Self-adaptive Surge Adjustment based on Fast Correlation Integral Method in Centrifugal Compressors

Abstract

The research result that correlation integral (i.e. CI) curve shows a continuous decrease of 1~2 second in CI amplitude until the advent of flow instability such as stall and surge opens a way for self-adaptive surge adjustment. However, the existing computational method of correlation integral curve is very time-consuming, which makes the promising control strategy fail to be used for real-time application. A fast algorithm of the curve computation is presented based on the concept of segmentation in this paper. The improved method is much faster than the existing one for all reduplicate computation of the number of correlated pairs is eliminated using our proposing propositions and criterions. Applying the improved method to some cases of our experiment, it is demonstrated that the correlation integral had a remarkable drop of about 0.25 order of magnitude prior to the development of surge and the elapsed time is much shorter than the requisite time of the existing method. The fast algorithm will motivate the real-time application of correlation integral curve to the control system using the value as the feedback threshold, which will extend the normal operating range of the centrifugal compressor.