Chaotic characteristics of pressure fluctuations and operation state recognition in S-shaped region of a pump turbine
Boxing Liu, Jianjun Feng, Zhu Guojun, Zhenguo Ge, Luhan Gao, Xingqi Luo
Abstract
Strong pressure fluctuations in pump-turbine S-shaped regions threaten grid stability. This study investigates vaneless space (VS) and draft tube (DT) pressure fluctuations within the S-shaped region through tests, combining time and frequency domain analysis and chaos theory. Time and frequency domain analysis reveals that the root mean square of pressure fluctuations peaks at the runaway condition in VS and DT. When the operating condition enters the S-shaped region, the probability density function evolves from unimodal to multimodal. Low-frequency broadband fluctuations increase in both regions before zero-flow. Based on the improved complete ensemble empirical mode decomposition with adaptive noise (ICEEMDAN) denoising and phase space reconstruction, the minimum covering sphere theory was employed to quantify the apparent characteristics of the attractor trajectory. Three chaos indicators are extracted: Correlation Dimension, Largest Lyapunov Exponent, and Kolmogorov Entropy. The results demonstrate that VS pressure fluctuations are more sensitive to operating condition transitions, with significant changes observed in the chaos indicators. Conversely, DT pressure fluctuations maintain stable high-dimensional chaos but exhibit greater sensitivity to the emergence of the runaway condition. The constructed chaos feature space effectively identifies operating condition changes and delineates instability regions, providing theoretical support for monitoring, control optimization, and fault early warning in pumped storage systems.