In a large number of industrial semi-batch reactor applications, a single reaction takes place in the reactor and the operational objective is to compute the optimal feed rate of reactants and the optimal batch temperature that optimizes an objective function at the end of a fixed batch time. In this work, it is shown that the optimal operating policy can be split into several time intervals. Analytical expressions for the optimal input are developed for each interval. Parametric uncertainty is handled by implementing this strategy in closed-loop. This methodology is illustrated via simulation of two semi-batch reactor applications.