Methods for handling of dependencies in HRA: Fortum case study

This report is a part of collaboration carried out in 2015 between a Nordic PSA Group (NPSAG) project and the SAFIR2018 research programme project PRAMEA (Probabilistic risk assessment method development and applications) with the objective to evaluate existing methods and develop supplementary guidance for the assessment of dependencies in Human Reliability Analysis (HRA). As part of the collaboration, Lloyd's Register Consulting performed a literature study on existing methods and guidance for assessment of dependencies in HRA and performed case studies in Sweden, financed by NPSAG. As part of PRAMEA two HRA dependency case studies were performed: One by VTT on Fortum's assessments and one by Risk Pilot on shutdown loss of cooling accidents (LOCA) in boiling water reactors (BWR). This case study performed by VTT examines the probabilistic assessments of dependency between multiple pre-initiator actions in Fortum's HRA as part of Loviisa 1 and 2 Probabilistic Safety Assessment (PSA). The chosen case example was the calibration error of two redundant thermostats due to a common cause failure (CCF) leading to loss of certain cooling systems in emergency situations due to no actuation of auxiliary air conditioning. Fortum's dependency model has been developed based on the Technique for Human Error Rate Prediction (THERP) method, from which it has been somewhat modified to better meet Fortum's needs. It can be concluded that the model gives similar results as THERP, seems reasonable and differentiates between different situations. The clearly defined questionnaire for determining the dependency level adds transparency and reproducibility to the method. It is also in reasonable agreement with NUREG-1792 Good Practices for Implementing HRA.