Test problem of the flow modeling in axial compressor cascades

Abstract

The flow of gas in the flow path of a gas turbine engine (GTE) is accompanied by a rather complex phenomenon. These are a three-dimensional boundary layer, an incoming vortex, a paired vortex, flow turbulence, aerodynamic wakes behind the trailing edge, separation of the boundary layer from the blade surface, pressure pulsations, uneven and unsteady flow, secondary overflows, changes in the angles of flow exit, etc. Flow R&D of a GTE remains a rather complex process, and requires the use of reliable research methods and techniques. Nowadays, two known methods are used to study a gas flow through the flow path of a GTE ˗ experimental and calculated. Calculated, in turn, can be divided into analytical and numerical. An important stage of the numerical experiment is the solution to test problems for the possibility of setting the parameters of the numerical experiment. In this work, two test tasks were carried out. The object of the research was two compressor cascades, consisting of the identical airfoils series KR-33. The profile chord was 52 mm; the pitch cascade was 52 mm. The difference was in the installation angle of these profiles: variant 1 of the compressor cascade has an installation angle of 63.5º; variant 2 of the compressor cascade has an installation angle of 89.5º. A computational domain was constructed for each compressor cascades of airfoils and consisted of 5 million cells. Air under normal atmospheric conditions was chosen as the working fluid. The flow regime of compressor cascades varied in the range of coefficient λ = 0.26…0.9 and λ = 0.265…0.8, where the coefficient λ is the reduced velocity. The unstructured mesh method with an adaptation for the boundary layer was chosen to construct the computational mesh. Such a combination makes it possible to correctly model the flow in the boundary layer near the walls. The turbulence model SST was taken to close the Navier-Stokes equations. A comparison of the results of numerical and physical experiments for two variants of compressor cascades shows that the flow simulation error is less than 5%. Because of the calculation, the choice of this turbulence model for subsequent studies of the flow in the stages of the compressor, fan, and propfan will be justified.