Discrete Event Simulation as a Predictor for Factory Traffic Management

Abstract

Manufacturing environments increasingly rely on automation and data-driven decision-making to optimize efficiency and production rates. This study explores the application of Discrete Event Simulation (DES) to model material flow and predict AGV (Automated Guided Vehicle), crane, and cart movements within a factory. The goal is to develop a digital twin that enables real-time decision-making, optimizes scheduling, and minimizes bottlenecks.

To achieve this, we utilize SimPy, an open-source Python-based DES library, in conjunction with a custom-built API and React.js front-end interface. The study evaluates available DES software options and justifies the selection of SimPy based on flexibility, integration capabilities, and its suitability for modeling custom business rules. The solution is structured into modular components handling path planning, transporters, flows, stations, hot-cold starts, and utilities, ensuring adaptability to future improvements.

A validation framework was established, utilizing historical data comparison and real-time validation to assess the simulation’s predictive accuracy. Over a 40-day testing period, the simulation achieved 89.6% accuracy and a sensitivity, or true positive rate (TPR), of 80.2%. The simulation provides a reliable first-pass scheduling tool that can be further refined with improved data collection.

The findings indicate that while full automation of AGV deployment is not yet feasible, this study lays the foundation for future integration with the factory’s Vehicle Management System (VMS). Business implications include the potential for automated scheduling, enhanced material flow visibility, and optimization of capacity planning. Future work should focus on improving data accuracy, integrating live factory data streams, and refining algorithms for predictive scheduling.