Optimization of batch reactions in series with uncertainty

Certain types of chemical reactions, such as the global deprotection of a polypeptide, are extremely complex. As a result, it may be very difficult or expensive to develop accurate models of these chemical reactions. Without a satisfactory kinetic model for the reaction, it is difficult to develop an optimum operating policy that will maximize the profit. Stochastic optimization is applied in this work to an example process step to obtain the optimum reaction temperature and reaction time. In the case of the "here and now" problem, the optimal conditions are a lower reaction temperature and a longer reaction time than obtained from the deterministic problem. The average reaction time for the "wait and see" problem is also longer than the deterministic case, but the average reaction temperature is very close to that of the deterministic problem. Both normally and uniformly-distributed uncertain parameters are considered.