ML19210C781
| ML19210C781 | |
| Person / Time | |
|---|---|
| Issue date: | 11/14/1979 |
| From: | Cullingford M, Oh C NRC OFFICE OF NUCLEAR REGULATORY RESEARCH (RES) |
| To: | NRC OFFICE OF NUCLEAR REGULATORY RESEARCH (RES) |
| References | |
| RTR-WASH-1400 NUDOCS 7911200097 | |
| Download: ML19210C781 (2) | |
Text
'w
- TE W P4 AN CVERVIEW 0F THE PROBABILISTIC ANALYSIS STAFF PROGRAllS ON HUf1AN FACTORS C. B. Oh, M. C. Cullingford Probabilistic Analysis Staff Office of Nuclear Regulatory Research U.S. Nuclear Regulatory Commission liashington, DC 20555
Background
The Probabilistic Analysis Staff (PAS) is sponsoring several projects to study human error in nuclur reactor plants.
Two of these projects focus on characterizing and quantifying human error.
The resulting characterization and quantification of human error should be useful in an assessment of the impact of human error on risk and in the fonaulation of new regulatory requirements.
This paper outlines the following:
(1) problem statement, (2) program objectives of two on-going projects, and (3) plans for additional research.
Problem Statement In October 1975, the Reactor Safety Study (WASH-1400) was released.
Among its inny conclusions, the study identified human error as a major contributor to some serious hypothetical accident sequences.
Failure to restore valves or switches to their r9per ".tandby alignment following tests or maintenance was found to dominate system unaval lability for a few systems in 14 ASH-1400.
Other potentially important human errors before an initiating event include flawed de-ign, personnel training, procedures, and maintenance.
Human error may also play an important role in that class of accidents in which the initiating event itself degrades the operability of the safety systems thus challenged.
For example, failure to follow procedures in safety system surveillance testing might disable the system and precipitate a trip.
Third, human reliability is extremely important during the course of an accident.
Particularly in those accidents which take a long time to develop, operator intervention can be expected.
The analyst has a difficult task to identify and quantify the likelikhood of the many possible courses of action the operators may take.
Approach - The Qaantification of Human Error In an attempt to better understand human error, the Probabilistic Analysis Staff (PAS) is sponsoring two projects to model and quantify human error.
One project involves the evaluation of nuclear reactor operator error from an analysis of past Light llater Reactor experience.
This project is being conducted 1366 127 g 911200 @ g
s N
2 by Iowa State University (ISU) fluclear Safety Research Group.
The other project is the development of a human factors handbook through the Statistics and Reliability Group of Sandia Laboratories.
The objectives of the ISU project are to (1) model quantitatively human perfonnance under a range of reactor operating conditions, and (2) evaluate operator error based on commercial nuclear plant experience reflected in Licensee Event Report (LER) data.
This project involves the analysis of all LER records from 1972 to present, extracting and evaluating those records which contain indications of possible human error, and coding the infonaation based on an ISU-developed classification scheme for statistical and modeling analysis.
In addition to other results, the project has provided frequency counts and cross-tabulations of human errors with respect to many variables including, but not limited to, component type (e.g., valves), reactor type (PWR, BUR), system affected, and type of human error such as "left valve closed."
Using the above results, we may be able to derive the probability of human error under varying operating procedures and instrument design and configuration conditions.
Since psychological factors as well as physical factors contribute to human error, the PAS contract with San < iia Laboratories is aimed at providing a qualitative methodology to identify psychological and procedural conditions which may be used to condition the estimate of human error probability.
The objective of this project is to prepare a handbook for use in evaluating human errors in operating reactors.
This handbook includes the identification of environmental conditions and operating procedures which have a high likelihood of contributing to or reducing the probability of human error.
The handbook also attempts to provide a qualitative method to estimate error rates relating to given environmental conditions, operating procedures, and psychological conditions (e.g., fatigue).
Future Intentions _
PAS human factors research may be expanded to complement current research projects in order to neet the needs of the T4RC mission.
Additional research may be aimed at developing a nuclear industry human error data base and to,
eventually predict the occurrence of human errors in operating reactors, particularly where human error dominates and significantly contributes to the risk.
Future research may include the analysis of operator behavior in experimental (simulation) settings, evaluation of other available nuclear industry data which may include a record of the occurrence and description of human error.
Additional research is being planned to support future regulatory initiatives aimed at improving-instrumentation. The specific objectives include better system status indications and improved in-plant accident response.
Conclusion The importance of human error in the man-machine interface of operating reactors is widely recognized.
PAS is attempting to broaden our knowledge of human reliability, to develop a data base, and to develop prediction models.
1366 128'
.