This study focuses on an optimization of start-up operating conditions of a rapid pressure swing adsorption (RPSA) process, which is operated in a cyclic pressure variation mode. The objective function is defined not only to reduce the operating power but also to shorten the time to reach the cyclic steady state (CSS), as well as to increase the purity of the desired product at CSS. A general mathematical model considering the dynamic variation and spatial distribution of properties within the bed has been formulated and described by a set of integrated partial differential algebraic equations (IPDAE). The number of variables for optimization is 16 825 and both the single discretization of a spatial domain and the double discretization of spatial/time domains have been used for the numerical integration. As the computation result the optimal cycle time is 14.46 s and the optimal feed pressure is 597 kPa. Under the optimal condition the purity of desired product at CSS is calculated as 96.42% and the CSS convergence rime is 5857 s.