In attempting to optimize a chemical synthesis, competing side reactions often interfere, which degrades the reactants into other undesired substances. In those cases, the course of the reaction can be influenced by acting on the manipulated variables temperature and feedrate as the reaction proceeds. The determination of such strategies is usually referred to as optimal control. In the present paper, a methodology of optimal control of batch reactors, the main processing tool in fine chemistry, is presented. The implementation of the methodology involves two phases : (i) an optimization phase where a dynamic mechanistic model is determined and used to optimize the process based on a technical-economic objective function and (ii) a control phase where a dynamic empirical model is used to optimize a performance objective function which penalizes deviations of process variables from their optimum values given by the optimization phase. The first part of the paper presents the global strategy and the necessary theoretical developments. The second part of the paper is an illustration of the strategy using the epoxidation of furfural as a model reaction. The optimization based upon an approximate mechanistic model is compared to a more conventional approach involving the response surface methodology. Then, control studies by simulation and experimentation on a 21 pilot plant reactor for the epoxidation of furfural are performed. The previously determined optimal trajectories are implemented in practical operation.