Chemical batch processes are typically used to manufacture specialty and fine chemicals. In 2017, these specialty chemicals represented more than 25% of the EU chemical sales, and their global market is predicted to grow in the upcoming years. This growth is accompanied by the need for upgrading existing production plants and building new plants.
As the construction of batch plants requires major investments, capacity decisions form the main subject of the strategic batch plant design problem. Hence, the strategic design of chemical batch plants defines the number, size and connectivity of equipment (i.e. tanks and reactors), the optimal production batch sizes of the required products and the preferred production planning policy. This multiproduct batch plant design problem is a challenging optimization problem, with a wide range of opportunities to investigate. Indeed, current models optimize often only capital costs and (limited) operating costs without further business context. Our aim is to extend this problem to incorporate realistic production environments (Make-To-Order, Make-to-Stock, Finish-to-Order), frequently used design options (parallel production lines, dedicated or temporary storage tanks) and state-of-the-art business objectives such as responsiveness, reliability and flexibility (conform with the industrial SCOR performance attributes).
Such plant design models are formulated as MILP models and solved with exact solvers. However, with all the additional features and decisions, the combinatorial complexity becomes extremely high so that meta- and matheuristics are needed. Finally, to examine the performance of a plant with a fixed design for a realistic stochastic set of customer orders, greedy scheduling heuristics are developed.