Real-Gas Centrifugal Compressor Design and Optimization for Hydrogen Blends and CO2-Rich Mixtures

Abstract

Abstract. The transition toward low-emission energy systems requires centrifugal compressors capable of operating with real-gas working fluids and multicomponent mixtures. This paper presents the meanline and streamline curvature design tool (MSCDT), an automated multi-objective optimization framework for impellers and vaned diffusers. MSCDT integrates one-dimensional meanline and two-dimensional streamline curvature methods with the NSGA-II genetic algorithm, incorporating real-gas thermodynamics via refprop. The framework is applied to compressors operating with H2, CO2, CH4, and their mixtures, as well as CO2 in the liquid-like region. Optimized geometries are automatically exported for three-dimensional computational fluid dynamics (CFD) validation. Compared to traditional meanline-based designs, MSCDT achieved isentropic efficiency improvements of up to 3%, with all pure-fluid cases exceeding 90% efficiency in model-based predictions. For mixture cases, agreement between meanline predictions and CFD results remained within 4% for pressure ratio and 3% for efficiency. Thermodynamic analysis shows that mixture composition affects compressor performance primarily through changes in density, molecular weight, and speed of sound, thereby modifying the compression work. MSCDT requires less than 0.1% of the computational cost of conventional 3D CFD-based optimization. By combining automation, real-gas compatibility, and multifluid capability, MSCDT provides a scalable and efficient tool for preliminary centrifugal compressor design in advanced energy systems, including sCO2 Brayton cycles, H2 infrastructure, and CCS applications.