Vulnerability Assessment of a Physical Protection System Design Using a Multi Path Analysis and Moving Critical Detection Points

Abstract

A modified Monte-Carlo-based multi-path method was developed to assess the probability of interruption (PI) offered by a physical protection system (PPS) at a nuclear facility against an adversary attacking the facility. To evaluate the effectiveness of a PPS, the estimate adversary sequence interruption (EASI) model developed by Sandia National Laboratories was used. To perform a multi-path analysis based on an adversary sequence diagram (ASD), a stochastic approach (Monte Carlo) script was developed. Three types of distributions (Gaussian, Poisson, and Uniform) were used to determine the differences in choosing the values of probability of detection (PD) provided by the PPS elements enabling the estimation of PI as a distribution. The multi-path analysis approach presented in this study was not limited to the adversary’s single path analysis approach used in the EASI approach. The EASI model in its usual usage does not estimate uncertainty in the estimated value of PI. However, such an uncertainty estimate is useful to evaluate the performance of the PPS. Moreover, in this model, the critical detection point (CDP) was not fixed at the same PPS protection layer for all attack scenarios as in the case of EASI approach. Instead, CDP was moved to enable the analysis of the different types of actions adversaries take to achieve their goals based on their perception of the PPS. These adversary actions consist of random, rushing, covert, deep penetration, and most vulnerable path (MVP) strategies. The script developed was enabled to evaluate the value of PI for situations with a moved CDP, depending on the paths an adversary may attempt to reach the target. This type of CDP movement resulted in more realistic PI values and the corresponding uncertainties. Also modeled in the script was bypass mode and the threats from insiders by eliminating the corresponding detection or delay elements of the PPS for the chosen adversary path. The prices of each PPS element such as for sensors and cameras were integrated into the script. Taking into consideration the unit price of the detection elements of the PPS, a relationship between cost and PI was examined. As a result of sampling PD values from three different distributions, the corresponding PI values and their uncertainties were compared for each sampling strategy. This study demonstrated that the inclusion of a moving CDP in the EASI model to assess the effectiveness of the PPS had a significant impact. The relationship between the total cost of the detection elements and the mean PI values indicated that these parameters did not follow a linear relationship. The analysis of PI distribution showed that the lower tails of the distribution with minimum values instead of the average PI shall be considered while evaluating the performance of the PPS.