Relationship between the leakage rate and acoustic emission of dry gas seals

Leakage is a critical defect in dry gas seals, and online monitoring based on acoustic emission (AE) offers a promising solution. However, quantitative relationships between AE signals and leakage rates remain underexplored. In this work, a theoretical model linking the root mean square (RMS) of AE signals to the leakage rate of dry gas seals is established and validated experimentally. AE signals were collected under inlet pressures of 2, 3, and 4 Mpa and rotational speeds ranging from 200 to 15,000 r/min. Through comparative no-load tests, the characteristic frequency band of leakage-induced AE was identified as 40 to 200 kHz, and a Butterworth bandpass filter was applied to remove environmental and mechanical noise. The filtered AE RMS values and measured leakage rates exhibited a strong power-law relationship. At 2 Mpa, the fitted relationship yielded an average relative error of 8.76% and a correlation coefficient of 96.31%; at 3 Mpa, the average relative error was 5.88% with a correlation coefficient of 98.05%; at 4 Mpa, the average relative error was 11.27% with a correlation coefficient of 94.23%. The maximum prediction error among all tested conditions was 14.87%. The results demonstrate that the proposed AE-based method can effectively predict the leakage rate of dry gas seals, providing a quantitative tool for online condition monitoring.