Ebook Logistic optimization of chemical production processes: Part 2

Part 2 book "Logistic optimization of chemical production processes" includes content: Engineered mixed-integer programming in chemical batch scheduling; milp optimization models for short-term scheduling of batch processes; uncertainty conscious scheduling by two stage stochastic optimization; scheduling based on reachability analysis of timed automata; integrated short and midterm scheduling of chemical production processes – a case study; integration of scheduling with ERP systems.
Nội dung trích xuất từ tài liệu:
Ebook Logistic optimization of chemical production processes: Part 2 135 Part IV Optimization Methods 137 7 Engineered Mixed-Integer Programming in Chemical Batch Scheduling* Guido Sand 7.1 Introduction After more than two decades of academic research on mixed-integer programming in chemical batch scheduling, the relevant literature exhibits a variety of modeling frameworks, which claim to be “general” or “rather general”. A review and compar- ison of related modeling concepts can be found in the chapter “MILP Optimization Models for Short-Term Scheduling of Batch Processes”. However, the diversity of batch scheduling problems makes it impossible to include all potential problem characteristics in a unified model. Moreover, from a practical point of view this may even be undesirable as general, unspecific models typically suffer from their high computational effort. Nevertheless, the general modeling frameworks serve as an indispensable means to convey and to compare basic modeling concepts and techniques. An alternative to mixed-integer programming based on general modeling frame- works is engineered mixed-integer programming based on tailored modeling and solution techniques [1]. In this chapter, a real-world case study is used to demon- strate how to develop and to solve a specific short-term scheduling problem. It will be shown that: Ĺ The case study does not fit into the general modeling frameworks. Ĺ The scheduling problem can be decomposed into a core problem and a subprob- lem. Ĺ The specific problem characteristics are modeled most appropriately by a com- bination of concepts from various general modeling frameworks leading to a mixed-integer nonlinear programming (MINLP) model. Ĺ A mixed-integer linear programming approximation can be derived following a problem specific approach. * A list of symbols is given at the end of this chapter. 138 7 Engineered Mixed-Integer Programming in Chemical Batch Scheduling This chapter is organized as follows. First, the case study, the short-term schedul- ing of the production of ten kinds of polymer in a multiproduct plant, is presented (Section 7.2). In Section 7.3, the engineered approach is first motivated, the core problem is then worked out, and the modeling approach is finally sketched. The engineered MINLP-model with its binary and continuous variables, its nonlinear and linear constraints and its objective is developed and discussed in Section 7.4. In Section 7.5, a problem specific linearization approach is presented and applied, leading to a simplified mixed-integer linear programming (MILP) model. The so- lution of the MINLP-model and the MILP-model by various standard solvers is compared with respect to the solution quality and the computational effort (Section 7.6). In Section 7.7, some general conclusions on the application of engineered mixed-integer programming in chemical batch process scheduling are drawn. 7.2 The Case Study The real-word case study considered here is the production of expandable polystyrene (EPS). Ten types of EPS are produced according to ten different recipes on a multiproduct plant which is essentially operated in batch mode. In this sec- tion, the multiproduct plant, the production process and the scheduling problem are presented. 7.2.1 Plant The topology of the plant can be taken from Figure 7.1. It consists of a preparation stage for the production of two dispersion agents D1 and D2 and an organic phase OP, a polymerization stage and a finishing stage with two lines. The supply of the raw materials F1, F2 and F3 and the storage of the final products A1. . .A5, B1. . .B5 is assumed to be virtually unlimited. The preparation stage and the polymerization Fig. 7.1 Flowchart of the EPS-plant. 7.2 The Case Study 139 stage are operated in batch mode, whereas the finishing stage is operated in con- tinuous mode. The dispersion agents are produced in two stirred tank reactors with a capacity of two (D1) and four (D2) batches along with two (D1) and four (D2) storage tanks with a capacity of one batch each. The organic phase is produced in one out of two stirred tank reactors with a capacity of one organic phase batch each; no intermediate storage is provided for the organic phase. The polymerization stage comprises four identical stirred tank reactors along with a common safety ventilation system designed for one runaway reaction (not shown in Figure 7.1). In the polymerization stage no intermediate storage is provided. The preprocessing stage, the polymerization stage and the finishing stage are fully networked by dedicated piping such that sever ...