Dynamic and Classic PSA Model Comparison for a Plant Internal Flooding Scenario

EasyChair Preprint no. 11233

Abstract

Different classic and dynamic PSA (Probabilistic Safety Assessment) codes have different capabilities. However, few comparisons between codes have been published. GRS compares two classic PSA codes, RiskSpectrum^® and SAPHIRE, and the dynamic PSA tool MCDET (Monte Carlo Dynamic Event Tree) by GRS. MCDET includes the Crew Module which allows simulating human interactions. The plant internal flooding scenario chosen results from a extinguishing water pipe leakage within the reactor building annulus of a pressurized water reactor. After the leakage, leak detection and human actions are needed to interrupt the water flow before items important to safety are damaged. An available and validated RiskSpectrum^® PSA plant model of the scenario was used and automatically transferred to SAPHIRE by applying the GRS tool pyRiskRobot. For the Crew Module, the scenario was extended by different steps and more time-dependent elements. The comparison shows: Both classic models lead to nearly identical flooding induced damage probabilities of the systems. However, qualitative differences between the codes exist. Preliminary results with the dynamic model show a lower probability because of the additional steps and large time available for mitigation measures. Concluding, dynamic PSA codes can enhance results from classic ones, particularly regarding aggravating conditions delaying mitigation measures outside buildings. It has been demonstrated that pyRiskRobot can transfer the most relevant parts of a classic PSA model increasing the analysists’ flexibility.

Keyphrases: aggravating conditions, code development, dynamic model, human action, internal flooding, probabilistic safety analysis

@Booklet{EasyChair:11233,
  author = {Florian Berchtold and Tanja Eraerds},
  title = {Dynamic and Classic PSA Model Comparison for a Plant Internal Flooding Scenario},
  howpublished = {EasyChair Preprint no. 11233},

  year = {EasyChair, 2023}}