BibTeX

@inproceedings{DenmanANS2013,
title = {Proof of Principle Framework for risk informed Severe Accident Management Guidelines},
author = {M R Denman and K M Groth and T A Wheeler},
year = {2013},
date = {2013-11-01},
booktitle = {Proceedings of Risk Management for Complex Socio-Technical Systems (RM4CSS)},
address = {Washington DC},
organization = {American Nuclear Society},
abstract = {PRA is being used to demonstrate the safety basis and to answer regulatory questions for Small Modular Reactor (SMR) designs. One key regulatory question associated with SMRs relates to staffing strategies for multiple co-located SMR units. In order to evaluate staffing requirements, the cognitive toll of multiple simultaneous severe accidents must be evaluated. The IDAC human cognitive model can be used to evaluate cognitive demands on operators, given a set of operating procedures. Severe Accident Management Guidelines (SAMGs) direct the operators’ actions when the accident proceeds into the residual-risk arena and thus outside of normal operating procedures. Unfortunately, SAMGs for most SMRs have not yet been develop, which limits PRA applicability. This paper provides proof of principle that coupling of Discrete Dynamic Event Trees (DDETs) with Bayesian Networks (BNs) can be used to create risk-informed SAMGs; these SAMGs enable design-stage insight into staffing and, longer term, have implications for improving operator response during severe accidents.},
keywords = {Artificial intelligence, Decision support systems, Dynamic PRA, nuclear power, Probabilistic risk assessment (PRA), risk-informed procedures},
pubstate = {published},
tppubtype = {inproceedings}
}

Abstract

PRA is being used to demonstrate the safety basis and to answer regulatory questions for Small Modular Reactor (SMR) designs. One key regulatory question associated with SMRs relates to staffing strategies for multiple co-located SMR units. In order to evaluate staffing requirements, the cognitive toll of multiple simultaneous severe accidents must be evaluated. The IDAC human cognitive model can be used to evaluate cognitive demands on operators, given a set of operating procedures. Severe Accident Management Guidelines (SAMGs) direct the operators’ actions when the accident proceeds into the residual-risk arena and thus outside of normal operating procedures. Unfortunately, SAMGs for most SMRs have not yet been develop, which limits PRA applicability. This paper provides proof of principle that coupling of Discrete Dynamic Event Trees (DDETs) with Bayesian Networks (BNs) can be used to create risk-informed SAMGs; these SAMGs enable design-stage insight into staffing and, longer term, have implications for improving operator response during severe accidents.