Development of a boiler monitoring system for predictive maintenance of boiler tube failures

Abstract Unplanned outages in coal-fired power stations are significantly driven by premature boiler tube failures, contributing to over 30 % of forced shutdowns across certain utility fleets. This paper presents a case study on the design, development, and pilot implementation of a predictive boiler monitoring system targeting supercritical boiler reliability. The boiler monitoring system showcases a transformative path toward predictive maintenance and operational efficiency in legacy power infrastructure. The system integrates a 3D interactive model, digital shadow, and predictive analytics to monitor and forecast boiler tube degradation. Over 240 000 inspection data points across multiple boiler circuits were digitized and analyzed with the goal of determining the remaining wall thickness and time to minimum allowable limit for the boiler tube walls. A predictive model based on the power law regression was trained successfully using historical data measured during 3 outages. The monitoring system was able to forecast boiler tube wall thickness with success. At 1 year the deviation was 2.5 %, which increased to 8.1 % for 9 years, a value which is acceptable based on ASME Boiler and Pressure Vessel standards. As such, by knowing the expected boiler tube wall thickness, maintenance planning is enhanced through reduction of expected outage time of 4 weeks per year to address repairs. Another benefit is the automated generation of quality assurance, quality control and cutting instruction reports.