This paper presents a decomposition algorithm for the integrated scheduling and redesign problem of a multi-stage batch plant dealing with multipurpose units and heterogeneous recipes. First, the procedure solves the scheduling problem considering the existing plant configuration with the main goal of minimizing the makespan. Then, a second objective of minimizing the number of units utilized without worsen the makespan achieved in the first stage is considered. The units released can be reallocated to other compatible processing stages in order to minimize the initial makespan value. In order to tackle large industrial examples, both scheduling and redesign problems are solved through a decomposition algorithm, which has a MILP model as its core. The procedure is tested on several realistic instances, demonstrating its robustness and applicability.