Effect of Wakes and Secondary Flow on Re-Attachment of Turbine Exit Annular Diffuser Flow

Abstract

In this paper numerical results of wake and secondary flow interaction in diffuser flow fields are discussed. The wake and secondary flow are generated by a rotating wheel equipped with 30 cylindrical spokes with a diameter of 10 mm as a first approach to the turbine exit flow environment. The apex angle of the diffuser is chosen such that the flow is strongly separated according to the well-known performance charts of Sovran and Klomp [1]. This configuration has been tested in an experimental test rig at the Leibniz University Hannover [2]. According to these experiments, the flow in the diffuser separates as free jet for low rotational speeds of the spoke-wheel as expected by theory. However, if the 30 spokes of the upstream wheel rotate beyond the value of 500 rpm the measurements indicate that the flow remains attached to the outer diffuser wall. It will be shown by the present numerical analysis with the commercial solver ANSYS CFX-10.0 that only an unsteady approach using the elaborate SAS-SST turbulence model is capable of predicting the stabilizing effect of the rotating wheel to the diffuser flow at larger rotational speeds. The favourable comparison with the experimental data suggests that the mixing effect of wakes and secondary flow pattern is responsible for the reattachment. As a result of our studies it can be stated that the considerably higher numerical costs associated with unsteady calculations must be accepted in order to increase the understanding of the physical flow phenomena in turbine exit flow and its interaction with the downstream diffuser.

On the Numerical Prediction of the Influence of Tip Flow on Diffuser Stability

“…Viewed in an historical context, the SAS model is an improvement of the Detached-Eddy Simulation (DES) approach by [14], with lower sensitivity to grid resolution. Ergorov and Menter [10] validated the SAS modification of the SST turbulence model for a series of applications Kluß et al [6] showed the advantages of the SAS-SST model concerning the flow separation prediction of a highly loaded diffuser downstream of a single stage rotor.…”

“…the Reynolds shear stresses of the SAS simulation (6) are the sum of the resolved stresses by the SAS term and the modeled turbulence (k-term). The resolved stresses include solely statistical variations, whereas the stresses derived from the Bousinesqhypothesis includes both deterministic and stochastic stresses (as long as they are not resolved by the SAS approach).…”

“…Turbine outflows are inherently unsteady and incorporate swirl and inhomogeneous flow conditions due to tip leakage flows and wakes. These conditions were also found to affect the diffuser flow considerably [3][4][5][6][7]. Sieker and Seume [5] showed that highly turbulent, unsteady inflow can stabilize the flow field even if the diffuser was predicted to separate according to the diffuser charts.…”