Accounting for the Effects of Experimental Setting in Propeller Flow Computation

Abstract

Various computational approaches differing in complexity and accuracy can be used to simulate flows around rotating lifting surfaces. Here, the flow field around a hovering propeller is estimated by two disparate methods, Reynolds-averaged Navier-Stokes (RANS) equations vs. wall-modelled large eddy simulation (WMLES), and compared to experimental data. In both cases, the computed thrust is somewhat lower than the measured value, mostly because the measuring equipment, located behind the propeller, disrupted the wake, thus changing the aerodynamic loads over the blades. However, the effect of the computational grids and approaches to resolving the rotational motion and turbulence are not negligible. They can be quantified in a particular case of a small-scale propeller operating at nominal angular frequency. For that reason, different computational domains, grids and numerical set-ups are tried and compared. It is observed that by careful geometric modelling, fine meshes, LES and inclusion of actual experimental set-up, a much-improved correspondence between the two sets of data can be achieved.