A model predictive control framework for advanced planning and scheduling in process industries

This paper presents a model predictive control (MPC) formulation for the planning and scheduling problem in process industries. The main idea is to use a moving horizon technique as well as a feedback control concept to continuously update production schedules and plans. In existing planning and scheduling formulations, a mixed integer (non)linear programming is very often solved over a chosen horizon which depends on the level of granularity considered. To account for new orders or plant disturbances such as equipment failures or some parameters deviations (production yields, rates or state deviation affecting the production quality), updated production schedule and plan are needed. In this paper, the process of rescheduling (replanning) is compared to a classical feedback control where the result of current schedule which is the inventory and/or desired profit.