Implementation of a Digital Twin-Based Safety Service Platform for Risk-Based Operation and Maintenance of Hybrid Regenerative Thermal Oxidizers : Focusing on Risk Assessment and Platform Technologies

Abstract

This study aims to propose an implementation framework for digital twin-based safety service monitoring technology to support the safe operation and advanced maintenance of regenerative thermal oxidizers(RTOs). Because RTOs involve multiple risk factors such as high-temperature combustion, regenerator clogging, valve stiction, burner misfire, and abnormal differential pressure, individual sensor alarms alone are insufficient to identify abnormal causes and fault propagation paths. Therefore, this study proposes a safety service model that integrates key risk and performance factors with a 3D equipment model, operational data, risk assessment logic, and SOP-based response procedures. The research procedure consists of risk factor identification, data collection and preprocessing, 3D object mapping, risk assessment, SOP linkage, 3D safety service implementation, and KPI-based evaluation. Operational data collected from PLC/DCS layers, including temperature, pressure, differential pressure, VOC concentration, valve status, burner ignition status, alarm logs, and trip logs, are processed through time synchronization, noise removal, and outlier detection in the Edge-IoT layer. The processed data are stored in a digital twin database composed of Redis, MySQL, time-series databases, sensor metadata databases, and object mapping tables, and are linked to 3D equipment objects. Risk assessment is designed by combining trip conditions, a Health Index, and HAZOP rules to determine equipment health, risk priority, and the need for corrective action.The results show that RTO operational data can be transformed from numerical monitoring data into spatial safety information by linking them with equipment objects, sensor locations, piping routes, and maintenance histories. The 3D digital twin model can visualize the status of the combustion chamber, ceramic regenerator, ducts, valves, burner, fans, and rotary unit, while providing abnormal locations, hotspots, fault propagation paths, and shutoff or bypass points. In addition, by linking HAZOP scenarios with SOPs, the system can automatically provide cause analysis, standardized inspection procedures, maintenance priorities, and worker safety information when alarms occur. In conclusion, this study expands RTO safety management from individual sensor monitoring and reactive maintenance toward data-driven condition diagnosis, spatial risk visualization, and risk-based maintenance decision-making.The proposed safety service can contribute to early abnormality detection, alarm refinement, improved SOP compliance, reduced hazardous work exposure, and a transition to predictive maintenance. Future research should quantitatively validate KPIs such as false alarm rate, missed detection rate, MTBF, MTTR, unplanned downtime, and energy consumption changes using long-term operational data from industrial sites.