An adsorber dynamics model of a simulated moving-bed separator (SMBS) has been tested against experimentally determined concentration waveforms for the separation of propylene and dimethyl ether, in a SMBS (Fish et al. AIChE J. 1993, 39, 1783). The axially dispersed, linear adsorption, linear driving force model was solved numerically by an efficient finite-difference method with adaptive gridding. The numerical simulations predict average raffinate (dimethyl ether) and extract (propylene) concentrations that are within 5% of those of the experiment, and the experimental concentration waveforms of propylene are reasonably well reproduced. However, the model gives a poor representation of the dimethyl ether waveforms and fails to, accurately predict impurity levels in raffinate and extract streams. These failures may be attributed to the inadequacy of a uniform adsorption site model to describe the adsorption of dimethyl ether on Chromosorb 101 and are a reminder of the potential inadequacy of simple isotherms.. Simulations were used to optimize the flow rates required to achieve the most effective separation. A study of flow rate optimization for a smaller separation factor of 1.2 shows that satisfactory separations can still be obtained. It is also shown that modifying the SMBS to a 2-1-2-1 column configuration leads to an increase of the product purity.