Methodology for the analysis of dynamic human actions

A methodology for the analysis of human actions under accident conditions has been developed, which uses information from plant simulator runs, plant procedures, and plant systems information. The objective is to enhance the completeness of the event sequence model (event trees) with respect to both favorable and unfavorable operator actions. Routine human actions that impact the plant at or below the systems level, such as test and maintenance actions, are handled in the systems analysis. Types of dynamic operator actions analyzed in this paper are actions taken during an event sequence that: supplement the automatic response of plant systems for event mitigation, change or detract from the automatic response of plant systems, or lead to recovery of failed systems. The derived results can be used directly in a probabilistic risk assessment. It is judged that the major cause of possible error is misdiagnosis, which can lead to either errors of omission or errors of commission. Operator mistakes may occur when a situation is misclassified or when inappropriate decisions and response selections are made in the operator action sequences. The operator action sequences are modeled in a natural progression of human response, including observation of plant parameters and the diagnosis of the event or the decision to take action