The impact of technological implementation decisions on job-shop scheduling simulator performance using secondary storage and parallel processing

Abstract

Cyber-Physical Production Systems require scalable simulations to make decisions in close to real-time when facing highly dynamic environments. When the environment changes, new simulations assist the production planning by running scheduling algorithms. We studied the impact of technological implementation decisions on the execution performance of a heuristic-based Flexible Job-Shop scheduling simulator. We developed a simulation application that implements a heuristic-based scheduling algorithm and explored two approaches: the use of secondary storage for storing auxiliary calculations and using parallel processing to improve execution performance. The results show that the overhead of accessing secondary storage is high when compared to a main-memory only solution. A parallel processing approach based on distributed memory was implemented and compared to the sequential version. The results show that while it is accessible to implement a parallel solution, an adequate communication pattern is necessary to improve the execution time of the parallel algorithm.