Data quality problems in discrete event simulation of manufacturing operations

TLDR

High‑quality input data are essential for discrete event simulation, yet the shift from design to daily manufacturing operations has revealed a lack of practitioner‑based understanding of data quality problems, despite research on automation and interoperability. The study aims to extend knowledge on data quality in DES by empirically describing simulation data quality problems, data production processes, and their interrelations in a real‑world automotive context. A multiple‑case study in the automotive industry was conducted to collect these empirical descriptions. The study produced practical guidelines to help manufacturing companies improve data quality in DES.

Abstract

High-quality input data are a necessity for successful discrete event simulation (DES) applications, and there are available methodologies for data collection in DES projects. However, in contrast to standalone projects, using DES as a daily manufacturing engineering tool requires high-quality production data to be constantly available. In fact, there has been a major shift in the application of DES in manufacturing from production system design to daily operations, accompanied by a stream of research on automation of input data management and interoperability between data sources and simulation models. Unfortunately, this research stream rests on the assumption that the collected data are already of high quality, and there is a lack of in-depth understanding of simulation data quality problems from a practitioners’ perspective. Therefore, a multiple-case study within the automotive industry was used to provide empirical descriptions of simulation data quality problems, data production processes, and relations between these processes and simulation data quality problems. These empirical descriptions are necessary to extend the present knowledge on data quality in DES in a practical real-world manufacturing context, which is a prerequisite for developing practical solutions for solving data quality problems such as limited accessibility, lack of data on minor stoppages, and data sources not being designed for simulation. Further, the empirical and theoretical knowledge gained throughout the study was used to propose a set of practical guidelines that can support manufacturing companies in improving data quality in DES.