ML19209C896
| ML19209C896 | |
| Person / Time | |
|---|---|
| Site: | Crane  |
| Issue date: | 07/26/1979 |
| From: | Farmer W NRC OFFICE OF NUCLEAR REGULATORY RESEARCH (RES) |
| To: | Cintula T, Huan Li, Milhoan J NRC OFFICE OF MANAGEMENT AND PROGRAM ANALYSIS (MPA), Office of Nuclear Reactor Regulation, NRC OFFICE OF STANDARDS DEVELOPMENT |
| Shared Package | |
| ML19209C834 | List: |
| References | |
| TASK-TF, TASK-TMR NUDOCS 7910180449 | |
| Download: ML19209C896 (5) | |
Text
.
i f
[
dljl 2 C i
?
L Those on Attached List i
Gentlemen:
6 l
Subject:
Human Factors Evaluation of TMI Accident (Proposal by A. Swain.
[
Sandia Laboratories)
L I
Enclosed is a copy of a proposal I recently received from Dr. Alan Swain
{_
of Sandia Laboratories for a human factors evaluation of the TMI accident.
[
Please review the proposal and provide me with your co:Taents.
If suffi-cient interest in tnis study exists within NRC, I would propose holding i
a Human Engineering Research Review Group heeting in August to discuss y
the proposal and now it might be implemented.
Dr. Swain could be requested
(
to attend if needed for the discussion.
I am requesting further infonna-t tion fror.; Dr. Swain on the probable funding required if the proposal were implemented in FY 198].
[
Sincerely, L
Original Signed by Willia = S.Far=er j
William S. Farmer, Chairman Human Engineering Research Review Group j
Division of Reactor Safety Research l
Enclosure:
as stated l
Distribution I
Subj l
Circ r
Chron Branch rf Farmer rf Bennett
' F armer i
Johnson Tong 70101804 %
,/
0in R.SR;y$il.(,,.vR....;B.,
...... RS R.:KfG n
L M.
orric,,
. WS FaGRdJr... Q[S.. e.tt.,
. h son /.Iong.
1/.24/.79...
.71?L n.9..
.7ng/.7.9..
mc mau m o.m neu aua 865
'43
f f
4 f
L p?o nuores;_es for Letter dated L_
Revie.. Crour Merrers:
T. Cintula, OMPA J. Milnoar., SD H. Li, DSS D. Tonai, PSYS M. Chiramal, PSYB s
J. Stone, IE W. Vesely, PAS Others:
i:
T. Murley, RSR F. Rowsome, PAS L. Beltracchi, DSS 1
l P. Collins, ULB l
- 8. Holman, CLB k
J. Pittran, PAS S. Richardson, SD G. Chipman, DSE S. Newberry, DSS T. Scarbrough, SD 1
P. Bemis, IE L. Reidinger, IE R. DiSalvo, PAS
[
l F. Gavigan, DOE /HQ D. Becknam, QAB A. Swain, Sandia l
l l
.h
.I I
r e
865
'44~
t ac= >
eu.= a = = >
an*
mc -
m,,.,
- om t
i
June 19 0 Revised Proposal Idea for Quantitative Human Factors Evaluation of Three-Mile Island Accident Alan D.
Swain, Ph.D.
Statistics, Computing, and Human Factors Division Although all details are not yet in, it is clear that human error probably played a role as an accident initiator and also was probably instrumental in less-than-adequate accident mitigation.
Human factors special'ists have been saying for years that NPPs are no different from other complex man-machine systems, and that it is not sufficient to rely on automatic redundant systems.
The TMI incident supports this assertion.
It also seems clear that the probability of TMI type incidents could be substantially reduced by the application of human factors technology to existing plants, and that this could be done with relatively little expenditure of time and funds.
In the " Preliminary Human Factors Analysis of Zion Nuclear Power Plant" (NUREG76-6503),
I described inexpensive changes to man-machine intelfaces, to written procedures, and to provisions for in-plant operator practice in responding to unusual conditions which could have reduced the probability of the incident at TMI.
In short, the technology is available now to effect substantial increases in human reliability in NPP operations.
I propose we demonstrate these points quantitatively, in a human factors analysis of the TMI incident, which would have implication for NPPs in general.
The work would be performed in five, somewhat overlapping phases.
PHASE 1.
Perform a human factors analysis of selected tasks at TMI, with emphasis on the human errors in the TMI incident.
PHASE 2.
Perform a human reliability analysis of the error events which occurred.
PHASE 3.
Generate a set of human factors changes to TMI equipment, operating procedures, and administrative control, again with emphasis on the errors which occurred in the incident.
PHASE 4.
Redo the human reliability analysis performed in Phase 2, using the revised work situation developed in Phase 3.
PHASE 5.
Prepare a report comparing the two analyses, which includes guidelines for incorporating in existing NPPs the improved human factors practices identified in the study.
(J a
' zi. R G
Following are'some details pertinent to each of the five phases.
In Phase 1 a qualified human factors subcontractor will perform a human factors analysis at TMI similar to the one I performed at Zion.
Prior to this work, all relevant reports about the incident which are on file at NRC will be reviewed.
The analysis at the plant will include the usual task analysis and other human factors procedures, including interviewing of plant personnel.
If possible interviews will include those persons directly involved in the human errors that occurred.
When the appropriate task analyses and interviews have been completed, the human events related to the incident will be plotted in an event tree which will enable conditional events to be delineated.
Phase 2 applies THERP, the Sandia human reliability model, to the event trees developed in Phase 1.
All limbs in the event trees will be assigned estimated conditional probabilities of success and failure.
Most of the estimates will be from the Handbook of Human Reliability Analysis for Nuclear Power Plant Operations, whicn will be completed in draft form (t'cr hRC review) by about the end of FY79.
Phase 2 (as well as Phase 4) the subcontractor will work closely with H. E. Guttmann and me, as we will supply the appropriate estimates of human failure probabilities and the necessary technology for the human reliability analysis.
The outcome of Phase 2 will be a prediction, in retrospect, of the probabilities of the human errors that occurred, including the common cause failure modes.
In Phase 3, the subcontractor will develop a set of recommendations for improving the existing human factors at TMI, with emphasis on reducing the probabilities of the errors that occurred in the incident and improving the recovery factors related to the errors.
These design changes will be ones that can be implemented with minimal expenditure of funds, but which would materially improve human reli-ability at the plant.
Examples of such inexpensive changes can be found in my Zion report.
Additional outputs from Phase 3 include the necessary event trees to predict, again in retrospect, the sequence of events that could have occurred in place of those that brought about the incident.
In Phase 4, the Phase 2 human reliability analysis will be redone, using the estimated probabilities derived from assuming the improved human factors developed in Phase 3.
This analysis will show quanti-tatively what the human impact could have been had " good human factors practices" been in effect at the time the first initiating event occurred at TMI.
865-?46
- Finally, for Phase 5 the subcontractor will develop a set of guidelines for NPPs in general to reduce the probability of a TMI type incident by implenentation of the human factors practices id<ntified in the study.
It is not intended that these guidelines would represent a complete set of human factors practices for NPPs.
The development of a complete set is a logical follow-on to the proposed work described herein.
The subcontractor will incorporate the guidelines into a final report describing the results of the study.
The work described above could be initiated by Sandia Laboratories early FYSO, with a final draft for government review completed by the end of FY80.
I propose that we use the service of Human Factors Technologies as the subcontractor.
They have considerable experience in nuclear power plants, and the work that they are currently performing for us in the evaluation of test and maintenance pro-cedures (for the Inspection and Enforcement Directorate, NRC) is directly applicable to the proposed work.
865 747 9