Implementation of Task Variance Model in Security Assessment Models

The Light Water Reactor Sustainability program developed a new method to modernize how access delay timelines are computed. This approach utilizes Bayesian methods to combine subject matter expert judgement and small performance test datasets in a consistent and defensible way. It also allows a more holistic view of delay performance that provides distributions of task times and task success probabilities to account for tasks that, if failed, would result in failure of the attack. Further development of these models/distributions should be integrated into tools that are used to assess the full performance of physical security systems. This effort will research and prototype this integration. To evaluate the impact of task time uncertainty on system effectiveness, research will be conducted into the best means to integrate and map these models into the existing task structure provided in physical security related COTS modeling and simulation tool used by nuclear power plant operators. Although COTS modeling and simulation tool are used extensively in DOE/NNSA, most tasks do not currently support an ability to randomize task times in the same manner developed. If successful, and the supporting models/methods can be verified and accepted by regulators, this method for representing uncertainty in tasks times could have a significant impact on neutralization, security system effectiveness and the Risk Informed System Analysis process at existing and future power plants. Phase I will focus on the feasibility of integrating these methods into a handful of critical and often used tasks used in actual/realistic adversary attack scenarios. This will allow for a comparison of traversal times at security layers and overall probability of neutralization using both fixed times and the those provided by the new model as a basis for comparison. This Phase I effort will focus on design of a application programming interface so that uncertainty models can inform task times for each relevant task modeled in any adversary scenario. To test this new methodology using real world scenarios, a prototype will be developed that applies uncertainty factors developed using this Bayesian method to existing scenarios developed for NRC licensees. Phase II would continue the broader API development effort to allow for more task options and explore the ability of the method to employ simulation-based player data back into the tool to conveys the state the adversary individual or team.