The Multi-Level-Reactor-Design (MLRD) methodology is a novel model-based optimization approach. It enables the development of innovative and tailored chemical reactor concepts by following a three level approach .The basic idea is to describe the reaction system of interest in a fluid element of arbitrary shape. Optimality with regard to a certain objective is reached by tracking this fluid element over time in state space and by manipulating its inner state accordingly. This is achieved via variation of outer fluxes (across the fluid element boundary) over time. The fluxes are either optimized directly (level one) or indirectly by optimization of suitable control variables (level two). Subsequently, an actual reactor design concept is deduced from the identified optimal control variable profiles. The resulting design parameters are then determined by optimizing the system within the specified setup (level three).
From this it follows that on level one, the system is limited by the applied reaction kinetics only. Thus the full potential of the reaction system can be revealed independent from known apparatuses without any further restrictions. On level two, the kinetics of mass and energy transport are additionally considered, thereby disclosing the positive or negative influence of these phenomena. Finally, level three represents the influence of the chosen apparatus configuration including losses due to technical feasibility.
This approach enables an early-stage detection of optimization potential and was already applied to various systems including heterogeneously catalyzed reactions, namely the oxidation of sulphur dioxide  and the partial oxidation of ethylene . In these examples, the fluid element was described as pseudo-homogeneous system. This implicitly assumes the utilized reaction kinetics to be intrinsic, and thus neglects effects of intraparticle mass and energy transport. This, however, is a severe simplification which often is neither applicable nor justified for many industrial processes. Therefore, in this work a major extension of the optimization methodology described above is presented by incorporating the effects of intraparticle transport processes. Accordingly, the fluid element is now described as pseudo-heterogeneous system resulting in two distinct advances: Firstly, the included transport effects now affect the reactor design directly returning more realistic results. Secondly, the catalyst pellet is treated and modelled as separate subsystem, enabling its properties to be tuned and optimized as well. In our model, the catalyst pellet is regarded as non-isothermal and the diffusive fluxes inside are described using a sophisticated Stefan-Maxwell based approach. Besides common numerical methods, the pellet governing equations were also solved using the Adomian Decomposition Method (ADM). This method results in an algebraic series approximation of the differential equations reducing the computational effort during optimization.
First results originating from the application of the extended MLRD method demonstrate that the simultaneous consideration of reactor and catalyst pellet in the optimization clearly changes the resulting reactor design with regard to optimality. To conclude, this approach allows for the concurrent optimal design of reactor and catalyst and therefore represents a valuable tool to unlock the full potential of heterogeneously catalyzed reaction systems.
 Peschel, A., Freund, H., Sundmacher, K., Ind. Eng. Chem. Res. 49 (2010), 10535-10548
 Peschel, A., Karst, F., Freund, H., Sundmacher, K., Chem. Eng. J. 66 (2011), 6453-6469