Abstract

Refrigerant centrifugal compressors have been increasingly used in HVAC industry in chillers. The efficiency of the compressors poses significant impact on the energy consumption of these systems. Danfoss Turbocor has pioneered the oil free centrifugal compressors with magnetic bearing. The efficiency of the Turbocor compressors are much higher than the conventional centrifugal compressors. The design of the compressor involves large amount of modeling. It ranges from 1D mean line modelling to 3D Computational Fluid Dynamics (CFD) modeling. CFD modeling has become a standard practice in predicting the compressor performance. In HVAC industry, due to the complexity of the refrigerant, the accuracy of numerical modeling has been historically challenging. In this paper, attempts were made towards developing a CFD model approach that minimizes the deviation from test data for a centrifugal compressor using real refrigerant gas. CFD models were developed with different levels of the physics fidelity. The first model considered the basic components of a two-stage centrifugal compressor: inlet guided veins, first stage impeller, first stage diffuser, return channel, second stage impeller, second stage diffuser, and volute. The second model accounted for the leakages from the gaps between the compressor parts. The third model added surface roughness and accounted for the interaction between the refrigerant and the back of the second stage impeller. The pressure ratio and efficiency curves predicted by each model were compared to experimental data from a chiller test. Results showed that as additional physics were introduced to the model, the error in the numerical model reduced significantly, leading to higher fidelity in the prediction of performance curves. The most comprehensive model showed an average deviation in the pressure ratio prediction from the test data of 1% for every speed line analyzed, and a deviancy of 3.5%, 1.5%, and 1% in the compressor efficiency for a low, medium, and high rotational speed, respectively. The model provided a conservative prediction of choke and stall flow rates for each speed line analyzed. Additionally, the influence of each new physic over the flow conditions was analyzed, highlighting their effect over the thermophysical properties at the inlet and outlet of the compressor. Particularly, it was noticed that the leakages have a more pronounced effect over pressure ratio and corrected mas flow rate, while surface roughness plays a significant role over the efficiency prediction. This happens because a rough surface tends to generate more friction during the process of compression of the working fluid.