• CFD
• SA
• UQ

The present research aims to contribute to the validation of the use of Computational Fluid Dynamics (CFD) analysis for the safety assessment of nuclear applications. The research focuses on investigating Thermo-Fluid-Dynamics phenomena occurring inside complex flow geometries and characterized by heterogeneous distributions of power and/or coolant fields. Additionally, the study involves conducting forward uncertainty propagation of the CFD model input parameters to provide a best estimate calculation and the uncertainty bounds for the considered CFD model.
The first validation case involves studying the in-vessel flow mixing within the reactor pressure vessel of a pressurized water reactor. The significance of such phenomena is evident in their ability to perturb the spatial and temporal distribution of coolant properties such as temperature, velocity, and boron concentration at the core inlet under transient conditions that deviate from the nominal operating conditions which may lead to reactivity insertion accidents. In addition, in-vessel mixing phenomena can also affect the thermal interaction between the coolant and the pressure vessel, also known for pressurized thermal shock scenarios, which is crucial for maintaining the integrity of the reactor pressure vessel.
The second validation case involves studying the heat transfer mechanism in a wire-wrapped fuel bundle cooled by heavy liquid metal under a wide range of operating conditions. The significance of the study stems from selecting Heavy Liquid Metals (HLM) as coolants, such as lead or lead bismuth eutectic, and introduces unique thermal-hydraulic characteristics. While HLM coolants offer superior heat transfer properties compared to traditional water cooling, including higher thermal conductivity and natural circulation capability, they also present specific challenges such as potential flow stagnation and coolant freezing that must be carefully evaluated through detailed heat transfer studies to ensure safe reactor operation.
Furthermore, the research addresses the challenge of performing uncertainty quantification analysis for the CFD models considering the uncertainties of the model’s input parameters.
Consequently, Deterministic Sampling is proposed as a methodology to propagate uncertainties through CFD models, due to the extensive resources needed to perform the CFD analysis.
Moreover, the study covers the validation of the Deterministic Sampling approach against other Stochastic Sampling approaches; Wilks' methodology and Random sampling together with a Sparse representation of Polynomial Chaos Expansion, where both of which have been considered for the second validation case and applied for a low fidelity CFD model.
The research activity is supported by extensive CFD code investigation, validation and code-to-code and model-to-model comparison for results obtained within the framework of several international research projects and collaborations.