Virtual Fit: A Digital Twin Framework for Geometric Deviation Prediction in Automotive Assembly

Abstract Ensuring product quality is paramount in modern manufacturing processes, where a multitude of variables can lead to manufacturing defects and resultant geometric variations and deviations. To address this challenge, this study proposes a comprehensive framework for in-line diagnosis of accumulated geometric deviations in the assembly process. This innovative approach leverages data analysis to construct a 3D virtual simulation model capable of predicting potential failures that might otherwise disrupt the assembly of components: the geometric deviations measured on the car body are the input of the framework, and the components added later in the assembly, such as the windshield, are considered to predict whether the assembly conformity will be met. The Design Science Research (DSR) methodology was adopted to develop the framework, which was evaluated through a real industrial case in the automotive sector. The implementation of Digital Twin (DT) technology, a highly promising tool for monitoring the manufacturing state of a product, is encompassed by our approach. Through the use of virtual simulation and the best-fit alignment technique, process control is enhanced, and the prediction of car assembly quality is enabled. This study on Digital Twin utilization, virtual simulation, and alignment method offers a promising avenue to significantly enhance production process optimization and overall quality. The proposed framework facilitates a streamlined approach, ensuring a comprehensive and efficient oversight of each step and procedure, thereby marking a significant advancement in manufacturing and quality control practices.