Convergence Criteria for Axial Compressor Flow Calculations

Abstract

Computational fluid dynamics is routinely used in the turbomachinery industry to aid in the design of axial flow compressors. The predictive capability of such codes is related to the quality of the numerical convergence of the flow solutions they produce. In certain cases convergence cannot be attained and various publications have linked this to the boundary conditions used within the code. In this paper an investigation in to how common types of boundary conditions affect the numerical convergence is described. The point at which steady-state calculations fail to predict the increasing non-axisymmetric flowfield at off-design, part speed operation is identified. The analysis of the convergence process is combined with numerical experiments to show that the rate of convergence of steady-state mixing-plane multi-stage axial compressor calculations depends upon the operating point on the pressure-rise versus mass flow rate characteristic. Intrinsically, as the calculated overall characteristic reaches its peak the rate of convergence decreases to zero. Ways to enhance the rate of convergence, for example the technique of adding a downstream nozzle, and conditions under which such techniques are likely to be successful are discussed.