Hydrogen purification with optimum recovery and energy consumption from various process streams represents one of the major commercial uses of pressure swing adsorption (PSA) technology. In the plasma process for hydrogen production, the effluent stream from a reactor has to be processed to obtain the high purity hydrogen before being fed to fuel cells for power generation or stored in tanks. The major impurity in the effluent stream is unreacted methane. A rigorous PSA model for simulation and optimization purposes has been developed. The non-isothermal, bulk separation with variable superficial velocity and dispersion coefficient, linear driving force approximation for particle uptake, and Langmuir isotherm to represent adsorption equilibrium were applied in the PSA modeling. The model was solved using gPROMS software. The extensive simulation results involving parametric studies of PSA separation performance were well matched with the experimental results using an activated carbon made from coconut shell as an adsorbent. The optimal conditions for separation of 50%H-2/50%CH4 mixture in a laboratory-scale PSA unit, and separation of 25%H-2/75%CH4 mixture, representing exit gas from plasma reactor, in a pilot-scale PSA unit were obtained by carrying out optimization routines in gPROMS. (c) 2008 Elsevier B.V. All rights reserved.