Simulation of hydrogen tank refueling

Abstract

Several simulations of hydrogen filling process for a type IV tank were completed using openFOAM Computational Fluid Dynamics (CFD) open-source code. CFD codes have been proven to be a useful method for estimating temperature and pressure distribution developed inside the tank during refueling operations. All simulations were completed in a 2-D mesh and for a simulated time of 20 seconds with varying initial tank pressures and mass flow rate at the inlet. The transient, compressible and heat transfer rhoCentralFoam solver was used along with the k-ω SST turbulence model to reproduce the effects of compressibility of the gas during the filling. The simulation cases were designated into two groups for comparing a 10 to 50 g/s mass flow rate, along with different initial tank pressures of 1, 5, 10, 20, 30 and 35 MPa. The gas inlet temperature was 273.15 K, with an initial tank and walls temperature of 298.15 K. Surrounding temperature was selected as 298.15 K with a constant convective transfer coefficient of 10 W/m2 K. The simulated wall thickness is 5 mm with a thermal conductivity of 0.4 W/m K for simulating a polymer liner. Results for the temperature and pressure distribution within the tank are in agreement with previous research papers’ findings for which larger initial pressures lead to minor increments in temperature during refueling. Larger initial pressures resulted in less gas velocities at the inlet. Temperature rise is heavily dependent of the inlet gas mass flow, as a higher flow will result in more mass dispensed into the tank. The applicability of these findings suggests that at the start of the gas refueling process, a low mass flow rate can be dispensed, as the instant change in temperature is smaller, and once a higher pressure is developed within the tank, increase the mass flow to achieve a faster fill. The convective heat transfer coefficient is influenced by the mass flow rate. For lower initial pressure simulations, higher values are achieved at the end region of the tank, far away from the inlet, due to gas compression. In some instances, negative coefficient values were developed, as tank wall temperature went lower than initial temperature condition, leading to a negative heat flux into the tank. Higher initial pressure cases resulted in oscillations of the variables through the simulation. Finer mesh analysis is suggested to verify mesh dependency.