. Users' choices of options or module procedures.
minor factors, as well as mistakes made by newcomers to the
PFM area, might lead to variations in outcomes. As a result, a 2.1 Evaluation of stress intensity factor
more straightforward PFM code is needed for PFM to be used Evaluation processes of stress intensity factor consist of an
in Japan. evaluation of the distribution of temperature and stress, crack
CRIEPI is developing a new PFM analysis code FERMAT setting, and stress intensity factor evaluation.
(Fracture mechanics Evaluation of RPV MATerials). The The temperature distribution is evaluated by using user
concept of FERMAT is a minimal design for practical use in input of material properties and transients of temperature and
the structural integrity assessment of RPVs based on heat transfer coefficient. By using those user inputs to solve the
JEAG4640. In this study, the results of FERMAT were transient heat conduction equation assuming an aximmetric
compared with the results of PASCAL4 for the model case as a model, the time dependency of temperature distribution may be
part of validation. determined. The evaluation of axial and hoop stress distribution
takes into account the primary stress induced by internal
2. EVALUATION OF CRACK INITIATION IN FERMAT pressure, the secondary stress caused by temperature
JEAG4640 contains basic framework to calculate distribution, and the residual stress from welding. Using an
frequency of crack initiation (FCI) and through-wall crack analytical solutions based on linear elastic theory, the primary
frequency (TWCF). FERMAT can assess FCI and TWCF based stress caused by internal pressure and the secondary stress
on JEAG4640. Aleatory uncertainty of fracture toughness and generated by temperature distribution is determined [8][9]. For
crack arrest toughness are considered in FERMAT. FERMAT welding residual stress, a user input table is employed.
also considers uncertainty of neutron fluence, chemical Through-clad surface cracks and embedded cracks in base
compositions, RTNDT, crack distribution, and occurrence metal and weld metal are considered in FERMAT according to
frequency of PTS transients as epistemic uncertainty by Monte JEAG4640. All surface cracks are regarded as circumferencial
Carlo method. cracks in accordance with the appendix's description in
In FERMAT, all processes, including pre- and post- JEAG4640. The proportion of surface cracks in weld metal
processes, may be managed using a single graphical user corresponds with the proportion of weld area in the whole
interface. Figure 1 depicts an illustration of the graphical user internal surface of RPV. The proportion of axial cracks and
interface for FERMAT. There are tabs for general information, circumferencial cracks is considered to be 1:1 for embedded
plant information, transient, operation history, welding cracks in base metal. The direction of embedded cracks in weld
information, crack distribution, chemical composition, radiation part is designed to coincide with the direction of the fusion line.
condition, 𝑅𝑇𝑁𝐷𝑇 , fracture toughness, and post-processes. When distributing cracks, you can enter settings for different
methods of distribution or tables created by VFLAW [10],
which is code for generating cracks based on crack information
of canceled RPV in the United States.
The stress intensity factor is assessed by using the method
specified in the Japanese Fitness-for-Service code [11] in
FERMAT. The approach is based on the influence function
method, which uses coefficients obtained by a polynomial
approximation of the stress distribution as input. By using
JAEA's approach [12], the coefficient table for embedded
cracks was expanded. Stress distribution with the discontinuous
part on the interface of the base metal and clad part (described
in Section 3.2) cannot be accurately approximated by a
polynomial function. Therefore, stress distribution in clad part
FIGURE 1: EXAMPLE OF THE GRAPHICAL USER is processed as shown in Figure 2 in FERMAT code. Offset in
INTERFACE OF FERMAT stress distribution of clad part (the red area) is converted into
the first-degree polynomial equation with equivalent internal
Specifications study and verification for the crack arrest force and 0 offset on outer surface (the green line). Processed
model of FERMAT are carefully performed now since the distribution (the blue line) can be obtained by adding the first-
crack arrest model is complex. This paper primarily focuses on degree polynominal equation for the extrapolated distribution
crack initiation results and evaluation techniques. Three steps of stress in base metal (the red line). A polynomial function

2 Copyright © 2023 by ASME


may properly estimate the processed stress distribution since it 𝑐𝐾𝐼𝑐 = 4 (5)
lacks any discontinuous parts. The processed distribution is ∆𝑇𝑅𝐸𝐿𝐴𝑇𝐼𝑉𝐸 = 𝑇(𝑟, 𝜏) − 𝑅𝑇𝑁𝐷𝑇_𝐶 (6)
used for evaluation of stress intensity factor of cracks across
clad part and base metal, i.e. surface cracks. Where 𝑅𝑇𝑁𝐷𝑇_𝐶 is defined by the following equations [13, 14].

𝑅𝑇𝑁𝐷𝑇_𝐶 = 𝑅𝑇𝑁𝐷𝑇(0) − ∆𝑅𝑇𝑒𝑝𝑖𝑠𝑡𝑒𝑚𝑖𝑐 + ∆𝑅𝑇𝑁𝐷𝑇 (7)


Original distribution
∆𝑅𝑇𝑒𝑝𝑖𝑠𝑡𝑒𝑚𝑖𝑐 = −15.6 + 67.56[− ln(1 − 𝑃𝑐 )]1⁄4.31 (8)
Extraporated distribution
Converted distribution Where 𝑅𝑇𝑁𝐷𝑇(0) [°C] is initial 𝑅𝑇𝑁𝐷𝑇 , ∆𝑅𝑇𝑒𝑝𝑖𝑠𝑡𝑒𝑚𝑖𝑐 [°C] is
Stress [MPa]

Processed distribution the term expressing epistemic uncertainty, ∆𝑅𝑇𝑁𝐷𝑇 [°C] is the
degree of 𝑅𝑇𝑁𝐷𝑇 shift caused by radiation embrittlement, 𝑃𝑐
is cumulative probability. 𝑅𝑇𝑁𝐷𝑇(0) is sampled based on user
input distribution. ∆𝑅𝑇𝑁𝐷𝑇 is basically calculated based on
Clad Base metal JEAC4201 in FERMAT as specified in JEAG4640. In addition,
offset and variation of prediction method specified in
JEAC4201 is also considered in ∆𝑅𝑇𝑁𝐷𝑇 after calculation
based on JEAC4201. The value calculated based on JEAC4201
Position [mm] (hereinafter referred to as ∆𝑅𝑇𝑁𝐷𝑇_𝐽𝐸𝐴𝐶4201 ) is calculated from
FIGURE 2: SCHEMATIC FIGURE OF PROCESSING METHOD flux, fluence, Cu contents, and Ni contents. Based on the
OF STRESS DISTRIBUTION IN CLAD PART distribution of user input, those parameters are sampled. In
FERMAT, flux distribution versus the beltline region's
2.2 Evaluation of fracture toughness and CPI configuration and neutron attenuation up till the cracks' tips can
A schematic figure of the calculation process for CPI is be taken into account.
shown in Figure 3. Fracture toughness 𝐾𝐼𝑐 [MPa√m] can be
calculated by following equations. 150
KIc
1⁄𝑐𝐾𝐼𝑐
𝐾𝐼𝑐 = 𝑎𝐾𝐼𝑐 + 𝑏𝐾𝐼𝑐 [− ln(1 − 𝛷𝐾𝐼𝑐 )] (1) KI
100
Where, 𝛷𝐾𝐼𝑐 is the cumulative probability, 𝑎𝐾𝐼𝑐 、 𝑏𝐾𝐼𝑐 are Φ4
Φ5 Φ3
values defined as functions of ∆𝑇𝑅𝐸𝐿𝐴𝑇𝐼𝑉𝐸 , 𝑐𝐾𝐼𝑐 is constant, 𝑇
KI , KIc

[°C] is the material temperature on the crack tip, 𝑅𝑇𝑁𝐷𝑇_𝐶 [°C] Φ6 K4 Φ2


50 K5 K3
is the reference temperature for nil-ductile transition. Equation Φ1
(1) is derived from the following equation [4]. K6 K2
Time evolution K1
𝐾𝐼𝑐 −𝑎𝐾𝐼𝑐 𝑐𝐾𝐼𝑐 0
𝛷𝐾𝐼𝑐 = 1 − exp [− ( ) ] (2) 0 100 200 300
𝑏𝐾𝐼𝑐
Temperature [°C]
𝛷𝐾𝐼𝑐 for 𝐾𝐼𝑐 curve crossing on cirtain evaluation point of FIGURE 3: SCHEMATIC FIGURE OF CPI EVALUATION
stress intensity factor 𝐾𝐼 is obtained by substituting 𝐾𝐼𝑐 = 𝐾𝐼
for equation (2). For example, 𝛷𝐾𝐼𝑐 = 𝛷5 is obtained by 2.3 Evaluation of FCI
substituting 𝐾𝐼𝑐 = 𝐾5 for equantion (2) for the condition shown Calculating the conditional probability of crack initiation
in Figure 3. 𝛷𝐾𝐼𝑐 corresponds with 𝐾𝐼𝑐 curve as indicated in for all cracks and all events using the aforementioned
this example. 𝛷𝐾𝐼𝑐 takes maximum value when 𝐾𝐼𝑐 curve techniques is the first step in the procedures to acquire FCI for
touches the curve of 𝐾𝐼 like Figure 3. 𝛷𝐾𝐼𝑐 in this case, is the entire RPV. The probability of crack initiation (𝑃𝐼) for each
occurrence is calculated in the second phase using the
equivalent to CPI. FERMAT calculates 𝛷𝐾𝐼𝑐 for all evaluation
following equations. [3]
points of 𝐾𝐼 , and determine CPI as the maximum value of
𝛷𝐾𝐼𝑐 . 𝑎𝐾𝐼𝑐 , 𝑏𝐾𝐼𝑐 and 𝑐𝐾𝐼𝑐 are input by users on the graphical 𝑃𝐼 = 1 − (1 − 𝐶𝑃𝐼1 ) × (1 − 𝐶𝑃𝐼2 ) × ⋯ × (1 − 𝐶𝑃𝐼𝑛 ) (9)
user interface in FERMAT. The analyses in this paper make use
of the following equations, which are stated in commentary Where 𝐶𝑃𝐼𝑖 is CPI for crack No. 𝑖, 𝑛 is the number of cracks
section of JEAG4640 as equations for Japanese RPV [3]. in the area to be assessed. FCI is determined by summing up
𝑃𝐼 multiplied by the average of occurence frequency of each
𝑎𝐾𝐼𝑐 = 13.18 + 6.71 ∙ exp[0.0337 ∙ ∆𝑇𝑅𝐸𝐿𝐴𝑇𝐼𝑉𝐸 ] (3) event sampled from cumulative probability deistribution for all
𝑏𝐾𝐼𝑐 = 15.88 + 42.21 ∙ exp[0.0121 ∙ ∆𝑇𝑅𝐸𝐿𝐴𝑇𝐼𝑉𝐸 ] (4) events.

3 Copyright © 2023 by ASME


3. RESULTS AND DISCUSSION TABLE 1: Main analysis conditions
Results of FERMAT and PASCAL4 were compared for Category Item Value
model cases targeting Japanese RPV. Comparisons were made Inner radius (to base metal) 2000 mm
for FCI and the methods used to obtain it, such as deterministic
analyses. Temperature distribution, stress distribution, and RPV Clad thickness 5.5 mm
stress intensity factor are evaluated as deterministic analyses. geometry Base metal wall thickness 200 mm
Comparison of findings with other codes is one of techniques Height 4000 mm
used for validation in the PFM field [15].
SBLOCA TH003, TH114
3.1 Analysis conditions MBLOCA TH007
Analysis conditions used in analyses in this study were LBLOCA TH056
demonstrated in Table 1. The geometry of RPV, material
MSLB TH103
properties, residual stress, radiation condition, chemical Transient
composition, and initial 𝑅𝑇𝑁𝐷𝑇 were determined based on the TH60, TH71, TH94,
example of analysis conditions written in the appendix of TH97, TH123,
JEAG4640 [3]. To determine the specific analytical conditions, SOV
TH126, TH129,
JAEA’s data for the Japanese model plant [14] was also
consulted. Each transient number is related to the identification TH130
number of thermal-hydraulic analyses performed in the United Orientation Circumferencial
States [16]. Distribution for depth and aspect ratio of cracks Surface
Depth 6.5 mm
were basically determined by VFLAW data. Based on data cracks
from a Japanese model, the depth of the surface crack was fixed Aspect ratio 2, 6, 10, 100
at 6.5mm [14]. Flux and fluence depicted in Table 1 are values Circumferencial
on inner surface of RPV. Consideration is given to neutron Orientation
and Axial
attenuation before they reach the crack tip. Although FERMAT
Initial Determined by
can take into account warm pre-stress and the flux distribution Depth
against the configuration of the beltline region, they are not cracks VFLAW data
Embedded
taken into account in the analyses in this study. Mean values of 1.125, 1.375, 1.75,
offset (−1.1°C) and standard deviation (8.9°C) in prediction cracks
Aspect ratio 2.5, 3.5, 4.5, 5.5, 7, 9,
method of radiation embrittlement based on JEAC4201[2] are
considered in analyses. 12.5, 20
Latin hypercube sampling is adopted in PASCAL4. This Inner 3/8 of base
method helps PASCAL4 to converge FCI and TWCF with the Position
metal thickness
small number of RPV samples. 1000 was chosen as the number
of RPV samples by JAEA, which is developer of PASCAL4 Neutron Mean value 7×1019 n/cm2
[14]. For FERMAT, dependence of FCI on the number of RPV fluence
samples was investigated and shown in Figure 4. Result in Radiation Std. deviation 13.1% of mean value
(E>1MeV)
Figure 4 suggests that 1000 is not enough for the numer of RPV condition
samples for FERMAT to converge results sufficiently. The Neutron flux(E>1MeV) 4.62×1010 n/cm2/s
difference in required number of RPV samples can be Radiation temperature 288 °C
explained by difference between Latin hypercube sampling and
Mean value 0.16 wt%
Monte Carlo method, which is adopted in FERMAT. This Base metal
paper shows resuls with 1000 and 10000 samples for FERMAT Cu Std. deviation 0.01 wt%
and results with 1000 samples for PASCAL4. content Mean value 0.14 wt%
Table 2 displays the analysis parameters used for Weld metal
Std. deviation 0.01 wt%
deterministic analyses. Deterministic evaluations for the small
break loss of coolant accident (SBLOCA) transient, including Mean value 0.61 wt%
Base metal
assessments of temperature distribution, stress distribution, and Ni Std. deviation 0.02 wt%
stress intensity factor, were performed. Figures 5, 6, and 7 content Mean value 0.80 wt%
illustrate, respectively, the transients of temperature, heat Weld metal
transfer coefficient between coolant and internal wall of RPV, Std. deviation 0.02 wt%
and pressure for SBLOCA [16]. A surface crack in base metal Mean value -5.0 °C
and an embedded crack in weld metal were chosen as examples Base metal
Initial Std. deviation 9.4 °C
for deterministic analyses as shown in Table 2.
RTNDT Mean value -50.0 °C
Weld metal
Std. deviation 9.4 °C

4 Copyright © 2023 by ASME


TABLE 2: Analysis conditions for deterministic

Heat transfer coefficient [W/(m2·K)]


30000
analyses

Category Item Value


20000
Transient SBLOCA TH003
Location Base metal
Surface Direction Circumferencial 10000
cracks Depth 6.5mm
Aspect ratio 6
0
Location Weld metal
0 5000 10000 15000
Initial Direction Axial Time [sec]
cracks 5% of the RPV wall FIGURE 6: TIME HISTORY OF HEAT TRANSFER
Depth COEFFICIENT BETWEEN COOLANT AND INTERNAL WALL
Embedded thickness
OF RPV IN SBLOCA TRANSIENT
cracks Aspect ratio 2.5
0.1 mm from the 20

Position interface of clad and


base metal
Pressure [MPa]
1.0E-4 10
FCI [/ry]

1.0E-5 0
0 5000 10000 15000
Time [sec]
FIGURE 7: TIME HISTORY OF INTERNAL PRESSURE IN
SBLOCA TRANSIENT
1.0E-6
0 5000 10000 15000 20000 3.2 Results of deterministic analysis
The number of RPV samples The distribution of temperature, axial stress, and hoop
FIGURE 4: DEPENDENCE OF FCI ON THE NUMBER OF RPV stress evaluated by FERMAT and PASCAL4 were shown in
SAMPLES FOR FERMAT Figure 8, Figure 9, and Figure 10, respectively. Those
distributions were evaluated for 60sec, 3000sec, and 15000sec
300 from event occurence. The result of temperature distribution by
FERMAT and PASCAL4 in Figure 8 are well corresponding
with each other for all evaluation times. Stress distribution
evaluated by FERMAT using analytical solution based on
Temperature [°C]

200 linear elastic theory is well corresponding with stress


distribution evaluated by PASCAL4 using finite element
analysis in Figure 9 and Figure 10.
The discontinuous part of stress distribution on the
100
interface of the base metal and clad part mentioned in Section
2.1 are observed in both axial stress in Figure 9 and hoop stress
in Figure 10.
0
0 5000 10000 15000
Time [sec]
FIGURE 5: TIME HISTORY OF COOLANT TEMPERATURE IN
SBLOCA TRANSIENT

5 Copyright © 2023 by ASME


400 Japanese Fitness-for-Service code solution [11] is employed in
FERMAT (60sec) PASCAL4 (60sec) FERMAT. Using the technique described in Section 2.1, stress
FERMAT (3000sec) PASCAL4 (3000sec)
FERMAT (15000sec) PASCAL4 (15000sec) distribution in the clad component is taken into account for
300
Temperature [°C]

surface cracks. On the other hand, the solution of CEA [17]


based on the existence of clad parts is adopted in PASCAL4 for
surface cracks. As a result, FERMAT's method for taking into
200
account clad parts and solution to the stress intensity factor
differ from PASCAL4's addressing surface cracks. However,
100 the differences in the stress intensity factor of surface cracks
derived by those two codes were not as substantial. Stress
Clad Base metal intensity factor for embedded cracks shown in Figure 12 was
0 also well corresponding with each other.
-50 0 50 100 150 200 250
Position [mm] 80

]
FIGURE 8: DISTRIBUTION OF TEMPERATURE FERMAT
70
PASCAL4
60
400

Stress intensity factor [


FERMAT (60sec) 50
FERMAT (3000sec)
300
FERMAT (15000sec) 40
Axial stress [MPa]

PASCAL4 (60sec)
200 PASCAL4 (3000sec) 30
PASCAL4 (15000sec)
20
100
10
0
0
0 5000 10000 15000
-100
Clad Base metal Time [sec]
-200 FIGURE 11: TIME HISTORY OF STRESS INTENSITY FACTOR
-50 0 50 100 150 200 250 FOR A SURFACE CRACK
Position [mm]
FIGURE 9: DISTRIBUTION OF AXIAL STRESS 50
]

FERMAT
400 40 PASCAL4
FERMAT (60sec)
FERMAT (3000sec)
Stress intensity factor [

300 30
FERMAT (15000sec)
Hoop stress [MPa]

PASCAL4 (60sec)
200 PASCAL4 (3000sec)
PASCAL4 (15000sec) 20
100
10
0

0
-100
Clad Base metal 0 5000 10000 15000
-200 Time [sec]
-50 0 50 100 150 200 250 FIGURE 12: TIME HISTORY OF STRESS INTENSITY FACTOR
Position [mm] FOR AN EMBEDDED CRACK
FIGURE 10: DISTRIBUTION OF HOOP STRESS
3.3 Results of probabilistic analysis
The stress intensity factor of a surface crack in base metal Variation of 𝑅𝑇𝑁𝐷𝑇_𝐶 derived by equation (7) consists of 2
and an embedded crack in weld metal for SBLOCA was types of variations. One is a variation of ∆𝑅𝑇𝑁𝐷𝑇_𝐽𝐸𝐴𝐶4201
evaluated by FERMAT and PASCAL4 with analysis conditions calculated based on JEAC4201 brought on by crack tip
of Table 2. Evaluated stress intensity factors are illustrated in placement, radiation circumstances, and chemical composition
Figure 11 and Figure 12 for a surface crack and an embedded (Cu and Ni contents). The other is variation of 𝑅𝑇𝑁𝐷𝑇_𝐶 −
crack respectively. In evaluation by both codes, the same ∆𝑅𝑇𝑁𝐷𝑇_𝐽𝐸𝐴𝐶4201 , which is caused by variation of 𝑅𝑇𝑁𝐷𝑇(0) ,
residual stress distribution was taken into account. For ∆𝑅𝑇𝑒𝑝𝑖𝑠𝑡𝑒𝑚𝑖𝑐 and variation of ∆𝑅𝑇𝑁𝐷𝑇 considered after
determining the stress intensity factor for surface cracks, the

6 Copyright © 2023 by ASME


calculation of ∆𝑅𝑇𝑁𝐷𝑇_𝐽𝐸𝐴𝐶4201 . The variation of 𝑅𝑇𝑁𝐷𝑇_𝐶 − 1.0E+0
∆𝑅𝑇𝑁𝐷𝑇_𝐽𝐸𝐴𝐶4201 is depicted for base metal and weld metal in FERMAT (10000 samples) VS PASCAL4 (1000 samples)
Figure 13 and Figure 14 respectively. Mean values of FERMAT (1000 samples) VS PASCAL4 (1000 samples)

Probability of crack initiation (FERMAT)


𝑅𝑇𝑁𝐷𝑇(0) , ∆𝑅𝑇𝑒𝑝𝑖𝑠𝑡𝑒𝑚𝑖𝑐 and offset in the prediction method of 1.0E-1
JEAC4201 are also considered in the horizontal axes of Figure
13 and Figure 14. Although histogram for FERMAT with
10000 samples are smoother than that with 1000 samples,
distribution is sufficiently expressed with 1000 samples. 1.0E-2
Significant difference depending used code was not found in
the results with 1000 samples in Figure 13 and Figure 14.
1.0E-3
0.05
FERMAT (10000 samples)
0.04 FERMAT (1000 samples)
PASCAL4 (1000samples) 1.0E-4
Frequency

0.03

1.0E-5
0.02
1.0E-5 1.0E-4 1.0E-3 1.0E-2 1.0E-1 1.0E+0
Probability of crack initiation (PASCAL4)
0.01
FIGURE 15: COMPARISON OF PROBABILITY OF CRACK
INITIATION
0.00
-200 -150 -100 -50 0 50 100 FCI evaluated by FERMAT and PASCAL 4 is shown in
RTNDT_C - ΔRTNDT_JEAC4201 [°C] Figure 16. Since FCI is generated for every EFPY by a single
FIGURE 13: VARIATION OF RTNDT_C FOR BASE METAL execution of FERMAT, FCI for each EFPY are plotted for
FERMAT in Figure 16. PASCAL4 assessed FCI for 8, 16, 32,
0.05 and 48 EFPY. For every EFPY depicted in Figure 16,
FERMAT (10000 samples) FERMAT’s results were lower than PASCAL4’s. However,
0.04 FERMAT (1000 samples) they barely differed from one another. FCI for 48 EFPY was
PASCAL4 (1000 samples) slightly changed by the number of samples as shown in Figure
4. However, results of FERMAT with 10000 samples
Frequency

0.03
overlapped with those with 1000 samples except quite low
EFPY. In other words, the tendency of FCI was not
0.02
significantly changed depending on the number of RPV
samples.
0.01

1.0E-4
0.00
-200 -150 -100 -50 0 50 100
1.0E-5
RTNDT_C - ΔRTNDT_JEAC4201 [°C]
FIGURE 14: VARIATION OF RTNDT_C FOR WELD METAL 1.0E-6
FCI [/ry]

Figure 15 shows a visualization of the probability of crack


1.0E-7
initiation on 48 EFPY for each transient after FERMAT and
PASCAL4 evaluations. Results of FERMAT with 10000 FERMAT (10000 samples)
samples overlapped with those with 1000 samples, and 1.0E-8 FERMAT (1000 samples)
significant difference was not observed depending on the PASCAL4 (1000 samples)
number of samples. When compared to PASCAL4, FERMAT's 1.0E-9
estimate of the probability of crack initiation was a tiny bit 0 10 20 30 40 50 60
lower in Figure 15. However, their tendency for each transient Effective full power year [EFPY]
was systematically corresponding with each other. These FIGURE 16: COMPARISON OF FCI
variations may result from variations in the stress intensity
factors indicated in Figures 11 and 12. However, There was 4. CONCLUSION
very little difference between the results of the two codes A new PFM analysis code FERMAT is developed aiming
illustrated in Figure 15. actual application of PFM for the integrity assessment of RPVs

7 Copyright © 2023 by ASME


by CRIEPI in Japan. In this study, the evaluation procedure and [12] K. Lu, J. Katsuyama, Y. Li, “Stress Intensity Factor
outcomes for FCI were compared to those for PASCAL4. Solutions for Subsurface Flaws in Plates Subjected to
The distribution of temperature and stress evaluated by Polynomial Stress Distributions”, ASME Pressure Vessels
FERMAT was well corresponding with those evaluated by and Piping Conference, PVP2016-63479, (2016).
PASCAL4. Stress intensity factor calculated by those two [13] J. Katsuyama, K. Osakabe, S. Uno, Y. Li, S. Yoshimura,
codes was well corresponding with each other. The significant “Guideline on Probabilistic Fracture Mechanics Analysis
difference was not observed in sampled parameters related to for Japanese Reactor Pressure Vessels”, ASME Pressure
𝑅𝑇𝑁𝐷𝑇 . The probability of crack initiation and FCI assessed by Vessels and Piping Conference, PVP2017-65921, (2017).
FERMAT and PASCAL4 were well corresponding each other. [14] Japan Atomic Energy Agency, “Sophistication of Ageing
Those results indicate that FERMAT and PASCAL4 provide Management Technical Evaluation on Nuclear Power
equivalent results for assessment of crack initiation in RPV. Stations, Sophistication of Methods for Integrity
Assessment of Primary Components of Nuclear Reactors”,
REFERENCES Nuclear Regulatory Authority, 291203, (2018).
[1] Japan Electric Association Code, “Verification Method of [15] L. Hund, J. Lewis, N. Martin, M. Sarr, D. Brooks, A.
Fracture Toughness for In-Service Reactor Pressure Zhang, R. Dingrecille, A. Eckert, J. Mullins, P. Raynaud,
Vessel”, JEAC 4206-2016, (2016). D. Rudland, D. Dijamco, S. Cumblidge, “Technical Basis
[2] Japan Electric Association Code, “Method of Surveillance for the use of Probabilistic Fracture Mechanics in
Tests for Structural Materials of Nuclear Reactors”, JEAC Regulatory Applications” NUREG/CR-7278, U.S. Nuclear
4201-2007, (2013 addenda), (2013) Regulatory Commission, (2022).
[3] Japan Electric Association Guide, “Guideline for [16] W.C. Arcieri, R.M. Beaton, C.D. Fletcher, D.E. Bessette,
Calculating Failure Frequency of Reactor Pressure Vessel “RELAP5 Thermal Hydraulic Analysis to Support PTS
Based on Probabilistic Fracture Mechanics”, JEAG 4640- Evaluations for the Oconee-1, Beaver Valley-1, and
2018, (2018). Palisades Nuclear Power Plants”, NUREG/CR-6858, U.S.
[4] T. Dickson, M. Smith, A. Dyszel, "Fracture Analysis of Nuclear Regulatory Commission, (2004).
Vessels – Oak Ridge FAVOR v20.1.12 Theory and [17] S. Marie, S. Chapuliot, “Improvement of the Calculation
Implementation of Algorithms, Methods, and Correlations", of the Stress Intensity Factors for Underclad and Through-
Technical Letter Report, TLR-RES/DE/REB-2021-03, clad Defects in a Reactor Pressure Vessel Subjected to a
(2021) Pressurized Thermal Shock”, International Journal of
[5] J. Katsuyama, K. Masaki, Y. Miyamoto, Y. Li, “User’s Pressure Vessels and Piping, 85, 517-531, (2008).
Manual and Analysis Methodology of Probabilistic Fracture
Mechanics Analysis Code PASCAL4 for Reactor Pressure
Vessel”, JAEA-Data/Code 2017-015, (2018).
[6] Y. Li, S. Uno, M. Koichi, J. Katsuyama, T. Dickson, M.
Kirk, “Verification of Probabilistic Fracture Mechanics
Analysis Code through Benchmark Analyses”, ASME
Pressure Vessels and Piping Conference, PVP2018-84963,
(2018).
[7] Y. Li, G. Katsumata, K. Masaki, S. Hayashi, Y. Itabashi, M.
Nagai, M. Suzuki, Y. Kanto, “Verification of Probabilistic
Fracture Mechanics Analysis Code for Reactor Pressure
Vessel”, Journal of Pressure Vessel Technology, 143,
041501, (2021).
[8] S. Timoshenko, J.N. Goodier, “Theory of Elasticity. 2nd.
Ed.”, (1951).
[9] Stephane Marie, “Analytical Expression of the Thermal
Stresses in a Vessel or Pipe with Cladding Submitted to
Any Thermal Transient”, International Journal of Pressure
Vessels and Piping 81, 303–312, (2004).
[10] S.R. Simonen, S.R. Doctor, G.J. Schuster, P.G. Heasler.,
“A Generalized Procedure for Generating Flaw Related
Inputs for the FAVOR Code,” NUREG/CR-6817,
Revision 1, U.S. Nuclear Regulatory Commission, (2013).
[11] The Japan Society of Mechanical Engineers, “Rules on
Fitness-for-Service for Nuclear Power Plants”, JSME S
NA1-2012, The Japan Society of Mechanical Engineers,
(2012).

8 Copyright © 2023 by ASME

You might also like