Predictions of transient pressure pulsations in a pump-turbine from measurements for enhanced flexible operations

With the growing demand for ancillary services nowadays, pump-turbine units are increasingly tasked with operating under off-design conditions, managing extensive peak load regulation across a wide range. These frequent transitions into transient processes induce pronounced pressure pulsations within the vaneless zone of the pump-turbine, causing substantial stresses on the runner and severe vibrations throughout the unit. Consequently, accurately and realistically determining pressure pulsations is crucial, although it remains a challenging task. This study proposes a data-driven method to predict transient pressure pulsations in the vaneless zone, including their amplitude-frequency characteristics, based on hybrid measurements that incorporate both steady-state and transient data. The discretized and filtered transient start-up data is innovatively used to construct the prediction database. After mode decomposition, the amplitude-frequency characteristics of the pressure pulsation data are comprehensively considered for predictions. Compared with experimental data, the reduced-order model achieved satisfactory predictions in both start-up and load variation processes, with a maximum prediction error of 2.8191 % in the frequency domain and a relative error of 1.7090 % for the peak amplitude, demonstrating the efficacy of the proposed framework. These results further corroborate that the amplitude-frequency characteristics of pressure pulsations under steady-state conditions closely resemble those affected by transient dynamic disturbances. This study not only provides a reference for analogous signal processing and prediction, but also offers promising applications in evaluating hydraulic stability during transient operations and optimizing control strategies to mitigate pressure pulsations, thereby enhancing the stability and operational flexibility of pump-turbine units.