Strategies for using a computer-operated reaction system for the evaluation of catalyst activity by optimization of product yield

Abstract

A fully automatic, computer-operated reaction system (FACORS) comprising a gas-flow reaction apparatus, several gas chromatographs, a desk-top computer, etc., was applied to unattended feedback optimization experiments for catalyst activity tests. The experiments shown as examples are those in which the computer executes a catalytic reaction, changing one or more experimental variables(T, F, P j) =X based on an optimization strategy and the data obtained until an optimum reaction condition(X o p t) that maximizes the yield(Y) of an object product and the maximum yield(Y m a x) are obtained. The experiments also give us catalytic information around the optimum condition. The reactions studied as examples include the dehydration of ethanol to ether over alumina and the partial oxidation of propene to acrolein over a Bi 2O 3·MoO 3-based catalyst. The strategies proposed are a one-variable-swing method(S1) and a ridge point method(S2). They were tested and compared with traditional methods, including a modified, Nelder-Mead simplex method, etc., in simulations before use in the experiments. The results showed that method S1 is one of the most powerful and useful strategies for one-variable optimizations and method S2 for 2- to 4-dimensional ones, while the simplex method is preferable in 5 or 6 dimensions. Catalyst activity tests using such feedback optimization experiments are superior to those using conventional methods conducted at given reaction conditions, since they give us information which is more effective and reasonable not only in the selection or development of catalysts but also in the design of a process using the selected catalysts.