Practical Application of Age-Dependent Reliability Models and Analysis of Operational Data
Piping Service Experience Related to
Aging in OECD Pipe Failure Data
Exchange (OPDE) Database
Oct. 6, 2005
Sun Yeong Choi*, Joon Eon Yang*, and Young Hwan Choi**
*Korea Atomic Energy Research Institute (sychoi@kaeri.re.kr)
**Korea Institute of Nuclear Safety
1
Korea Atomic Energy Research Institute
Practical Application of Age-Dependent Reliability Models and Analysis of Operational Data
Backgrounds
Aging PSA and Passive System Reliability Study in KAERI
We have performed
Participation in OPDE Project to collect piping failure data
Development of DB for the number of weld of safety piping
systems in Korean NPPs for piping reliability analysis
Analysis for piping damage frequency / rupture frequency
We are going to
Investigate the OPDE database for the piping aging trend
Develop the estimation method for aging parameter
2
Korea Atomic Energy Research Institute
Practical Application of Age-Dependent Reliability Models and Analysis of Operational Data
Purpose of Study
Purpose of Study
To Get Insights for Piping Aging Trend from OPDE Database
How to obtain piping aging trend from OPDE DB
What is the meaning of the results obtained from piping aging trend
analyses
Methods
Understanding of Data Completeness in OPDE DB
Selection of Data from OPDE DB for Aging Trend Analysis
Various Aging Trend Analyses with the Selected Data
3
Korea Atomic Energy Research Institute
Practical Application of Age-Dependent Reliability Models and Analysis of Operational Data
OPDE (OECD Piping Failure Data Exchange)
Objective of OPDE Project
To develop a well structured, comprehensive database on pipe failure
events
To enable project member organizations to utilize the collected data for
various application fields including PSA, Aging Management, or LBB
OPDE Project
May 2002 ~ September 2008
Project Members: 12 Countries
OPDE Database
Scope of OPDE DB
Non-through Wall Defect
Through-Wall Defect without Active Leakage
Leak / Rupture / Severance
3,244 records (by May 2005) 4
Korea Atomic Energy Research Institute
Practical Application of Age-Dependent Reliability Models and Analysis of Operational Data
Structure of OPDE DB
Regulatory inspection reports
Dev iation Reports
LERs / AOs / ROs / WOs Plant Operating Records / Logs
RIISI Applications ISI Summary Reports ISI Program Plans / ISI Database
Isometric Drawings ASME Section XI Relief Requests Piping Design Inf ormation
Population Data Event Data Service History
Plant A&E Sy stem Component Ev ent Date Af f ected Sy stem(s) Component ISI FAC Program
Locations per isometric Metallurgical Narrativ e w/description of Method of Water Chemistry /
drawing(s) data impact on operations Detection H2-injection (BWR)
Fabrication data Installation WOR IHSI Replacement
Weld data data
Data Retrieval & Validation
Statistical
Analysis Classif ication according to Failure
attributes and inf luences Modes
Identif y degradation mechanism(s)
Data
Analysis
Applications Piping Reliability Parameter dB
PSA PFM RIISI Conditional Rupture
Leak f requencies by
Probabilities by Attributes &
Attributes & Inf luence Factors
Inf luence Factors
Statistical Distribution Parameters 5
Korea Atomic Energy Research Institute
Practical Application of Age-Dependent Reliability Models and Analysis of Operational Data
Characteristics of OPDE DB
Data Source of OPDE DB
Data from SKI-PIPE Database
The SKI-PIPE database collected piping failure occurred up
to 1998 from 1970
Data from 29 Countries (including 12 OPDE Member
Countries)
Data from OPDE Project Members
OPDE project mainly focused on piping failures occurred
after 1995
To use OPDE DB for piping aging trend analysis,
understanding of data completeness is required
6
Korea Atomic Energy Research Institute
Practical Application of Age-Dependent Reliability Models and Analysis of Operational Data
Data Completeness of OPDE DB (I)
Data Collection for OPDE DB
Start of
End of Base Date
SKI-PIPE
SKI-PIPE (12/31/04)
′1960 ′1970 ′1990 ′2000 ′2008
(1) First Rx-Criticality
Cease of Data
(2) Collection
(3) Pipe Failure
(4) First Record in
OPDE DB
(5)
Base Date : Closing Date of Data Collection for Analysis 7
Korea Atomic Energy Research Institute
Practical Application of Age-Dependent Reliability Models and Analysis of Operational Data
Data Completeness of OPDE DB (II)
Case (1) and (2)
Piping failures have been collected successively during both
periods of the SKI-PIPE and OPDE DB
Case (3) and (4)
Data collection ceased at the end date of the SKI-PIPE
Collection of pipe failures from 17 countries stopped after 1998
Case (5)
There exists a long interval between the first operation date and
the event occurrence date stored into the DB for the first time
For some plants, there is no pipe failure record occurred during
the period of the SKI PIPE project
8
Korea Atomic Energy Research Institute
Practical Application of Age-Dependent Reliability Models and Analysis of Operational Data
Data Scope for Aging Trend Analysis
1,424 Records from 212 PWRs
Plant from OPDE Member Countries (Except for Canada)
Plant Starting Operation after Jan. 1970
Total Plant Operation year: 4609.7 yr
Data having Clear Failure Description
Plant Operating BE CH CZ DE ES FI FR JP KR SE US
Year
0 ~ 10 3 4 1 5 1 14
10 ~ 20 2 4 5 3 22 8 6 23 73
20 ~ 30 4 1 7 3 2 32 10 3 2 31 95
≥30 1 1 3 4 1 20 30
Total 7 2 7 15 6 2 58 23 14 3 75 212
9
Korea Atomic Energy Research Institute
Practical Application of Age-Dependent Reliability Models and Analysis of Operational Data
Method 1 for Aging Trend Analysis
Description
Three Groups Categorized by Different Base Date (End date of 1984, 1994,
and 2004)
Common Method for Data Upgrade for Component Reliability Analysis
We can get more data when base data is Dec. 31, 2004
′1970 ′1984 ′1994 ′2004
(I)
125 Plants (819.7 yr) (II)
198 Plants (2583.7 yr) (III)
212 Plants (4609.7 yr)
Plant Operation 10
Korea Atomic Energy Research Institute
Practical Application of Age-Dependent Reliability Models and Analysis of Operational Data
Results of Method 1
Pipe Leak Trend Pipe Crack & Wall Thinning Trend
0 .6 0 0.05
l pi
Al P i ng l nni
W al Thi ng
O ccurrence Frequency (/yr)
0 .5 0 ≤1 i nch
)
ack
Cr
equency (/yr
> 1 and ≤ 4 i
nch 0.04
0 .4 0 >4 i nch
0.03
0 .3 0
O ccur ence Fr
0.02
0 .2 0
r
0.01
0 .1 0
0 .0 0 0.00
1984 1994 2004 1984 1994 2004
B ase D ate e
Base Dat
Leak frequency dominates the whole piping failure frequency
Ratio of Leak frequency and other frequencies is more than a factor of 10
There is no significant trend in crack and wall thinning frequencies
Leak frequency is decreasing as data collection progresses
11
Korea Atomic Energy Research Institute
Practical Application of Age-Dependent Reliability Models and Analysis of Operational Data
Review of Results from Method 1
Aging Effect
It is very difficult to extract aging effect from the results
Each group categorized by base date consists of younger plants
and older plants
Operation experiences of older plants and younger plants are mixed
in each group
12
Korea Atomic Energy Research Institute
Practical Application of Age-Dependent Reliability Models and Analysis of Operational Data
Method 2 for Aging Trend Analysis
Description
Three groups are categorized by different total plant operation years
(0~15, 15~25, 25~ 35) with Dec.31 2004 as the starting point
Plants in Group (3) are the oldest plants, while plants in Group (1) are
the youngest plants
Jan. 1, 1970 Jan. 1, 1980 Jan. 1, 1990 Dec. 31, 2004
Group (3) Group (2) Group (1)
63 Plants (1863.0 yr) 119 Plants (2427.0 yr) 28 Plants (284.6 yr)
25~35 years 15~25 years 0~15 years : Plant Operating Years
1970’s 1980’s 1990’s (up to 2004) : Plant Operation Starting Date
13
Korea Atomic Energy Research Institute
Practical Application of Age-Dependent Reliability Models and Analysis of Operational Data
Results of Method 2
Pipe Leak Trend Pipe Crack & Wall Thinning Trend
0.50 0.05
l pi
Al Pi ng ack
Cr
0.40 ≤1 i nch 0.04 l nni
W al Ti ng
equency (/yr)
nch
> 1 and ≤4 i
equency (/yr)
0.30 >4i nch 0.03
O ccur ence Fr
0.02
Occurence Fr
0.20
r
r
0.10 0.01
0.00
0.00
0~ 15 15~ 25 25~ 35
0~15 15~25 25~35
aton
O per i Year
aton
Oper i Year
Leak frequency of older plants is generally higher than that of
younger plants.
Ratio of Leak frequency and other frequencies is more than a
factor of 10
14
Korea Atomic Energy Research Institute
Practical Application of Age-Dependent Reliability Models and Analysis of Operational Data
Review of Results from Method 2
Aging Effect
Although the leak frequency of older plants is generally
higher than that of younger plants, it does not express the
aging effect.
The leak frequency of older plants during the plant early
operation period is higher than that of younger plants
The reasons may be low levels of design, construction, and
maintenance in early operation period of older plants.
15
Korea Atomic Energy Research Institute
Practical Application of Age-Dependent Reliability Models and Analysis of Operational Data
Method 3 for Aging Trend Analysis
Description
Plant operation starting times of all the plants are moved to
identical time point (Jan. 1, 1970)
Six groups are categorized by 5 calendar years
′1970 0~5 yr 5~10 yr 10~15 yr 15~20 yr 20~25 yr 20~30 yr
212 Plants 207 Plants 196 Plants 179 Plants 112 Plants 61 Plants
(1044.5 yr) (1007.5 yr) (948.4 yr) (736.2 yr) (430.9 yr) (205.2 yr) 16
Korea Atomic Energy Research Institute
Practical Application of Age-Dependent Reliability Models and Analysis of Operational Data
Results of Method 3
Pipe Leak Trend Pipe Crack & Wall Thinning Trend
35
0. 0.1
3
0. Crack
0.08 l hi ng
W al-t nni
0.
25 l pi
Al Pi ng
equency (/ )
)
yr
equency (/yr
≤1 i nch
2
0. <1 and ≤4 0.06
<4
15
0.
O ccurence Fr
Occurence Fr
0.04
1
0.
r
r
05
0.
0.02
0
0
0-5 5 - 10 10 - 15 15 - 20 20 - 25 25 - 30
0-5 5 - 10 10 - 15 15 - 20 20 - 25 25 - 30
me Cal
Ti ( endarYear)
me endarYear
Ti (Cal )
Crack frequency increases progressively
The trend of leak frequency shows typical bath-tub shape
The leak frequency is relatively high in early operation period and
decreases by 20-year operation and then the leak frequency
increases 17
Korea Atomic Energy Research Institute
Practical Application of Age-Dependent Reliability Models and Analysis of Operational Data
Results of Method 3 (Modified)
Pipe Leak Trend
0.7
0-1 5
O ccurrence Frequency (/yr)
0.6 15-25 Leak data is categorized into
0.5 25-35
three groups by plant
0.4
operation years (0~15, 15~25,
0.3
25~35)
0.2
And then Method 3 is applied
0.1
0
in each group
0-5 5-10 10-15 15-20 20-25 25-30
m
Ti e
The leak frequency of the oldest plants decreases as the plants
ages and then increases slightly after 20-year operation
In the case of plants whose operation years are ranged from 15 to
25, the leak frequency is almost constant as plants age
18
Korea Atomic Energy Research Institute
Practical Application of Age-Dependent Reliability Models and Analysis of Operational Data
Review of Results from Method 3
Aging Effect
Based on the Method 3, the trend of leak frequency shows
typical bath-tub shape as the plants age
Considering the plant operation year (0~15, 15~25, 25~35),
the leak frequency of the oldest plants decreases by 20-
year operation and then increases slightly
19
Korea Atomic Energy Research Institute
Practical Application of Age-Dependent Reliability Models and Analysis of Operational Data
Conclusion
In this Study
We apply three kinds of analysis methods to figure out piping
aging trend from OPDE DB
The aging effect can be extracted by using Method 3 based on
the assumption of identical plant operation starting time
Further Works
Estimation of Aging Parameter
Application of Aging Parameter into Aging PSA
20
Korea Atomic Energy Research Institute