Numerical and Experimental Study of Topographic Speed-Up Effects in Complex Terrain

Abstract

Our research group is developing computational fluid dynamics (CFD)-based software for wind resource and energy production assessments in complex terrain called RIAM-COMPACT (Research Institute for Applied Mechanics, Kyushu University (RIAM)-Computational Prediction of Airflow over Complex Terrain), based on large eddy simulation (LES). In order to verify the prediction accuracy of RIAM-COMPACT, we conduct a wind tunnel experiment that uses a two-dimensional steep ridge model with a smooth surface. In the wind tunnel experiments, airflow measurements are performed using an I-type hot-wire probe and a split film probe that can detect forward and reverse flows. The results of the numerical simulation by LES are in better agreement with the wind tunnel experiment using the split film probe than the results of the wind tunnel experiment using the I-type hot wire probe. Furthermore, we calculate that the two-dimensional ridge model by changing the length in the spanwise direction, and discussed the instantaneous flow field and the time-averaged flow field for the three-dimensional structure of the flow behind the model. It was shown that the eddies in the downwind flow-separated region formed behind the two-dimensional ridge model were almost the same size in all cases, regardless of the difference in the length in the spanwise direction. In this study, we also perform a calculation with a varying inflow shear at the inflow boundary. It was clear that the size in the vortex region behind the model was almost the same in all the calculation results, regardless of the difference in the inflow shear. Next, we conduct wind tunnel experiments on complex terrain. In the wind tunnel experiments using a 1/2800 scale model, the effect of artificial irregularities on the terrain surface did not significantly appear on the airflow at the hub height of the wind turbine. On the other hand, in order to investigate the three-dimensional structure of the airflow in the swept area in detail, it was clearly shown that LES using a high-resolution computational grid is very effective.