PUMP

Modern power generation facilities depend heavily on auxiliary components such as centrifugal pumps, which ensure effective cooling and stable operation of gas turbines. The literature reviewed shows that conventional maintenance strategies reactive and preventive—are often costly and inefficient, leading to unexpected failures and operational losses. Predictive maintenance (PdM) has emerged as a superior, data-driven alternative that uses statistical and sensor-based models to forecast equipment failure. The review further highlighted the growing adoption of PdM techniques in African power systems, where the need for reliability and cost optimization remains high. This study focuses on developing a predictive maintenance model for the cooling water centrifugal pump at the Azura-Edo Independent Power Plant, using statistical trend and regression analysis to predict performance degradation. The research employed an analytical and quantitative design, utilizing two years (2023–2024) of historical operational data from Azura-Edo IPP. Key parameters included ambient temperature, discharge pressure, gas turbine active power, and vibration readings from different pump locations. Microsoft Excel served as the main analytical tool for data cleaning, descriptive statistics, correlation testing, and multiple regression modeling. The regression model related vibration amplitude to operating parameters, producing a mathematical expression capable of estimating degradation levels. A control chart was also developed to monitor vibration stability using calculated upper and lower control limits, forming an early warning system for predictive maintenance intervention. Results from the analysis revealed moderate variability among parameters, with vibration showing the strongest correlation to discharge pressure and turbine power. The developed regression model effectively predicted vibration trends with reasonable accuracy, confirming its suitability for maintenance forecasting. The study concluded that predictive maintenance can significantly improve pump reliability, reduce unplanned downtime, and optimize maintenance scheduling at Azura-Edo IPP. It is recommended that the model be integrated into the plant’s SCADA system for real-time monitoring, with periodic updates to ensure adaptive accuracy and sustainable performance.