Lean Resource Scheduling Algorithm with Maximized Resource Utilization Using Iterative Local Search

Abstract

This thesis presents a lean resource scheduling algorithm which merges traditional machine scheduling problems with Lean Manufacturing concepts to determine the resource levels, such as employee headcount or number of machines used in production, and the corresponding schedule which minimize resource idle time while keeping scheduled makespan within a neighborhood around the takt time. The algorithm begins by solving a relaxed problem to find a satisfactory makespan via iterative local search, then solving a secondary problem to minimize the idle time subject to a makespan neighborhood constraint.

Experiments were conducted on a randomly generated dataset with six different factors, and both the overall program run time and the amount of idle time reduction between the first feasible solution and final solution were measured. The algorithm executes in a relatively short time, even for moderately large problem instances, and the idle time reductions are promising at a grand average of twenty-five percent reduction.

The results of the algorithm are promising on the test sets, although the method has not been tested in a practical case study. Given the promising results, further study on the underlying model, algorithm performance, and testing in a practical application are recommended.