A heuristic for multi-level lot-sizing problems with bottlenecks under a rolling schedule environment
Lot-sizing scheduling techniques determine what amount is required to meet forecasted demand whilst minimising the sum of setup and holding costs. These techniques are not adequate to provide an optimal solution to bottleneck facility problems which do not meet demands placed on them. Thus, it is necessary to analyse how much should be produced from each product with bottleneck facilities. Therefore a lot-sizing problem with bottleneck(s) under rolling schedule environment is the subject of this thesis. This research proposes a simple heuristic for multi-level lot-sizing problems where there is a bottleneck. Previous methods to solve this problem have formulated the problem as an integer programming problem and solved the problem using a Lagrangian relaxation embedded within the branch and bound procedure. Then the proposed heuristic is extended for multiple bottleneck problems, and finally applied to the real life problem. In this research it is suggested that items to be produced can be grouped into two types and a simple but efficient heuristic can be used to determine the production quantities required. A program was developed to compute production levels and was found to require only a small fraction of the computer time required by the full integer programming approach and to produce solutions of reasonable quality. The heuristic is simple to implement. Keywords: heuristics, inventory, lot-sizing, scheduling, production, bottleneck, linear programming, integer programming.