Modeling of Supercritical Carbon Dioxide Centrifugal Compressors for Brayton Power Cycles

Abstract

Due to several perceived advantages, supercritical Carbon dioxide (sCO2) cycles are viewed as potential replacements for steam-based power generation. Their successful deployment depends on several factors - one important aspect being the realization of high efficiency turbo-compressors. Although the work of compression is reduced in a sCO2 Brayton cycle compared to an ideal gas cycle, the compression work is still a notable fraction of the turbine work. Consequently, a mild decrease in the achievable compression efficiency can adversely affect the overall cycle performance. Therefore, modeling and analyzing sCO2 centrifugal compressors is an important exercise. In the first part of this work, we show that 3D Computational Fluid Dynamics (CFD) coupled with Real Gas Tables (RGP) can satisfactorily model flows in real gas sCO2 centrifugal compressors. We also show that the existing loss models can be used to model the losses in the rotor satisfactorily. However, explicit loss model equations for the vaneless diffuser (VLD) are currently unavailable for real gas sCO2 flows. Hence, the vaneless diffuser governing equations are derived for real gas flows from first principles. These equations require the skin friction coefficient as input. From CFD data, the existing skin friction coefficient estimates are found to be inaccurate. Therefore, the functional relationship of the skin friction coefficient is established using an Artificial Neural Network (ANN) approach for real gas sCO2 flows. A two-hidden layer ANN is found to satisfactorily capture the functional relationship of the skin friction coefficient with real gas sCO2 flows in vaneless diffusers. The 500 kW and 5 MW net power scales are perceived as the lower and upper bounds for sCO2 Waste Heat Recovery (WHR) in conjunction with centrifugal compressors (Axial compressors are used for higher power scales). The achievable centrifugal compressor efficiencies are estimated for these two representative WHR power scales- 500 kW and 5 MW net power. The 1D loss analysis shows that an overall single-stage centrifugal compressor efficiency of 89.7 % is achievable for the 5 MW scale. The achievable single-stage compressor efficiency in the 500 kW WHR power scale case is only 65.9 % due to high windage and vaneless diffuser losses. A two stage intercooled compression process is subsequently examined as an alternative for the 500 kW scale. In the case of the two-stage inter-cooled compression process, the ratio of the single-stage isentropic power to the actual net power consumed is found to be 82.2 %, which is a marked improvement compared to the single-stage compressor isentropic efficiency. These design studies are expected to pave way for commercializing the sCO2 Brayton cycle as an alternative to steam-based power cycles for WHR.