Hybrid solution methods for operational production planning

The present project deals with production planning problems, in particular lotsizing and scheduling, which originate from planning problems in the semiconductor and chemical industry. This kind of problems is a combination of the classical lotsizing scenario and the short-term scheduling decisions. In most cases these two decision levels (lotsizing and scheduling) are strongly intertwined. So, in order to develop a realistic and efficient planning tool it is necessary to consider both aspects simultaneously. The main task of this project is to analyze production settings which include parallel, non-identical machines and sequence-dependent setup times and costs. Different problem settings (alternative objective functions, resource- constraint setup processes, ) will be analyzed. The goal is to develop strong model formulations by searching for valid inequalities, such that at least for small problems an exact solution method is applicable. Secondly, special solution methods will be developed and analyzed. Different approaches for the solution algorithms will be taken into consideration. We will start with developing an exact branch-and-bound method (in particular a branch-and- price algorithm) and combine it with heuristic strategies in order to construct an efficient exact algorithm. For larger problem instances different possibilities for decomposing the problem into smaller subproblems will be investigated. These subproblems should be small enough to allow a fast computation of a high-quality solution. In order to improve these decomposition approaches, metaheuristics will be used to guide the decomposition process. Because both exact methods and approximation algorithms will be developed during this project, it is possible to estimate the performance and the solution quality delivered by the newly created hybrid solution methods. The goal of the present research project is to develop and analyze efficient algorithms based on heuristics, metaheuristics and exact methods, in order to solve complex production planning processes more efficiently.