One of the major challenges for HRA is the unavailability of relevant data. Thus, an important research topic is data collection to support the understanding and modelling of human performance. For a variety of incident and accident situations, the best option is to study human performance in simulators, since there is no operational data available for these conditions.
Experiments in HAMMLAB can produce important basic information for HRA method development and data for informed use of existing HRA methods. An important issue is the understanding of human performance in accident conditions. In order to address this issue, it is necessary to study cognitive aspects of human performance, and investigate why and how errors occur. Here we can assess the effects of Performance Shaping Factors (PSFs) in accident scenarios, as defined in a broad sense in most HRA methods, as well as studying the interaction and dependencies between PSFs. HAMMLAB experiments can also produce important input to methods that quantify “human error” utilizing Bayesian methods.
Human Reliability Analysis (HRA) methods emphasize the importance of the context, i.e. all PSFs and plant conditions, in assessments of human performance. HRA makes predictions of human performance based mainly on an analysis of context factors. Yet, although there seems to be a quite high level of agreement about which context factors affect human performance, there is less agreement on how and to what extent. In particular, knowledge is needed on the effects of context on human performance in scenarios and events typically modelled in PRA/PSA.
An urgent need within PSA/PRA, is to validate HRA methods. In particular, there is a need to find similarities and differences in use and application areas for the various approaches and techniques. In 2006 we initiated a collaborative project with many of the member organisations in the OECD Halden Reactor Project called the International HRA Empirical Study.