Coordinating Project-Based Learning (PBL) projects across multiple courses requires each student to work in the same team throughout all of them. When enrollment is partial and practical subgroups exist, forming balanced teams manually is complex and difficult to reproduce. This paper proposes several solutions to this problem through a Mixed-Integer Linear Programming (MILP) model that minimises the number of teams while respecting constraints on size, minimum coverage and balance. Three variants are proposed: a basic one that balances team size only (A), another that guarantees a minimum representation of each course in every team (B), and a third that requires a core of fully enrolled students (C). The model is validated using real data from 94 students on the Degree in Transport and Logistics Management (GGTL) at the Universitat Politècnica de València (UPV).