Sensitivity analysis in severe accident simulations:A historical perspective

In recent years, the application of Best Estimate Plus Uncertainty methodologies in the frame of Severe Accidents (SAs) has gained significant momentum. Both researchers and regulatory bodies in the field recognize the importance of quantifying the uncertainties associated with SA codes results as well as sensitivity analysis’ relevance in understanding the variables driving the calculated uncertainty. In this framework, the current work aims to deliver a thorough overview of the historical evolution of the application of sensitivity analysis techniques within the SA domain over the past five decades, detailing their primary focus, geographical context, main techniques and key documents. Highlighting how sensitivity analysis evolved over the years, the paper underscores its critical role within nuclear safety assessments. This review offers both a detailed historical perspective and insights into future directions for research, emphasizing the need for a balance between computational efficiency and model accuracy, and suggesting the integration of machine learning techniques to enhance future analyses.