A novel constraint programming model for large-scale scheduling problems in multiproduct multistage batch plants: Limited resources and campaign-based operation

NOVARA, FRANCO MATÍAS; NOVAS, JUAN MATÍAS; HENNING, GABRIELA

COMPUTERS AND CHEMICAL ENGINEERING

PERGAMON-ELSEVIER SCIENCE LTD

Lugar: Amsterdam; Año: 2016 vol. 94 p. 101 - 117

0098-1354

Multiproduct multistage batch plants are characterized by the manufacturing of multiple products having similar recipes. These plants generally operate on the basis of a set of orders, each having its due date. In these environments, the short-term scheduling has an important impact on the costs and benefits of the organization, as well as on the effective use of the limited resources. Therefore, it is crucial to develop efficient scheduling approaches capable of solving big size problems and finding good quality solutions in reduced computational times. In multiproduct multistage environments, each order is fulfilled by means of a single or multiple batches. Most of the academic contributions, such as Harjunkoski and Grossmann, 2002; Gupta and Karimi, 2003, Marchetti and Cerdá, 2009a, 2009b, among others, have assumed that a single batch per product is demanded. When multiple batches per order are required, it is usual to operate under a campaign mode. Thus, a given number of batches of the same product are sequentially processed in order to reduce changeover times and costs. The number and size of the batches required to satisfy a given order can be specified before the scheduling activity takes place by decoupling the batching and the scheduling decisions (sequential approach). On the contrary, the scheduling problem can include the definition of the number and size of batches (monolithic approach). However, the number of batches that comprise each product campaign is a decision that must be taken in any case.