The Benefits of Using a Risk-Informed Approach to Resolving GSI-191

Abstract

The emergency core cooling system (ECCS) and containment spray system (CSS) in a pressurized water reactor (PWR) are designed to safely shutdown the plant following a loss of coolant accident (LOCA). The assurance of long term core cooling in PWRs following a LOCA has a long history dating back to the NRC studies of the mid 1980s associated with Unresolved Safety Issue (USI) A-43. Results of the NRC research on boiling water reactor (BWR) ECCS suction strainer blockage of the early 1990s identified new phenomena and failure modes that were not considered in the resolution of USI A-43. As a result of these concerns, Generic Safety Issue (GSI) 191 was identified in September 1996 related to debris clogging of the ECCS sump suction strainers at PWRs. Although plants have taken steps to prevent strainer clogging (by increasing the screen area, for example), satisfactory closure of this issue has proved elusive due to long term cooling issues and the effect of chemical precipitates on head loss. Previous investigators have identified bounding scenarios using conservative inputs, methods, and acceptance criteria. The acceptance criteria are applied in a “pass/fail” fashion that ignores risk. That is, if the results are acceptable, the issue has been resolved. Otherwise, it is necessary to either redo the analysis with partial relaxation of analytical conservatisms or perform additional plant modifications to ensure that the acceptance criteria are met. This article describes a new approach to close out the GSI-191 issue by evaluating the risk associated with ECCS performance on post-LOCA core cooling as a basis to change the plant license. The approach includes an assessment of LOCA frequencies as a function of break size at locations along the reactor coolant system, as well as a full quantification of the uncertainties associated with LOCA frequencies and the generation, transport, accumulation, and subsequent impact of debris on ECCS performance. The overall frameworks for the deterministic and risk-informed approaches are summarized with emphasis on the risk-informed method. The differences between the deterministic approach taken in the past and the new risk-informed approach are described. Advantages and disadvantages between the two methods are described and contrasted for the GSI-191 issue. The South Texas Project (STP) GSI-191 risk-informed closure efforts are presented.

Global importance measure methodology for integrated probabilistic risk assessment

“…103 Such debris can be transported into the containment sump and, during the recirculation cooling phase, the transported debris poses two potential hazards: (a) it plugs up emergency core cooling system (ECCS) strainers and (b) it penetrates the ECCS strainers and blocks coolant channels in the reactor core. In 2011, the South Texas Project Nuclear Operating Company initiated a pilot plant project for risk-informed resolution of GSI-191, 12,104,105 where an I-PRA framework was developed to deal with the dynamic and location-specific nature of the physical failure phenomena. 12 This I-PRA framework consisted of two modules: (a) a plant-specific PRA module and (b) the Containment Accident Stochastic Analysis Grande (CASA Grande), which simulates several key physical processes associated with GSI-191, for example, location-specific LOCA occurrence, jet formation, debris generation, debris transport, strainer head loss, chemical effects on head loss in debris beds, and thermal hydraulics in the reactor coolant system.…”

“…To generate the failure probability of basic event CB, Uncertainty Quantification (UQ) is performed on the reduced-order CASA. The uncertainties associated with the input parameters are represented by probability distributions (Table 7) based on the technical report, 107 Morton et al, 41 and engineering judgment. The uncertainty associated with the fitted filtration and shedding parameters are treated based on the range of their values listed in Tables 5 and 6, respectively.…”

“…The total flow rate of each ECCS train ( A, B , and C ) is represented by discrete probability distributions based on the ECCS design of the plant being analyzed. 107…”

“…The uncertainty associated with M crt is represented by a uniform distribution with upper and lower bounds of 50 and 4 (gram/FA), respectively, 107 and is propagated by another layer of MC simulation using 5000 samples. As a result, the point estimate of p CB is calculated as p ^ CB = 0.294.…”