The enterprise-wide optimization, involving set-point optimization, scheduling, planning, and strategic business, is progressively moving from a steady-state towards a dynamic real-time approach. The general idea, which gained ground in recent years, is to combine the classical mixed-integer programming approach for scheduling and planning with the moving horizon methodology typical of model predictive control (MPC), by sequentially solving a problem characterized by a rolling horizon. The main objective is to achieve a predictive enterprise-wide optimization, such as the dynamic re-scheduling or re-planning of production. This manuscript investigates also the feasibility of modeling directly the generictransients into a mixed-integer superstructure by adopting the Boolean logic. First, a generalizedframework is proposed, then, an industrial case study is analyzed and discussed by introducing a start-up procedure for critical process units.