Dynamic and Uncertainty‑Aware Risk Ranking in Nuclear Cooling Systems Using a BN‑Driven p‑VIKOR Approach

Authors:

Abstract
Reliable risk prioritisation in the Emergency Core Cooling System (ECCS) of nuclear power plants requires decision‑support methods that can represent causal dependencies and update failure likelihoods under changing operating conditions. This work develops a dynamic approach that integrates (i) Bayesian Networks (BN) for evidence‑responsive estimation of component‑level failure probabilities, (ii) Analytic Hierarchy Process (AHP) derived weights for Severity, Likelihood, and Detectability, and (iii) a probabilistic VIKOR (p‑VIKOR) formulation that incorporates BN posterior variance into the compromise ranking. The approach supports uncertainty‑aware decision‑making by combining operational evidence with expert judgement in a structured multicriteria setting. A case study involving four representative ECCS failure modes evaluates the proposed method under two tested scenarios: nominal operating conditions and a vibration‑induced degradation condition. The Bayesian Network produces scenario‑dependent shifts in failure likelihoods that static AHP–VIKOR does not capture, while the proposed p‑VIKOR extension applies uncertainty based penalisation without altering ranking order. Within the scope of the evaluated scenarios, the results indicate that the proposed approach produces scenario‑dependent changes in risk prioritisation that reflect both the updated BN likelihoods and the associated uncertainty in these estimates.