Relax and fix heuristics to solve one-stage one-machine lot-scheduling models for small-scale soft drink plants

Abstract

The production planning of regional small-scale soft drink plants can be modeled by mixed integer models that integrate lot sizing and scheduling decisions and consider sequence-dependent setup times and costs. These plants produce soft drinks in different flavors and sizes and they have typically only one production line. The production process is carried out basically in two main stages: liquid preparation (stage I) and bottling (stage II). However, since the production bottleneck of these plants is often in stage II, in this study we represent the problem as a one-stage one-machine lot-scheduling model that considers stage II as the bottleneck but also takes into account a capacity constraint of stage I. To solve the problem, we propose relax and fix heuristics exploring the model structure and we evaluate their computational performances solving different problem instances based on real data of a Brazilian small-scale soft drink company. The solutions obtained are compared to the company solutions and the solutions of a general-purpose optimization software.