Operational Evaluation of Digital Twin Systems in Continuous Industrial Production

Abstract

Digital twin platforms are widely promoted for monitoring and optimizing industrial assets. These systems integrate physical sensor data, simulation models, and maintenance records into unified virtual representations. This paper reports on the deployment of a digital twin system in a continuous production environment. Data ingestion pipelines, model synchronization mechanisms, and alerting workflows were evaluated over extended operation. While predictive maintenance accuracy improved, model drift and incomplete calibration limited reliability during abnormal operating conditions. Sustained value from digital twins requires systematic model management and domain expert involvement.