*3.1. Experimental Validation of the CFD Model*

The experimental tests, carried out in the subsonic wind tunnel owned by the University of Catania, demonstrated that the small VAWT rotor object of this study is much less efficient than a HAWT of comparable size. The maximum measured power of the micro rotor presented here was approximately only 0.55 W at a flow speed of 21 m/s. A micro HAWT with comparable dimensions, tested by the authors in a previous work [11], showed a power of approximately 10 W at the same flow speed.

Concerning the experimental validation of the CFD model, in Figure 6 the comparison between the measured and the calculated power coefficients is shown. The experimental data in Figure 6 refer to the three different braking loads equal to 25 g, 50 g, and 75 g in mass. Due to the low torque generated by the rotor, the torque dissipated by the bearings limited the range of measurement approximately at λ = 0.3. Owing to the unphysical prediction obtained with the RANS turbulence models, only the DDES simulation results are reported in Figure 6. Both the URANS SST transition and SST k-ω model predicted unrealistic negative power coefficients for almost the entire operating range of the rotor. Furthermore, in Figure 6, the Qblade Double Multiple Stream Tube Model (DMSTM) power coefficient prediction is plotted. Qblade is an open source 1D code [37] that required the use of suitable polars for the airfoils. In this case, the experimental data of the NACA 0012 were taken from the literature [38], for an average cord Reynolds number of approximately 40,000. The DMSTM implemented within Qblade uses advanced models and corrections for tip losses, dynamic stall and virtual camber, similar to those used in the BEM theory for HAWTs [39].

The close proximity between experimental measurements and simulation results proved an excellent accuracy of the CFD 2D model based on DDES [40]. Considering the unphysical predictions obtained with the RANS turbulence models despite the very fine spatial and temporal discretization level, this result is very meaningful. In the simulation of strong boundary layer instabilities in VAWTs, the use of an advanced turbulence model like DDES appears to be necessary.

**Figure 6.** Comparison between numerical and experimental power coefficient.

The 1D Qblade results in Figure 6 showed surprising accuracy as well. At least the first part of the Cp curve is well predicted in light of the extreme simplicity and rapidness of the Qblade model. This certainly deserves further investigations on different geometries and operating conditions, in order to verify whether it is a mere chance or a generalizable result.

Nevertheless, CFD still remains the only way to get a thorough insight into the rotor aerodynamics, which is of utmost importance for the comprehension of the causes of the poor efficiency of such micro VAWTs. For this purpose, the present work demonstrated that only an advanced turbulence modeling like the hybrid RANS-LES formulation, implemented in DDES, was able to provide accurate and physically reliable results. Moreover, the availability of an accurate CFD model will allow the authors to identify an optimization strategy for these rotors in order to increase their efficiency. The use of more suitable airfoils, specific pitch angles, and vortex generators are just some of the simplest and cheapest techniques, whose effectiveness will be evaluated thanks to the CFD model developed in this work.
