Stochastic, multi-path vulnerability assessment of a physical protection system using non-fixed critical detection points

Abstract

Physical protection system (PPS) at nuclear facilities must be assessed against diverse adversary strategies, uncertainties in detection performance, and potential insider actions. Traditional estimate of adversary sequence interruption (EASI) model assumes fixed critical detection points (CDPs) and fail to capture detection variability or multi-path vulnerabilities. This paper introduces a stochastic, multi-path framework with non-fixed CDPs (nf-CDPs) that accounts for uncertainty in detection probability, communication reliability, and response delays. A stochastic approach (100,000 simulations) is applied to adversary path generation under five adversary strategies: random, rushing, covert, deep penetration, and most vulnerable path (MVP). The framework incorporates simplified insider modeling and cost–performance analysis. Results show nf-CDPs shift dynamically with stochastic sampling, producing wider probability of interruption (P_I) distributions than fixed-point CDP assumptions. Sensitivity analysis highlights insider presence and response force variability, while regression confirms a nonlinear cost–P_I relationship. The study demonstrates nf-CDPs provide a more realistic PPS assessment and practical recommendations.