A fast POD prediction method for hydrogen leakage at different pressures

Abstract

The complexity of hydrogen leakage at different pressures leads to the challenge of developing a prediction method for both low pressure and supercritical pressure hydrogen leakage. In this paper, a reduced-order model (ROM) based on proper orthogonal decomposition (POD) is developed to rapidly predict the hydrogen dispersion caused by leakage in small holes at different pressures. The subsonic and underexpanded hydrogen jets are simulated and verified with experimental data. Numerical results of some pressures are collected as POD snapshots and POD interpolation methods are performed to predict the velocity and hydrogen concentration distributions at desired unsampled pressures. Different sample plans with different sampling intervals and sample numbers are investigated. POD-linear, POD-spline, and POD-RBF (radial basis function) interpolations are compared. The results show that the POD-linear and POD-spline methods can accurately predict the velocity and hydrogen fraction at different pressures, and both are better than the POD-RBF method. The relative errors in predicting the hydrogen mole fraction between the POD-linear and POD-spline methods and the simulation results are 2.26% and 2.37%, respectively. Best of all, the POD interpolation method proposed in this work significantly reduces the time required for prediction, costing only 3.3 ms.