Medium-term planning of a multiproduct batch plant under evolving multi-period multi-uncertainty by means of a moving horizon strategy

Abstract

In this contribution, a moving horizon strategy (MHS) based on two-stage stochastic mixed-integer linear programming with recourse (2S-MILP) for medium-term planning under multi-period multi-uncertainty (MPMU) (demand, plant capacity and yields uncertainties) is proposed. A dynamic 2S-MILP formulation is introduced to consider the evolution of the uncertainties in medium-term planning. The moving horizon strategy with explicit uncertainty model is applied to the medium-term planning of a polymer production plant with batch and continuous production steps where different amounts of final products are delivered by each batch according to the chosen recipes and the products can be assigned to different demands. The 2S-MILP results are compared to the results obtained from planning using the expected values of the uncertain parameters (EEV).