A Digital Twin for Condition Monitoring of an Industrial Chain Belt Conveyor System

Abstract

Chain belt conveyors have become a popular choice for transporting materials in diverse industries. The primary driving force behind these conveyors is the gearmotor, which is composed of an electric motor and a gear unit. The industrial electric motor market is estimated to be 21 billion USD and consumes up to 70\% of the electricity by the industry. However, research on chain belt conveyors systems remains limited, leading to poor maintenance strategies and inadequate understanding of their efficiency. The aim of this paper is to develop a digital twin for condition monitoring, applicable to industrial gearmotor-driven chain belt conveyor systems. To achieve this, the current state-of-the-art and industry practices were analysed, focusing on SEW Eurodrive as a representative for industrial industry. A literature review was conducted to explore predictive maintenance, digital twins and condition monitoring in the context of chain belt conveyors. Consequently, this has culminated in the development of a novel chain belt conveyor model, based on fundamental mechanical analysis of the chain belt conveyor, while also acknowledging the limitations associated with industrial settings. In order to accomplish this objective, parameter estimation was performed for system identification using data collected from physical chain belt conveyor systems. Additionally, various methods have been proposed for anomaly detection, which include the utilization of estimated parameter thresholds, frequency domain analysis, statistical residual analysis and the assessment of the transport loss factor. Verification and validation have been performed using data from a chain belt conveyor system. Eventually, a case study was conducted with promising results in terms of robustness and accuracy. In the industrial setting, dynamic weighing attained a benchmark accuracy of 95\%. Further research, with data over longer periods of time, is required to establish degradation patterns and optimize the model for broader applications.