This paper deals with pressure-swing-adsorption (PSA) experiments for removing CO2 and CH4 from a mixture with an 'inert' such as hydrogen or helium. Specific features are quasi cyclic steady-state and no breakthrough of the adsorbed impurity during the adsorption-production step, implying that information has to be taken inside the column, here by a set of temperature measurements. The purpose is to propose an approach for evaluating the effect on the behaviour and performance of different working conditions (feed composition, temperature, conditions of regeneration, adsorbent) under conditions of optimal bed length. We define a new criterion to represent the effective sorption capacity required in adsorption, which we call critical bed fraction (CBF), defined as the fraction of bed length occupied by the temperature and concentration fronts at incipient breakthrough of the less adsorbed component, at cyclic steady-state. The effects of the different parameters on performance (recovery and productivity) may then be expressed and analysed in terms of CBF. Some noteworthy observations are the following: for a given amount of purge, the flow velocity in the purge step may have an optimal value; a relatively high concentration of CH4 may have positive effects on both recovery and productivity when used with a high purge velocity, because it contributes to a better desorption of CO2; the advantages of an adsorbent having a high capacity per unit mass and fast internal kinetics may be offset if the density is low, implying a low capacity per unit volume.