CFD Simulations of Hydrogen Tank Fuelling: Sensitivity to Turbulence Model and Grid Resolution

Abstract

CFD modelling of compressed hydrogen fuelling provides information on the hydrogen and tank structure temperature dynamics required for onboard storage tank design and fuelling protocol development. This study compares five turbulence models to develop a strategy for cost-effective CFD simulations of hydrogen fuelling while maintaining a simulation accuracy acceptable for engineering analysis: RANS models k-ε and RSM; hybrid models SAS and DES; and LES model. Simulations were validated against the fuelling experiment of a Type IV 29 L tank available in the literature. For RANS with wall functions and blended models with near-wall treatment, the simulated average hydrogen temperatures deviated from the experiment by 1–3% with CFL ≈ 1–3 and dimensionless wall distance y+ ≈ 50–500 in the tank. To provide a similar simulation accuracy, the LES modelling approach with near-wall treatment requires mesh with wall distance y+ ≈ 2–10 and demonstrates the best-resolved flow field with larger velocity and temperature gradients. LES simulation on this mesh, however, implies a ca. 60 times longer CPU time compared to the RANS modelling approach and 9 times longer compared to the hybrid models due to the time step limit enforced by the CFL ≈ 1.0 criteria. In all cases, the simulated pressure histories and inlet mass flow rates have a difference within 1% while the average heat fluxes and maximum hydrogen temperature show a difference within 10%. Compared to LES, the k-ε model tends to underestimate and DES tends to overestimate the temperature gradient inside the tank. The results of RSM and SAS are close to those of LES albeit of 8–9 times faster simulations.