Operator Timing of Task Level Primitives for Use in Computation-Based Human Reliability Analysis

Abstract

Computation-based human reliability analysis (CoBHRA) provides the opportunity for dynamic modeling of human actions and their impacts on the state of the nuclear power plant. Central to this dynamic HRA approach is a representation of the human operator comprised of actions and the time course over which those actions are performed. The success or failure of tasks is time dependent, and therefore modeling different times at which the operator completes actions helps predict how timing differences affect the human error potential for a given task. To model the operators’ timing variability, Goals, Operators, Methods, and Selection rules (GOMS) task level primitives were developed based on simulator logs of operators completing multiple scenarios. The logs have sufficient detail to determine the timing information for procedure steps and to map the procedure steps into the task level primitives. The task level primitives can then be applied to other procedures that were not evaluated, since they represent generic task level actions applicable to all procedure steps. With these generic task level primitives, untested scenarios can be dynamically modeled in CoBHRA, which expands the usefulness of the approach considerably. An example is provided of a station blackout scenario, which demonstrates how the operator timing of task level primitives can enhance our understanding of human error in nuclear process control.