Abstract
Centrifugal compressors play a critical role in various industrial applications, and understanding their operational behavior, especially during surge conditions, is essential for enhancing efficiency and reliability. This study investigates surge transitions in centrifugal compressors by employing a comprehensive approach that integrates dynamic mass flow rate and pressure measurements together with dynamic structural response data. To gain insight into the transition from the stable to the unstable region of the compressor, an extensive experimental analysis on a centrifugal compressor was conducted. The activity involves measurements used to characterize and identify the behavior of the compressor in correspondence of surge inception conditions. Frequency and time-frequency data analysis techniques have been employed to examine the inflow pressure and the anemometric signal. The purpose is to identify their characteristics and define the compressor operation as stable or unstable. Synchronous averages, applied in the time domain, have proven to be effective for detecting incipient surge conditions. Additionally, statistical techniques, such as variance and spectral kurtosis, have been utilized for the detection in signals with strong additive noise and non-deterministic contents. A method for system identification in deep surge is introduced, particularly focusing on the correlation between the compressor circuit geometries and the operating conditions of the turbocharger. This approach enhances the diagnostic capability of the monitoring system by providing a better knowledge of surge occurrence. The paper is aimed to provide information for the development of control strategies and predictive maintenance protocols to mitigate surge-related issues.
