Steady-state hybrid modeling of economized screw water chillers using polynomial neural network compressor model

This paper presents steady-state hybrid modeling and analysis of water chillers with economized screw compressors using intermediate gas injection. The hybrid chiller model consists of the polynomial neural network compressor model and other component models grounded on the first principles. The polynomial neural network compressor model works for all full-load and part-load conditions, both economized and non-economized modes. Good agreement between the predicted and measured performance of two compressors from the manufacturer is reached ranging from full-load to unload conditions. The hybrid chiller model predictions on the performance of two chillers agree with the test data within ±5% errors. Based on the validated model, further investigation on the optimal switch points between the economized and non-economized mode is conducted, which can be used to precisely improve the part-load performance of chillers. A simple linear correlation for switching the economized and non-economized mode is proposed as well.