Real-Time Monitoring of Vacuum Leaks in Condenser Pipes Using Ultrasonic Acoustic Signals

Abstract Vacuums in air-cooled condensers (ACCs) are essential for maintaining the efficient operation of thermal power plants. Air ingress caused by vacuum leaks increases condenser pressure, degrades heat transfer performance, raises turbine back pressure, and ultimately reduces plant efficiency. Previous research has primarily focused on leak detection in pressurized pipeline systems, whereas vacuum leak detection in ACCs remains largely unexplored due to the limitations of existing techniques and harsh industrial operating conditions. This paper proposes a portable real-time ultrasonic leak detection (RULD) system for detecting air ingress into vacuum-operated condenser pipes using ultrasonic acoustic signals. The proposed system employs three ultrasonic microphones to acquire leak-generated acoustic signals and processes them using the root mean square (RMS) method with adaptive noise-based thresholds. Leakage is identified when the RMS values of at least two microphones exceed their respective thresholds. Experimental investigations were conducted on four irregular leak geometries (hydraulic diameters 1.19–1.43 mm) under a 0.5-bar vacuum, considering microphone distances of 5–20 cm and angles of 0°–90°. The results showed that the 20–40-kHz ultrasonic frequency band provides the highest sensitivity for leak detection. The 0° microphone orientation produced the maximum RMS response, which decreased with increasing microphone angle, while reliable detection was achieved up to 10–15 cm from the leak source. The developed RULD system processed each 1-s acoustic data segment in 0.0009 ms, enabling real-time leak detection without interrupting subsequent data acquisition. The proposed system was further validated on condenser pipes and a prototype ACC model, demonstrating its suitability for real-time monitoring of vacuum-operated condenser systems.