A Digital Twin for Detecting Liquid-Liquid Interface in Containers

Abstract

This study introduces a digital twin (DT) architecture for detecting liquid-liquid interfaces in containers, aimed at improving real-time monitoring and optimising the separation process in the petroleum industry. The research employs the DT concurrent to the physical twin to enhance the system performance by integrating response error measures with knowledge of industrial processes. The DT uses ultrasonic sensors to collect data and a high-fidelity numerical wave propagation model for an adaptive fitting algorithm to update the interface responses. Both experimental studies and numerical simulations confirm the ability of the DT to accurately determine liquid levels by analysing interfacial responses from ultrasonic signal features in transmission mode. Cross-validation and error metrics validate the DT’s adaptability and accuracy in interface detection, even with undersampled data.