Based on existing reports and databases, most of the installations in highly turbulent sites in fact fail to reach the expected energy yield, resulting in still or underperforming turbines that also give bad press for the technology. A better understanding of the real performance of wind turbines under highly turbulent conditions is then pivotal to ensure the economic viability of new installations. To this end, the possible use of Computational Fluid Dynamics (CFD) techniques could provide notable benefits, reducing the time-to-market and the cost with respect to experiments. On the other hand, it is intrinsically not easy to reproduce properly intense and large-scale turbulence with the techniques of common use for research and industry (e.g. CFD unsteady RANS), while the only methods that are granted to do so (e.g. DNS or LES) are often not computationally affordable. Moving from this background, this study presents the development a numerical strategy to exploit at their maximum level the capabilities of an unsteady Reynolds-Averaged Navier-Stokes (RANS) approach in order to reproduce fields of macro turbulence of use for wind energy applications. The study is made of two main parts. In the first part, the numerical methodology is discussed and assessed based on real wind tunnel data. The benefits and drawbacks are presented also in comparison to other existing methods. In the second part, it has been used to simulate the behavior under turbulence of a H-Darrieus vertical-axis wind turbine, for which unique wind tunnel data were available. The simulations, even if preliminary, showed good matching with experiments (e.g. confirming the increase of power), showing then the potential of the method.