Development of ERANOS modules for the sensitivity analysis of the effective delayed neutron fraction beyond cross sections

Sensitivity analysis is essential to understand core neutronics and optimize core design complying with safety margins, as it quantifies how input parameters affect key outputs and identifies components requiring tighter control to ensure stable reactor behavior. It also provides the basis for uncertainty analysis, which evaluates the impact of nuclear data uncertainties on integral quantities and supports regulatory decisions during reactor licensing. In this context, the present thesis contributed to the neutronic design of the ALFRED reactor - Europe’s demonstrator for Lead-cooled Fast Reactor (LFR) technology - by extending the ERANOS code system with new computational modules for the sensitivity analysis of the effective delayed neutron fraction, beta_eff, with respect to nuclear data beyond cross sections. The developed methodology, based on Generalized Perturbation Theory (GPT), enables the calculation of sensitivity coefficients S(beta_eff, alpha), where alpha includes delayed and prompt neutron yields (nu^d, nu^p) and spectra (chi^d, chi^p). All expressions were derived from first principles and implemented using validated ERANOS functionalities, ensuring consistency with the existing code framework. A progressive verification strategy was adopted, from analytically solvable cases to a simplified ALFRED-inspired Toy Model. Numerical verification focused on nu^d, benchmarking GPT-based sensitivities against results from a Direct Perturbation (DP) approach. The comparison showed excellent agreement, with discrepancies below 0.2% across all isotopes and precursor families considered. Additional verification for nu^p, chi^d, and chi^p was conducted using simplified models of increasing complexity. The tools developed in this thesis provide a robust foundation for advanced sensitivity and uncertainty analyses of beta_eff in realistic reactor models, directly supporting the safety-oriented design and licensing goals of the ALFRED project.