Prediction of particle erosion in the internal cooling channels of a turbine blade

A multi-phase computational approach was adopted for predicting particle erosion in a domain that is representative of the trailing edge region of a rotor blade. The flow field was solved by an in-house FV code adopting a non-linear k-eps-zeta-f elliptic relaxation RANS turbulence model. The model demonstrated to be able to reproduce the anisotropy of near-wall turbulence and partly the influence of streamlines curvature on the turbulent flows. Furthermore, URANS predicted very strong unsteadiness that allowed to reconstruct part of the turbulence spectrum and to identify the relevant frequencies. A Lagrangian particle tracking model was used for predicting particle dispersion and erosion of the solid surfaces. Impact mechanisms was modelled using the Tabakoff’s model, previously validated by the authors on a number of studies. Analysis of the eroded surfaces allowed to assess the importance of an accurate, unsteady description of the flow pattern to obtain a credible prediction of the erosion mechanism.