Finding the proper level of detail to achieve sufficient model fidelity using FlexSim: An industrial use case

Abstract

Discrete-Event Simulation (DES) has established itself as an appropriate instrument for optimizing production facilities in the manufacturing industry, enhancing the overall production system performance and therefore ensuring competitiveness concerning today's difficult market conditions. DES supports decision-making by providing model-based system analyses for given objectives and obtains meaningful and verifiable results. Because high-fidelity modelling is always a time-consuming and therefore expensive task, the necessary level of detail of a simulation model is an important question to answer. Therefore, the goal of this work is to examine the different levels of detail of a series of models created for a complex industrial production system utilizing the FlexSim simulation environment. First, an iterative modelling approach, starting with the development of a low-detail model, followed by the creation of further refined models with higher fidelity, is presented. A qualitative fidelity comparison between the models is stated using 4 fidelity dimensions. Furthermore, the necessary level of detail for a given accuracy target is evaluated by a comparison between simulated and real production system throughputs during model validation. Finally, achieved production system optimizations based on experiments using the adequate model are presented. Despite the large variety of different production environments, similar use cases may profit from the findings presented in this work.