The hydrogen (H2) diffusion through palladium (Pd) and Pd–copper (Cu) membranes was numerically investigated by developing a two-dimensional computational fluid dynamics model for predicting the performance of H2 separation. The momentum and mass transport phenomena in the laminar flow conditions were solved at different operating conditions in a vertical cylindrical-type reactor. The effect of feed-gap distance, H2 concentration, and reactor heating temperature on the H2 permeation processes were simulated and compared for both Pd-based membranes. The concentration, velocity, and convective and diffusion mass transfer flux distributions were analyzed using the designed model. The H2 concentration was proportional to the feed-gap distance/cross-sectional area. The smaller the feed-gap distance, the greater the probability of a H2 molecule being adsorbed by the membrane surface and the ionization energy increasing, leading to further H2 dissociation through the Pd-based membranes. It was found that the diffusion flux of all feed concentrations was substantially decreased 50 s after the start of the permeation process. Moreover, the diffusion flux of the Pd–Cu40% membrane was relatively larger than that of the pure Pd membrane under the same operating conditions. The distributions of the convective flux, diffusion mass transfer flux, and concentration of the Pd–Cu40% membrane were substantially increased up to 350 °C, then fell to a lower value at higher temperatures. The simulation results were validated with the experimental results, with analysis indicating a good agreement with the experimental results under the same operating conditions. It can be concluded that the simulation modeling for Pd-based membranes was able to predict the optimum operating conditions at high H2 diffusion rates.