Adaptive nonstationary operation of reactor systems for catalytic combustion

NOx removal from automobile exhaust by selective catalytic reduction is considered as an example of a system with competing chemical reactions and selectivity strongly dependent on temperature. In thermally coupled monolithic reactors, temperature profiles can be successfully kept within the required range by periodic operation - switching between a hot stream containing reactants and an inert cooling stream. This work presents a methodology for determining optimum (i.e. minimum emissions) operating parameters of a system of monolithic reactors in situations where the inlet gas properties Vary in time. An adaptive control strategy is employed, based on predicting the inlet properties by a feed-forward artificial neural network, and then using a set of heuristic rules combined with nonlinear optimization to select the mast favorable reactor configuration. This adaptive control scheme leads to significant improvements (in terms of integral emissions of pollutants), compared with a situation where system configuration and operating parameters are fixed a priori.