High-fidelity CFD simulations of two tandemly arrayed wind turbines under various operating conditions

In this study, we perform computational fluid dynamic (CFD) simulations for the NTNU Blind Test 2 experiment in which the wake of two turbines was measured in a closed-loop wind tunnel and operating in line. The Reynolds-Averaged Navier Stokes (RANS) equations with the k-omega SST turbulence model are adopted in the simulations. For each of the two wind turbines, geometries including the blades, hub, nacelle, and tower are fully resolved. The Moving-Grid-Formulation (MVG) approach with a sliding interface technique is leveraged to handle the relative motion between the rotating and the stationary portions of the wind turbines. In the simulations, three experimental configurations were numerically investigated. For each of the configurations, the tip- speed ratio (TSR) of the upstream wind turbine is fixed at 6, while the TSR of the downstream wind turbine is changing, i.e. TSR = 4.0 in test-case A, TSR = 7.0 in test-case B, and TSR = 2.5 in test-case C. The CFD-predicted wake profiles including the mean streamwise velocity (U) and the mean streamwise turbulent fluctuation (u2) profiles at different downstream locations behind the downstream wind turbine are obtained and compared with the experimental measurement. It is demonstrated that velocity profiles match the experimental data exceptionally in that the asymmetry in the measured profiles was successfully captured. For the turbulent fluctuations, although the trends are well captured, their magnitude is underpredicted by about 50% compared to the measurement. Then, it was found that the magnitude on one side of the turbulent fluctuation profiles is much higher than on the other side which is consistent with the experiment. Detailed analyses and discussions of the wake evolution were then followed to elucidate the mechanism behind this phenomenon. We then confirmed that the occurrence of this phenomenon is due to the influence of the tower-wake generated by the upstream wind turbine, and thus emphasizing that all geometric details, such as the tower, should be included in simulations to investigate wake interactions of multiple wind turbines. To the authors' knowledge, this work is the first to present a clear and detailed discussion of the mentioned phenomenon in the NTNU BT2 wind tunnel experiment.