A nuclear plant accident diagnosis method to support prediction of errors of commission
- Univ. of Maryland, Glenn Martin Hall, College Park, MD 20742 (United States)
- Paul Scherrer Inst., 5232 Villigen PSI (Switzerland)
The identification and mitigation of operator errors of commission (EOCs) continue to be a major focus of nuclear plant human reliability research. Current Human Reliability Analysis (HRA) methods for predicting EOCs generally rely on the availability of operating procedures or extensive use of expert judgment. Consequently, an analysis for EOCs cannot easily be performed for actions that may be taken outside the scope of the operating procedures. Additionally, current HRA techniques rarely capture an operator's 'creative' problem-solving behavior. However, a nuclear plant operator knowledge base developed for the use with the IDAC (Information, Decision, and Action in Crew context) cognitive model shows potential for addressing these limitations. This operator knowledge base currently includes an event-symptom diagnosis matrix for a pressurized water reactor (PWR) nuclear plant. The diagnosis matrix defines a probabilistic relationship between observed symptoms and plant events that models the operator's heuristic process for classifying a plant state. Observed symptoms are obtained from a dynamic thermal-hydraulic plant model and can be modified to account for the limitations of human perception and cognition. A fuzzy-logic inference technique is used to calculate the operator's confidence, or degree of belief, that a given plant event has occurred based on the observed symptoms. An event diagnosis can be categorized as either: (a) a generalized flow imbalance of basic thermal-hydraulic properties (e.g., a mass or energy flow imbalance in the reactor coolant system), or (b) a specific event type, such as a steam generator tube rupture or a reactor trip. When an operator is presented with incomplete or contradictory information, this diagnosis approach provides a means to identify situations where an operator might be misled to perform unsafe actions based on an incorrect diagnosis. This knowledge base model could also support identification of potential EOCs when detailed procedures are not available, such as during the design phase for the next generation of nuclear power plants. To illustrate the feasibility of this modeling approach, the diagnostic matrix has been used to analyze examples of operator misdiagnosis and subsequent EOCs obtained from industry operating experience. (authors)
- Research Organization:
- American Nuclear Society, 555 North Kensington Avenue, La Grange Park, IL 60526 (United States)
- OSTI ID:
- 22030006
- Resource Relation:
- Conference: NPIC and HMIT 2006: 5. International Topical Meeting on Nuclear Plant Instrumentation Controls, and Human Machine Interface Technology, Albuquerque, NM (United States), 12-16 Nov 2006; Other Information: Country of input: France; 21 refs.; Related Information: In: Proceedings of the 5. International Topical Meeting on Nuclear Plant Instrumentation Controls, and Human Machine Interface Technology| 1430 p.
- Country of Publication:
- United States
- Language:
- English
Similar Records
Simulating operator decision processes at Savannah River
Simulation and Non-Simulation Based Human Reliability Analysis Approaches
Related Subjects
DIAGNOSIS
ERRORS
FUZZY LOGIC
HUMAN FACTORS
INFORMATION
KNOWLEDGE BASE
NUCLEAR POWER PLANTS
PROBABILISTIC ESTIMATION
PWR TYPE REACTORS
REACTOR ACCIDENTS
REACTOR COOLING SYSTEMS
REACTOR OPERATORS
RELIABILITY
RUPTURES
STEAM GENERATORS
THERMAL HYDRAULICS