Empirical Loss Model Optimization for the Prediction of Centrifugal Compressor Off-Design Performance

Abstract

Abstract Certain critical process applications, such as sub-atmospheric re-compression of cryogenic helium gas supporting particle physics and other scientific research, require operating multi-stage centrifugal compressors over a wide operating range. Customarily, turbomachines are designed for a specified design condition, but in these applications it is essential that these machines behave stably at off-steady-state-design conditions due to actual processes conditions and requirements that are inherently transient. Predictive algorithms have been developed which adopt various empirical correlations that estimate individual losses, thus allowing the compressor performance to be estimated. Some of these algorithms require complete knowledge of the impeller geometry to enable performance estimation at varying process conditions. However, often such complete knowledge may not available for differing reasons, although basic dimensions and geometry information is obtainable. The objective of the present study is to examine the numerous loss correlations available in open literature and evaluate the effects of each proposed correlation on performance prediction accuracy, thereby permitting an accurate and robust method to predict compressor performance over a wide range of operation. This paper recommends a methodology and set of loss correlations to accurately predict the pressure ratio and isentropic efficiency of centrifugal compressors operating over a wide range of off-design conditions.