Constraints in designing simulator scenarios and identifying human failure events for testing HRA methods

A diversity of different Human Reliability Analysis (HRA) methods are applied in Probabilistic Risk Assessments (PRAs). The Halden Reactor Project has, together with its member organizations, initiated a project to address a desire to test these methods against empirical evidence. In the initial phase of the project, specific accident scenarios have been chosen and human failure events (HFEs) have been identified. As part of these first trials, we are learning about some of the constraints, issues, and characteristics of the scenario designs that have to be dealt with. For instance, scenarios should be PRA/HRA relevant, sufficiently challenging to be useful tests of the HRA methods and yet be plausible, and also be feasible in a simulator setting. We are also learning about issues on how to define the HFEs of interest to ensure usefulness of the results to PRA/HRA. For instance, HFEs must be measurable and defined so that success of the action and failure of the action are clearly distinguishable. Process-based success criteria can for example reflect on several HFEs, and to define success criteria for individual HFEs other measures might be needed. This paper will discuss lessons learned about scenario design and the defining of HFEs to be able to progress toward testing HRA methods using simulations of accident scenarios.