Using microworlds to support dynamic human reliability analysis

Human error data is invaluable for validating low probability events, in the absence of operation plant data, in existing traditional static HRA approaches, but it is also crucial for advancing computer-based dynamic human reliability research. Probabilistic risk analysis, with the aid of advanced simulation tools, such as RELAP5-3D, has outpaced the simulation capabilities of existing HRA methods. Human error data can be used for the development of a virtual human operator model, which is fundamental for bridging the gap between existing probabilistic risk and the newly emerging field of computation-based human reliability analysis (CoBHRA). The Rancor microworld is a simplified process control which is sufficiently simple to allow participants to successfully configure the plant and begin producing electrical power after as little as half an hour of training. Rancor is well suited to gather human error data, which can then be used to build a virtual operator. Currently, Rancor is being used to gather data on human error probabilities within the context of performance shaping factors. An error seeding method is proposed, in which participants follow manipulated procedures to induce errors and require the participants to reconfigure the plant within the normal operating envelope.