The analysis, modeling and design of the lift-driven Vertical Axis Wind Turbine (VAWT) has challenged the wind energy community for many decades; this limited progress in knowledge has severely impaired the development of the VAWT, giving rise to the myth that the VAWT rotor is inherently inefficient (in comparison with the more conventional Horizontal Axis Wind Turbine - HAWT) or too complex for commercial implementation. In this research work, we take a new path on the analysis of the VAWT: instead of considering a rotor that creates a perturbation on the flow (wake and induction field), we consider an unsteady wake, to which a rotor energy-conversion system is associated, obtaining the loading on the blade by better understanding the flow. The research aims at understanding the wake and its relation with energy conversion and the loading on the rotor system. Four main questions drive this research: • What is the relation between blade loading and energy conversion? • How does the near wake of the VAWT develop? • What is the difference between the 2D and the 3D wake? • How does understanding the near wake improve our design? At the end of this dissertation we achieve a clear and insightful view on the 2D and 3D aerodynamics from the point of view of the wake, that significantly improves the aerodynamic design and optimization of new VAWT rotors for energy conversion and propulsion, opening a new design space and methodology. The results and discussion presented in this dissertation are organized in five steps (see thesis outline, Chapter 1): • Part I: understand the fundamental VAWT aerodynamics and how these relate with the research presented in this dissertation. • Part II: understand the energy exchange process in the 2D plane by understanding the shedding of the wake over the rotation, and the wake expansion in 2D. Analyze the impact of dynamic stall on the near wake evolution, and how to extract blade load information from the near wake in dynamic stall. • Part III: understand the impact of the spanwise dimension of the rotor and the role of the consequent trailing vorticity. Investigate the little known skewed flow. • Part IV: understand better the energy exchange process, the wake’s generation and the decoupling between loading and energy conversion. Propose new approaches and guidelines for the aerodynamic rotor design. • Part V: discuss the main results and conclusions of the research, and its impact on new aerodynamic research and design approaches, both for 2D and 3D VAWT rotors. In Part I (Chapter 2) we frame our research approach, analyzing the VAWT from a wake perspective, by considering both 2D and 3D aerodynamics of the VAWT at two different scales: aerofoil/blade scale and rotor scale. We divide the rotor in windward (315◦ < θ < 45◦), upwind (45◦< θ < 135◦), leeward (135◦ < θ < 225◦) and downwind (225◦ < θ < 315◦) regions of the rotation. This approach obsolesces the conventional division of the rotor into upwind and downwind halves; while the upwind/downwind division is driven by angle of attack considerations (blade loading problem), this new segmentation is determined by the shedding of vorticity (energy conversion problem), a more useful and effective approach. The wake is also split into shed vorticity due to the time gradient of the bound circulation, and trailing vorticity due to the spatial gradient of the bound circulation; this division leads to our 2D and 3D analysis of the flow. In Part II, we analyze the 2D rotor and wake at two scales: rotor and blade. The two flow scales are obviously related, in the sense that the rotor’s aerodynamics are the result of the wakes generated at the several blades and the blade experiences an induction field due to the vorticity distributed over the wake at the rotor scale. The separation in blade and rotor scale is in fact a separation of two views on the total system: • The rotor, as an energy exchange system, where the energy exchange results in a wake and streamline expansion. • The blade, as an aerodynamic loading system, where the design-objective loading is associated with an equivalent bound circulation. The time variation of this bound circulation results in a shed wake. The 2D potential-flow analysis (Chapter 3) shows that: • the conventional breakdown of the VAWT into upwind and downwind actuator systems is incorrect, leading to an overestimation of energy conversion on the upwind half of the rotation and an underestimation in the downwind half. • contrary to the HAWT, the induction is not a function of the total loading, but only of the load component associated with the azimuthally varying circulation. • it is possible to significantly improve Double Multiple Streamtube models by incorporating a better description of the flow. The proposed improved model clearly surpasses conventional models on the prediction of the induction and loading. • the impact of the blade bound-circulation constant-term is small in comparison to the time-varying bound-circulation term; therefore, the induction field of the 2D VAWT in potential flow can be defined by only the number of blades, rotor solidity and tip-speed ratio. This allows for the obsolescence of streamtube momentum models, replaced by faster and more accurate potential flow vortex models. In Chapters 4 and 5 we visualize and quantify, experimentally and numerically, the flow field in the near wake of the blade during the upwind and leeward segments of the motion, at tip-speed ratios λ = 2, 3 and 4, using Particle Image Velocimetry (PIV). An interesting physical aspect of the vortical flow in dynamic stall, especially at low tip-speed ratios, is the transport of the shed vorticity with the blade. This transport of the vortical structures with the blade means that the geometry of the wake, due to viscous effects, differs from what is obtained with potential flow. A different spatial distribution of the shed vorticity implies a different induction field, which might imply a reduction of the effectiveness of momentum models and simple potential-flow models and change the rotor’s performance. The results show two important effects: • At the rotor scale, the transport of vorticity with the blade, rolled in the leading edge/trailing edge separated vortices. • At the blade scale, the importance of the small scale vortices for the oscillations on pressure distribution and loads. In Chapter 6 we use the experimental and numerical data from the previous chapters to evaluate the feasibility of extracting information from the flow/wake measured with PIV, even in dynamic stall, for improving flow analysis and model validation. In Part III we introduce the fourth dimension of our problem: the spanwise direction. The finite span leads to a non-constant spanwise distribution of circulation on the blade, and this distribution leads to the release of trailing vorticity, of which the blade tip vortex is the most prominent component. In Chapter 7 we measure the wake at the tip-vortex region of the VAWT; in Chapter 8 we combine these experimental results with 3D unsteady free-wake potential-flow simulations to: • experimentally and numerically observe, quantify and analyze the generation and convection of the 3D tip vortex of the VAWT. • experimentally, numerically and analytically investigate the effect of blade-tip shape on the generation and convection of the tip vortex, with focus on the added circulation due to the motion of the blade. • combine experimental measurements and numerical simulations to analyze: the 3D wake of the VAWT; the interaction between shed and trailing vorticity; the roll-up and expansion of the wake in the leeward and windward regions; the in-rotor convection and inboard/outboard motion of the tip vortex; the 3D induction field; the 3D blade wake interaction during the downwind blade passage; and the effect of trailing vorticity in the spanwise distribution of circulation, including the 2D to 3D load direction reversal in the downwind blade passage. The spanwise dimension of the flow also gives rise to a new form of misalignment between the flow and the axis: skewed flow. In Chapter 9 we analyze the physics of skewed flow, flow asymmetry, near wake development, blade-wake interaction and impact on energy conversion. The analysis of the VAWT from the point of view of the 2D and 3D near wake is shown to be very effective in understanding: the physics of the flow; the energy exchange process; how the total energy exchange over one rotation actually relates to the local aerodynamic loading on the blade; the impact of implementing an essentially 2D energy conversion process into a 3D aerodynamic system; and the resulting inefficiencies due to the finite span and trailing vorticity. In Part IV (Chapter 10) we show that it is possible to decompose the VAWT design problem into designing for loading and designing for energy conversion, opening a large design space and proposing a new methodology, impacting both 2D and 3D flow. We also show that, although the 2D wake does not vary significantly with variation in the pitching axis location and blade camber, the 3D wake and performance are significantly affected by these variations. This is due to the impact that varying the bound circulation has on the release of trailing vorticity; a larger trailing vorticity generated during the upwind blade passage implies a larger induction due to trailing vorticity, and a worse interaction at the downwind blade passage. The effects of variation of camber and/or pitching axis in 2D and 3D performance are contradictory and complementary and can be simultaneously optimized. In Part V (Chapter 11) we further develop these and other main conclusions, discussing their impact on VAWT aerodynamics. The research here presented implies a break from conventional approaches to the VAWT aerodynamics, allowing for the development of new research and models, both in 2D flow (aerofoil design, rotor energy conversion optimization) and 3D flow (blade and rotor shape, non-uniform flows).