Uncertainty and sensitivity analysis of turbulence models for flow in a bare rod bundle based on the Non-Intrusive Polynomial Chaos method

This research investigates the uncertainty of turbulence model parameters within bare rod bundles and its impact on numerical prediction outcomes. Turbulent phenomena in the fuel rod bundle affect heat transfer and flow. The Reynolds-Averaged Navier-Stokes (RANS) method is widely used, but uncertainty in turbulence model parameters challenges simulation reliability. This study employs the Non-Intrusive Polynomial Chaos (NIPC) method to quantitatively evaluate the parameter uncertainty of three turbulence models: RNG $k-\epsilon$, $k-w$ SST, and LPS-RSM. Sobol index analysis is used to identify the key parameters. The results show that all three parameters of the RNG k-ε model are key, while the key parameters for k-ω SST and LPS-RSM include various coefficients related to turbulent kinetic energy, dissipation, and viscosity. The study finds turbulence model parameter variations significantly affect eddy viscosity in the mainstream region but have a lesser effect on the near-wall region, and the influence on macroscopic flow characteristics is also limited. This research quantitatively reveals the impact of uncertainty in fluid turbulence model parameters within bare rod bundles on non-dimensional turbulent kinetic energy, identifies the key parameters for different turbulence models, and provides a theoretical basis and technical support for the improvement and optimization of turbulence models in future.