Low Mach preconditioning for turbomachinery flow simulations with cavities and variable gas compositions

Abstract

The optimization of turbomachines increasingly relies on highly accurate numerical performance predictions. Loss predictions require the cavities of the machine to be included in numerical simulations. Commonly, in cavities, the velocity of the simulated fluid is small. For density-based solvers, this results in slow convergence and inaccurate computations. Further, the fluid in cavities is often composed of several gases. This paper presents the low Mach preconditioning method for multi-component thermally perfect gas of DLR’s inhouse solver TRACE. Two low Mach academic test cases, a lid driven cavity and an air and exhaust gas mixing layer, are computed to validate the preconditioner. Both test cases show an accelarated convergence and an improved accuracy, when preconditioning is used. A 1.5 stage low-pressure turbine rig with a labyrinth seal is computed with thermally perfect air. The result shows a good agreement with the experimental reference. The fluid is then changed to exhaust gas, and two air inflows are added in the labyrinth seal, to analyze the effect of low Mach preconditioning on the mixing of the two gases. The preconditioned computation shows an improved convergence in the cavity. Moreover, the wall temperature and the gas distribution in the cavity differ, when preconditioning is applied.