Assessment of SU2 for radial compressor performance prediction

Abstract

The open-source SU2 suite has been subject to extensive verification and validation studies in the past few years. Through such studies, the ability of the SU2 CFD solver to produce high-fidelity results for external flows has been substantiated by data from numerous testcases. Interest in applying the suite to turbomachinery flow analysis and optimization has been on the rise in recent times. This has led to a significant extension of SU2 capabilities to deal with turbomachinery flows. Evidence of the ability of SU2 to carry out CFD analysis of ideal-gas turbomachinery flows is scarce; the suite has only been applied to a radial-inflow turbine case thus far. The present Thesis documents the first application of SU2 to radial compressor ideal-gas flow analysis. Two testcases are studied: an automotive turbocharger compressor including experimental data shared by MTEE and a mixed-flow Supercritical CO2 compressor including a baseline CFD result. SU2 capabilities are assessed via a solution verification study including a grid convergence study and through a comparison of SU2 results with those of CFX and experimental data for the turbocharger. The assessment is based on major compressor performance metrics: efficiency and pressure ratio. RANS steady-state, second-order accurate simulations are carried out for a single, voluteless compressor channel using the S-A and SST turbulence model with mixing-plane interfaces using non-reflective boundary conditions. The comparison includes an analysis featuring SU2 S-A and SST simulations as well as a three-point speedline comparison featuring SU2 S-A results. The Supercritical CO2 compressor testcase is preliminarily studied, again including results from CFX as well as SU2 for a single operating point. It is concluded that SU2 is capable of arriving to an asymptotically converged solution in the verification study while displaying a convergence rate similar to the theoretical value. In addition, SU2 SST results are in well agreement with simulation results. The speedline obtained using SU2 also matches that of CFX and experimental data to a good extent quantitatively as well as qualitatively. For the S-CO2 compressor the disparities between the solvers are notorious, which is likely due to combination of factors including near-wall mesh treatment in SU2 and the complex geometry of the testcase.