The implementation of wind energy conversion systems in the built environment has renewed the interest and the research on Vertical Axis Wind Turbines (VAWTs). The VAWT has an inherent unsteady aerodynamic behavior due to the variation of angle of attack and perceived velocity with azimuth angle. The phenomenon of dynamic stall is then an intrinsic effect of the operation at low tip speed ratios, impacting both loads and power. The complexity of the problem and the need for new design approaches for VAWTs for the built environment have driven the authors to focus this research on the CFD modeling of VAWTs on: • Comparing the results between commonly used turbulence models: Unsteady Reynolds Averaged Navier-Stokes - URANS (Spalart-Allmaras and k-e) and large eddy models (Large Eddy Simulation and Detached Eddy Simulation). • Verifying the sensitivity of the model to its grid refinement (space and time). • Evaluating the suitability of using Particle Image Velocimetry (PIV) experimental data for model validation. The current work investigates the impact of accurately modeling the separated shed wake resulting from dynamic stall, and the importance of validation of the flow field rather than validation with only load data. The structure and magnitude of the wake are validated with PIV results, and it demonstrated that the accuracy of the different models in simulating a correct wake structure has a large impact in loads. Copyright © 2009 John Wiley & Sons, Ltd.