Discrete event simulation using high performance computing for manufacturing processes

Abstract

This work aims to implement a tool based on the discrete event simulation, whose purpose is to organize and optimize resources for industrial manufacturing issues. This work provides a simulation-based solution to the job shop scheduling problem and analyzes the efficiency of high-performance computing systems. The problem mentioned above seeks to promote the best arrangement of tasks to improve time and optimize resources. It is considered an NP-Completeness problem due to the number of variables to be evaluated, hence its complexity and importance for industrial planning. In this study, the serial simulation was developed based on discrete events and tested with three different data sets for this type of approach, ft06, ft10, and ft20. This method is compared to the best optimization tool currently, OR-Tools, developed by Google. Finally, the simulation is implemented in serial and parallel versions to include it in a high-performance computing system. It collects information on the computational metrics to find alternatives for future implementations with more robust models and real scenarios. The results exhibited an implementation based on sustainable and scalable discrete event simulation for more complex models, and can also be implemented in high-performance computing systems.