The operation of enzyme cascades in microfluidic devices is a current field of research that promises manifold applications in biocatalysis. For an optimization of flow biocatalysis systems it is desirable to model the reactor in silico in order to enable a better understanding and thus an economic optimization of the reaction systems. However, due to their high complexity, it is still difficult to simulate coupled enzyme reactions. We here describe a new model for a plug flow reactor consisting of a porous bed of compact uniform particles functionalized with an immobilized ketoreductase (Gre2) which is overflown by a mobile phase containing the enzymatic NADPH cofactor regeneration system based on glucose dehydrogenase (GDH). By studying different flow rates, lengths and layer thicknesses of the catalytic bed, we show that the synergy of experiment and mathematical modeling can optimize the space-time yields of the reaction system.