ML20059C801
"Draft Meeting" is not in the list (Request, Draft Request, Supplement, Acceptance Review, Meeting, Withholding Request, Withholding Request Acceptance, RAI, Draft RAI, Draft Response to RAI, ...) of allowed values for the "Project stage" property.
| ML20059C801 | |
| Person / Time | |
|---|---|
| Issue date: | 06/05/1993 |
| From: | Lewis H Advisory Committee on Reactor Safeguards |
| To: | Advisory Committee on Reactor Safeguards |
| References | |
| ACRS-2877, NUDOCS 9311020020 | |
| Download: ML20059C801 (36) | |
Text
0Q3-M77 CERTIFIED BY H.
LEWIS 6/5/93
SUMMARY
/ MINUTES OF THE ACRS SUBCOMMITTEE MEETING ON PROBABILISTIC RISK ASSESSMENT MAY 11, 1993 BETHESDA, MARYLAtO PURPOSE The ACRS Subcommittee on Probabilistic Risk Assessment (PRA) held a meeting on May 11, 1993 in Bethesda, Maryland. The purpose of this meeting was to discuss a draft report of the PRA Working Group that summarized the activities of the Group and provided guidance for the staff regarding the application of PRA. A copy of the meeting agenda and selected slides from the presentations are attached. The meeting began at 8:30 am and adjourned at 4:45 pm and was held entirely in open session. No written comments or requests for time to make oral statements were received from members of the public. The principal attendees were as follows:
ATTENDEES ACRS HRC. STAFF H.
Lewis, Chairman M. Cunningham, RES P.
Davis, Member H. VanderMolen, RES T.
Kress, Member C. Ryder, RES C. Michelson, Member W. Beckner, NRR R.
Seale, Member G. Apostolakis, Consultant E.
Wilkins, Member T.
Brown, Consultant C. Wylie, Member H. Martz, Consultant W.
Kerr, Consultant D.
Okrent, Consultant D. Ward, Consultant D.
Houston, Cognizant Staff Engineer 020011 Also in attendance were representatives of NUS, TVA and Bechtel.
m
- ; sn
((
9311020020 930605 Dk 1
PRA Meeting Minutes May 11, 1993 DISCUSSION Chairman's Openin7 Remarks In his opening remarks, Dr. Lewis noted that the staf f had provided a draft copy of a report by the PRA Working Group and a set of external reviewer's comments on an earlier draf t. He indicated that a copy of the earlier draft had not been provided so that the comments made by these reviewers were impossible to follow in regard to either the basis of the concern or the resolution.
STAFF PRESENTATION Overview of PRA Working Grouc Reoort - M.
Cunningham, NRC/RES Mr. Cunningham, Chairman of the PRA Working Group, discussed the general activities of the PRA Working Group (PRAWG).
In his discussion, he addressed the following topics:
o Introduction (Objectives, Scope and Tasks of the PRAWG.)
o Characteristics of Present Staff Uses of PRA o Overview of the PRAWG Draft Report o Summary of External Reviewer's Comments ar.d Resolution o Future Plans of the PRAWG In regard to future plans, he indicated that a workshop on terms and methods was scheduled for October 1993, guidance for adapting PRAs would be issued in December 1993, and the methodology for Level 1 to Level 3 transformations would be complete in December 1993. He further noted that the PRA Working Group would continue to function in the future.
He indicated that one of the PRAWG recommendations was an effort to bring consistency to risk assessment and risk management to all of the organizations within 1
the Federal Government.
Guidance for Issue Prioritization and Screening - H. VanderMolen, NRC/RES l
Dr. VanderMolen discussed the prioritization of issues in regard to ordering and screening. He stated that appropriate PRA logic diagrams should be used and that the analyst may use point
)
estimates and some conservatism in the calculations. In response to a question, he indicated that he did not know the definition of a point estimate.
He also indicated that Level 1 (core damage frequencies-CDF) could be transformed to Level 3 (risk) results. He ended with a discussion of an example f or Generic Safety Issue resolution. In this discussion, one of the items to be addressed i
was the principal accident sequences. It was pointed out to him that ISLOCA would not have been identified as a concern if it had been left to such an examination.
4 e
t i
PRA Meeting Minutes May 11, 1993 PRA Methods: Improvements and Recommendations - M. Cunningham, NRC/RES Mr. Cunningham discussed the staff survey results in regard to methodology. He indicated that the majority of the ~ staf f 's. PRA applications relied on adaptation of results from existing PRAs. He also indicated that the staff had no preference over any specific PRA or PRA method and that the majority of PRA applications are-Level 1.
He concluded with a set of recommendations by the PRAWG which included:
o Guidance for adapting PRA methods and results o Level 1 to Level 3 transformation capability o Development of PC-based PRA tools o Feasibility study of " roll-up" PRA models o Develop "living" PRA models and database o Develop sensitivity / uncertainty analysis tools He presented a schedule for these activities that showed completion in 1994.
1 PRA Knowledge. Skills and Trainina - M. Cunningham, NRC/RES In his introductory remarks, Mr. Cunningham indicated that this presentation would focus on Appendix C of the draft report. As such, two basic areas were to be addressed:
(1) PRA terms and methods and (2) job and task analysis studies. In response to a question, he indicated that Appendix C was not intended to be a textbook but rather it was intended to be used by those who are not expert in PRA but use PRA in a routine manner. For the second area, he indicated that the Office of Personnel (OP) was in charge of all NRC training programs and that the PRAWG would help to develop a comprehensive PRA training program. The PRAWG would also coordinate the OP programs and the programs at the Technical Training Center (Chattanooga, Tennessee).
Probability and Statistics - H. Martz, LANL Dr.
Martz discussed probability and statistics.
In regard to probability, he addressed the following:
o Frequentist versus subjectivist probability i
o Basic rules for calculating probability o Concept of point and interval estimates o Proper use and interpretation of best estimate In regard to statistics, he addressed the following topics:
o Bayesian versus classical statistics o Confidence versus probability intervals o Common statistical methods in PRA o Sources of reliability data For the areas / topics noted above, he discussed the guidance that was being developed for each.
PRA Meeting Minutes May 11, 1993 l
Accident Secuence and Reliability Analysis - G. Apostolakis, UCLA Dr. Apostolakis indicated that the objective of this section was to present basic probabilistic concepts and analytical tools useful in performing a Level 1 PRA. He discussed the analysis methods for component reliability, system reliability, accident sequences, l
dependent failures, human reliability, external events and sof tware reliability. He also discussed time-dependent modeling and the availability of reliability codes. In closing, he described one computer code: IRRAS 4.0 (Integrated Reliability and Risk Analysis System), an integrated PRA software tool that creates and analyzes fault trees and accident sequences using an IBM-compatible 1
microcomputer.
Expert Judgment - G. Apostolakis, UCLA Dr. Apostolakis next addressed the area of expert judgment. In regara to this, he discussed the following topics:
o Formal elicitation processes o Biases in probability elicitation o Methods for processing expert judgments o Evaluating the quality of judgments o Mathematical methods for combining expert judgments In concluding, he described the potential pitfalls of this process.
These included failure to use a sufficient breadth of expertise, imprecisely defined issues, and failure to carefully document rationale. In response to a question about the maximum entropy
]
process, he indicated that the process had a theoretical basis in i
the area of thermodynamics but did not have a basis in information theory.
1 Accident Progression and Risk Analysis - C. Ryder, NRC/RES Mr. Ryder addressed the section of Appendix C entitled, " Accident Progression and Risk Analysis. " He discussed the following topics:
4 o CDF to risk transformation o Accident progression o Radionuclide release and transport o Consequences-methods and guidance o Risk calculation and treatment of uncertainty o Deterministic calculations and codes A concern was expressed that the guidance for risk integration was in conflict with the provisions of the Safety Goal Policy, that is, one had to assess the differences between high consequence-low probability events and low consequence-high probability ones even j
though the bottom line risk values were the same. His response was couched along the lines of surrogate goals and a desire to understand how the plant behaved.
j
l I
j PRA Meeting Minutes May 11, 1993
-1 Uncertainty and Sensitivity Analysis - T. Brown, SNL Dr.
Brown discussed the three objectives of the section on uncertainty and sensitivity analysis. These were as follows to introduce: (1) concepts of uncertainty and' sensitivity analysis and the role they play in PRAs, (2) some of the more common methods used to propagate uncertainties through PRAs and (3) some of the formats used to display the results. In addition, he indicated that the selection of variables and treatment of correlation and dependencies are important factors in an uncertainty ana)ysis.and that uncertainty results cannot generally be extended from one plant to another.
Subcommittee Comments. Concerns and Recuests In addition to the questions noted above, Subcommittee Members and Consultants expressed other comments and concerns as follows (random order):
(1) On a number of occasions, the staff was reminded cf the comment made by John Garrick (PLG), namely: "The fundamental question of why and how the NRC should use PRA have not been answered."
The staff indicated that this aspect was outside their charter as given in the Program Plan.
(2) It was recommended that the staf f look at those utilities which manage PRA well within their technical groups and try to gain insights from them.
(3) Concerns were expressed in regard to the transformation process to give Level 3 results from Level 1 analysis. It was indicated that the staff seems to ignore the inadequacy of knowledge of Level 2 analysis and its potential importance. The staf f stated that they would check the transformation process using Level 3 PRAs, for example, NUREG-1150.
Closino Remarks I
In closing, Dr. Lewis thanked the staff and its consultants for 2
their participation in the meeting. He asked the staff to present a shorter overview discussion for the Full Committee meeting on May 13, 1993. He also asked the other members present to provide.their thoughts on what a proposed Committee report on this matter should address.
1
PRA Meeting Minutes May 11, 1993 FUTURE ACRS ACTION The Subcommittee till schedule future meetings on this matter as appropriate. The Full Committee will consider this. draft report at the May Meeting.
ACTIONS. AGREEMENTS AND COMMITMENTS There were no specific actions, agreements or commitments developed at this meeting other than the topics to be covered in the Full-Committee presentation on May 13, 1993.
DOCUMENTS The review documents for this subcommittee meeting.were as follows:
(1) Memorandum For James M. Taylor, Executive Director for Operations, From Eric S. Beckjord, Director, Office of Nuclear Regulatory Research, dated April 22, 1993.
Subject:
Draft Report of the PRA Working Group with the following attachments:
Appendix A - Survey of Staff PRA Uses Appendix B - Review of and Recommendations on Agency PRA Uses Appendix C - Guidance on PRA Terms and Methods (All Documents Draft Predecisional) l (2) Firs" Report of the NRC PRA External Review Group dated March 10, 1993 NOTE:
Additional meeting details can be obtained from a transcript of this meeting available in the NRC Public Document Room, 2120 L Street, N.W.,
Washington, D.C. 20006, (202) 634-3273, or can be purchased from Ann Riley & Associates, 1612-K Street, N.W.,
Suite 300, Washington, D.C. 20006, (202) 293-3950.
P
-w r-
ADVISORY COMMITTEE ON REACTOR SAFEGUARDS PROBD.BILISTIC RISK ASSESSMENT SUBCOMMITTEE MEETING BETHESDA, MARYLAND ROOM P-110
- AGENDA -
Tuesday. May 11. 1993 T_IME I
A. Subcommittee Chairman's Opening 8:30 am Remarks - H. Lewis B. Overview of Draft Working Group Report 8:35 am
- M. Cunningham (RES)
- * * * * * * *
- B REAK * * * * * * *
- 10:25 am C. Guidance for PRA Uses -
NRC Staff o
Introduction - M. Cunningham (RES) 10:45 am o Guidance for Issue Prioritization and 11:00 am Screening - H. VanderMolen (RES) o Guidance for Issue Analysis - VanderMolen (RES) 11: 40 am
- LUNCH *********
12:00 (Noon)
D.
PRA Methods, Improvements and 1:00 pm Recommendation - M. Cunningham (RES)
E.
PRA Knowledge, Skills and Training o Introduction - M. Cunningham (RES) 1:20 pm o Probability and Statistics - H. Martz (LANL) 1: 35 pm o Accident Sequence and Reliability 2:05 pm Analyses - G. Apostolakis (UCLA) o Expert Judgment - G. Apostolakis (UCLA) 2:20 pm
- BREAK ********
2:40 pm o Accident Progression and Risk Analysis-2:55 pm C. Ryder (RES) o Uncertainty and Sensitivity Analysis -
3:55 pm T. Brown (SNL)
F. Subcommittee Discussion and Planning 4:15 pm
- ADJOURN ********
4:45 PM
I 1
Results and Recommendations of the PRA Working Group I
Presentation to ACRS Subcommittee i
by Mark Cunningham Probabilistic Risk Analysis Branch Office of Nuclear Regulatory Research (301) 492-3965 4
May 11,1993 1
Elements of Presentation 1.
Overview of draft Working Group report o
introduction o
Working Group survey results o
Overview of draft report o
Summary of external reviewers comments o
Future plans 11.
Guidance for PRA uses o
introduction o
Guidance for issue prioritization and screening o
Guidance for issue analysi's PRA methods improvements and recommendations 111. PRA knowledge, skills, and training o
introduction o
Guidance on PRA terms and methods o
Job and task analysis study of issue screening and analysis 2
~
Overview (cont.)
o
.i Characteristics of Present Staff Uses Categories of Staff. PRA Uses Licensing - reactors o
Reviews
- Advanced reactors
- Plant-specific licensing actions Regulation - reactors o
Monitoring Operations
- Inspection o
issue Screening
- Operational events
- Generic safety issues o
issue Analyses
- Operational events
- Operational data analyses
- Operational trending
- Generic safety issues i
- Severe accident issues o
Facility Analyses
- Staff studies
- Individual plant examinations o
Regulatory action
- Regulatory analyses i
Licensing - Fuel Cycle and Materials 0
Reviews
- High level waste facilities Regulation - Fuel Cycle and Materials o
Facility analyses
- Staff studies of medical devices 6
y..
Overview (cont.)
Characteristics of Present Staff Uses (cont.)
1 Staff Skills and-Experience o
Most of the staff surveyed have taken one or.
more of the NRC training courses; however, Staff-experience and fami.liarity with PRA techniques needs improvement.
1 Staff PRA-related technical skills such as statistical analysis and decision analysis 1
need development.
o More formal guidance is needed for H
performing and using PRA in s number of areas.
Technical Characteristics o
The majority of PRA applications by the staff rely on adaptation of results from existing PRAs.
o No preference over any specific PRA or PRA-method was reported. Past PRAs such as WASH-1400, NUREG-1150, and industry-
-l sponsored PRAs are equally used.
7
o The majority.of PRA applications and studies are level 1 PRAs (i.e., with the product being a core damage frequency or change in core damage frequency).
o No staff member surveyed identified a use of formal decision analysis methods in their work.
8
. ~
Overview (cont.).
General Recommendations i
o Working Group continue to function to-monitor implementation of recommendations o
Agency adopt risk. assessment and-risk-management definitions used.in other parts of the federal government o
Develop a single staff document summarizing staff risk assessment and risk management activities o
Pursue decision analysis 4
i b
9
Overview (cont.)
Use Guidance: Results and Recommendations Recommendation Responsibie Timing Office Develop interim detailed guidance (including decision AEOD,NRR, 1993 criteria) for issue screenings RES and analyses (beyond that in Appendix B).
Review standard review plan updated to reflect advanced PRA WG 1995 reactor PRA review process.
Complete development of guidance for PRA uses NRR 1994 (including IPEs) in plant-specific reactor licensing issues.
Develop guidance on how to use IPEs in risk-based NRR 1994 inspection process.
Maintain close coordination between high level waste NMSS Ongoing performance assessment process and reactor risk assessment process.
Maintain close coordination between medical facility NMSS Ongoing PRA and reactor risk assessment process.
11
1 Overview (cont.)
PRA Skills, Training, and Methods:
Results and Recommendations i
o Development of a. desk reference on PRA
~
terms and methods Summary of commonly-used PRA terms 1
and methods (Appendix C)
- Probability & Statistics
- Accident Sequence & Reliability Analysis
- Accident Progression and Risk Analysis
- Expert Judgment
- Uncertainty & Sensitivity Analysis Workshops for staff o
initiated job and task analysis of issue screening and resolution processes 1
Initial. analysis of. generic issue l
prioritization and resolution process Follow-on analysis of other staff screening and resolution processes i
12 L..
Overview (cont.)
Recommendations of the Working Group Recommendation Responsible Timing Office Complete job and task PRA WG analysis of issue screening (AEOD, NRR, RES) 1993 and analysis process using SAT methods Organize workshops on Working Group interim PRA WG 1993 guidance on PRA terms (RES) and methods.
Revise PRA training based on completion of job and OP 1993 task analysis.
Develop a comprehensive PRA training program, OP 1994 based on job and task analyses of major PRA uses.
Develop PRA proficiency OP 1994 certification process.
Recruit staff with critical OP and Ongoing PRA skills program offices 13
Overview (cont.)
Future Plans o
Finalize Working Group report; transmit to Commission (if no additional meetings with ACRS and external reviewers) - June 1993 o
Complete Working Group-initiated activities Test and update generic issue guidance Workshops on terms and methods (October 1993)
Job and task analyses for generic issue prioritization and resolution (July 1993)
Guidance for adapting PRAs (December 1993)
Level 1 to level 3 transformations (December 1993) o implement Working Group recommendations Single summary document Additional guidance development Training and recruitment Methods o
Periodic review by ACRS and Commission 19
Results and Recoimmendations of the PRA Working Group
- l lI.
Guidance for.PRA Uses, PRA Methods improvements Presentation to ACRS Subcommittee by Harold VanderMolen Probabilistic Risk Analysis Branch Office of Nuclear Regulatory Research.
(301) 492-3968' Mark Cunningham-Probabilistic Risk Analysis Branch-Office of Nuclear Regulatory Research-(301) 492-3965 May 1.1,1993 20
1 Introduction - Use Guidance (cont.)
Screening issue Other Comment PRA Use Analysis
_ Analysis or Recommendation Licensing - Reactors o Reviews Advanced reactors Review SRP revisions when available.
Plant-specific licensing actione NRR develop (e.g. technical specification guidance for changes)
PRA/IPE use in various licensing actions.
Regulation - Reactors o Monitoring Operations inspections NRR develop guidance for using IPEs.
o issue Screening Operational events X
Generic safety issues X
l o issue Analyses Operational events X
Operational data analyses X
Operational data trending X
Generic safety issues X
Severe accident research X
issues 1
o Facility analyses Staff studies X
X PRA needed (screening vs.
detailed) dependent on analysis use.
23
1 s
s Screening issue Other Comment PRA Use Analysis Analysis or Recommendation Licensing - Reactors Individual Plant Examinations Define how best to use IPE results in regulation.
o Regulatory action Regulatory analyses X
Licensing - Fuel Cycle and Materials o Reviews High level waste repositories Continue coordination with similar reactor studies.
Regulation - Fuel Cycle and Materials o Facuity Analyses Staff studies of medical Continue devices coordination with similar reactor studies.
i 24
PRA Methods (cont.)
Recommendations of the Working Group Recommendation Responsible Timing Office Complete initial guidance RES 1993 for adapting PRA methods and results.
Complete Level 1 to RES.
1993 Level 3 transformation capability.
Continue development of RES Ongoing PC-based PRA tools and plant data base.
Assess feasibility of RES 1994 agency-wide reactor classification system.
dN Complete feasibility of RES 1994
" roll-up" reactor PRA models.
Develop "living" PRA RES, AEOD 1994 models and data base for staff use.
Develop RES 1994 sensitivity / uncertainty analysis tools for staff use.
.s-40
Results and Recommendations of:
the PRA Working Group 111. PRA Knowledge, Skills, and Training d
Presentation to ACRS Subcommittee by Harry Martz, LANL George Apostolakis, UCLA Christopher Ryder, NRC Thomas Brown, SNL Mark Cunningham, NRC May 11,1993 i
31 -
.c Probability and' Statistics (cont'd)-
> Focus of a study of probability
> Frequentist versus subjectivist probability
> Basic rules for calculating probabilities.
t
- Inclusion / exclusion rule
- Chain rule
> Concept of point and interval estimates
> Proper use and interpretation of best estimate 1
5 4
49-
.:z Probability and Statistics (cont'd) 1 1
> Bayesian versus classical statistics
- Advantages / Disadvantages 1
~
> Confidence versus probability intervals i
> Common statistical methods in PRA
- Estimating the probability of failure per demand
- Estimating failure rates 1
> Sources of reliability data
- Generic
- Plant-specific i
50-
.l l
l!
lI
- l il
l l
so m
f s
a l
o t
A g
n s
s n
e o
m c
L i
e d
if r
n ia i
a o
c rt gr rr e
oa e o ai p
r/
v g
(d n
s l
eD i
n me e
e a r
hn t
i y
g s
at d n, a
t o i
fl f
e o oi os t
Ci m
l l
wi ei e e t
i cb o c cs cu n u n
r nn nq ia kt) ai a n s
o gncrtsemsi mi ail nrt n
Be ono nio n ircaf epf e R
e r
r e n,rme nor n e
po t
po n
u pr igot et p o
gisr smpsm f
neeaoxao F
EdpPcEPc a
m am
'.s i
Accident Sequence & Reliability Analysis (cont'd)
Component Reliability Analysis Methods
> Bathtub curve
> Unavailability System Reliability Analysis Methods
> Redundancy
> Unavailability
>FMEA
> Fault Trees Accident Sequence Analysis Methods
> Event Trees
> Large Event Trees vs. Large Fault Trees 55
Accident Sequence & Reliability Analysis (cont'd)
Dependent Failure Analysis Types of Dependence
> Failure Rate Coupling
> Common Cause Failure Models Human Reliability Analysis Methods
> Pre-and Post-Accident Models
> Time-oriented Models
> Rating-oriented Models External Event Analysis Methods
> Seismic
> Fires
> Floods 56
.t 1
~
Accident Sequence & Reliability Analysis (cont'd)
Software Reliability Methods
> New issues
> Methods Time-Dependent Modeling
> Accident Scenarios
> Aging Available Reliability Codes
> Fault Tree Analysis CAFTA, FRANTIC, RELTREE, SETS
- Uncertainty Analysis: TEMAC, LHS
- Workstation Packages IRRAS; NUPRA; RISKMAN 57
Accident Progression & Risk Analysis CDF to Risk Transformation Class 1: Simple transformation
> Advantages
- Quick
- Useful for screening & ranking
> Disadvantages
- Plant features imbedded in results
- Details of accident progressions are lost Class 2: Surrogate PRA
> Advantages More detail than Class 1 Less resource-intensive than Class 3
> Disadvantages
- Not totally modified
- Still requires significant resources l
Class 3: Plant-specific PRA
> Advantages Plant specific results
- Access to intermediate results
> Disadvantages
- May be difficult to generalize
- Resource-intensive 60
...+
re t-Accident Progression & Risk Analysis (cont'd) 4 Consequences.
Methods Defined in the Report
> Offsite consequence calculations
> Onsite consequence calculations
- Not well developed
- Several models recently tried in NRC shutdown risk studies Guidance
> MACCS is NRC code for performing offsite consequence
> Assessment of onsite consequences not usually considered in commercial reactor PRAs. but sometimes it is appropriate 64
j Accident Progression-& Risk Analysis (cont'd) i 1
?
Risk Integration-Guidance 1
4
> Term " risk" should be restricted to the total risk or risk triplet definitions t
> If other definitions are usedin discussions outside of NRC, definitions should be clearly defined
> Use of totalrisk estimates provides simple message, but can be misleading
- Mask differences in accident frequencies and consequences 65 i
.a 44 J
Expert Judgment (cont'd)
J i
Formal Expert Elicitation Processes ~
l
> Criteria for selection Organizing experts into teams
> Technical issue selection and definition (clairvoyance test, observables, foundations)
Elicitation of probability distributions (interval, j
pair, comparisons, decompostions) 0.7 -
Oase 1: T < 500* F 0.6 --
0.5 --
I 0.4 Weighting Factor 0.3 x
0.2 -
I l
0.1 -
0 x
x 84 psi 117 psi
' 138 psi 160Ni 225 psi Containment Failure Pressure (psig) l 69
w.
- O' a
Uncertainty & Sensitivity Analysis t
l 1
i Uncertainty and sensitivity studies should j
accompany most-PRA studies that are to be used.
in issue analysis and resolution.
Uncertainty j
Quantification of the imprecision in the PRA estimates that results from imprecisely formulated a
PRA models and imprecisely known input i
variables.
Sensitivity Analysis Determination of the impact of changes in the input variables and model structures on the PRA i
estimates and the overall uncertainty.
.I J
74
Uncertainty & Sensitivity-Analysis-(cont'd) 4 Model input PRA Models Parameters input Parameters Initial conditions
. Failure rates Model coefficients
- Human error rates Physical quantitles
. Timing information Etc.
. Physicalquantities
. Etc.
Data and 4
information sources PRA Output
[
V V
Models Describing PRA Input Paramters g
Thermal-hydraulic Human reliability PRA Model Component failre rates Radionuclide transport
. - Event trees -
Fault trees Structural response
. Source term models
. Etc.
i
- Consequence models 4.
3 5
76 q
N
.r Uncertainty & Sensitivity Analysis (cont'd)
Additional Issues Selection of variables and treatment of correlation-and dependencies are important factors in an-uncertainty analysis
> Key variables must be selected to capture-most uncertainty
> There is no rigorous analytical way to select the variables
> Current practice is to use past PRAs and sensitivity analysis.
> Effect of correlations and dependencies is to increase the magnitude of the uncertainties Uncertainty results cannot generally be extended to other plants
> Can be used as a reasonableness check-to see if all important uncertainties have been included n
. l
.., a t
1 Uncertainty & Sensitivity Analysis (cont'd)'.
Additional Issues z
Selection of variables and treatment of correlation and dependencies are important factors-in an j
uncertainty analysis
> Key variables must be selected to capture most uncertainty j
> There is no rigorous analytical way to select j
the variables i
> Current practice is to use past. PRAs and sensitivity analysis.
> Effect of correlations and dependencies is to increase the magnitude of the uncertainties Uncertainty results cannot generally be extended to other plants
> Can be used as a reasonableness check to see if all important uncertainties have been included 79
. -