In multicomponent batch distillation, several main cuts and slop cuts are collected into tanks. A general model (based on a superstructure) is proposed which allows the calculation of the proportion of each slop cut to be added to each initial load and to each new batch in order to optimise the global operation. The formulation of this problem leads to a non-linear programming (NLP) problem, where the set of decision variables are the proportions of each slop cut to be re-added to each batch and the reflux ratios. Two types of criteria and of reflux ratio policies, as well as the number of batches and the number of trays, are explored. Results show the importance of the criterion choice, the little difference between an optimal constant reflux policy for a batch and a variable reflux policy, and the increase of the production rate (for problems without liquid holdup) with the number of trays.