| dc.description.abstract | The aftermarket maintenance service of gas turbines came with a lot of technical challenges that included; low availability of the power plant (33.2) %, poor reliability (66%), and high maintenance costs (28.6%). This was because the guaranteed performance deteriorated even under strict adherence to the manufacturer’s recommended maintenance program. This deterioration was managed by condition-based maintenance (CBM), which was applied through Gas Path Analysis and performance monitoring. The Condition-Based Maintenance strategy was always applied rigidly without assessing the availability, reliability, and need for real-time maintenance requirements. The gas turbines were always subjected to unnecessary major maintenance and overhauls when still in good operating condition and also at times, failing before the guaranteed manufacturer’s approved safety margin of operation. This, therefore, called for an urgent need to develop a performance monitoring approach, that quantifies degradation and evaluates the realtime maintenance effectiveness. This performance monitoring approach used operational data from real installation and a novel model-free data analytics method. This had the advantage of the ease of applicability within the industrial setting. Five performance indices were proposed to be used and tested using the Anova Tool. Performance patterns for Parameter-dependent models were developed using the polynomial curve fitting methods. The maintenance system was evaluated and optimized regarding availability and reliability (safety margin) and overall effectiveness. These gas path parameters were Revolution per minute (N), Air inlet angle (A), Jet nozzle throat area (F), Turbine exhaust Temperature (T), and Vibrations (V). Results showed that Condition-based maintenance could be pushed to 6.3 weeks’ intervals with a range of (5.65±0.65) weeks instead of weekly intervals, however without forgetting the need for pre-post run inspections. This increased availability, and reliability by 26.2% that is to say (71.4 to 97.6) %, and lowered maintenance cost from 28.6% down to 2.5%. A higher performance level of up to 88.5% was achieved from 66.8%. For maintenance cost-effectiveness, a statistical selection method was a function of p-value and its weighted rank, percent error-value (E)and its weighted rank, and Restoration coefficient value (C) and its weighted rank. These were used to select the most effective and responsive parameters for performance indication and monitoring that had a percentage weighted index of above 80%. This boosted the maintenance effectiveness by 20.7% (from 68.8 to 89± 0.5) %. | en_US |
