*1.1. Impact of Turbulence in New Wind Energy Applications*

The installation of small wind turbines in built environments is being studied by the scientific community to complement wind energy conversion in large wind farms with a distributed generation. Smaller rotors positioned at the top of tall buildings could theoretically exploit a higher zone of the wind profile (more energetic) without the need for tall towers. The delocalized generation would also allow the production of energy where it is needed, saving transportation costs and contributing to the sustainable design of new buildings [1]. The real feasibility of this scenario, however, still needs to be proved in terms of energy conversion efficiency and social acceptance [2].

From a more technical point of view, the main challenge is the complexity of urban flows. The terrain presents high roughness, with the displacement height often at the level of the building

heights themselves [3]. In addition, the flow reaching the rotors is often modified by the interaction with multiple obstacles of different shapes and permeability, e.g., upstream buildings or street furniture. All these effects result in mean wind speeds significantly lower than those available in the countryside areas [4], and poorer flow quality in terms of skew angles [4], fluctuations and, in the end, of high values of turbulence [5], which is one of the preeminent characteristics of urban flows. Turbulence in these contexts is characterized not only by high intensities, but also by large length scales [6]. This kind of turbulence is then quite different with that usually considered both in the experimental testing of airfoils or turbine prototypes and in the design phase using either engineering models or computational fluid dynamics (CFD) [7]. Experimental evidence suggests, however, that it can play a relevant role in the effective turbine operation in terms of increase of fatigue, unpredictability of energy production, or influence on stall conditions [8].

#### *1.2. Evidence of VAWT Behaviour in Turbulent Flows*

Vertical-axis wind turbines (VAWTs), especially with the Darrieus configuration, are often suggested as a possible alternative to conventional horizontal-axis wind turbines (HAWTs) for power production in small-sized applications, and especially in the aforementioned peculiar conditions found in the urban environment [2]. Despite the lower nominal efficiency with respect to HAWTs, this research trend is connected to the inherent advantages of the Darrieus concept: the omnidirectionality with respect to the wind direction (so that no yaw system is needed [9]), the possibility of putting the generator on the ground, the low noise emissions [10], and the good performance in case of misaligned (skewed) flows thanks to an increased virtual swept area [11]. In addition, Darrieus VAWTs are often preferred to other turbine layouts in view of an integration with the landscape of populated areas, since they are commonly perceived as aesthetically more pleasant by people [12]. Up to now, the study of turbulence effects on Darrieus VAWTs has been based on little field data correlating the performance of installed turbines with the on-site wind turbulence measured by a meteorological station. These studies often draw unclear and contradictory conclusions, since the effects of turbulence are said to be positive [13,14], negative [15,16], or velocity-dependent [17]. The difficulty of properly acquiring experimentally the inflow condition in a real urban environment suggests that detailed laboratory measurements are needed in order to reproduce in an accurate and repeatable way the turbine power curves and wake characteristics in turbulence. Nevertheless, replicating the urban flow characteristics inside a wind tunnel is not an easy task. In fact, most of existing experimental facilities are designed for aeronautical purposes and, thus, have a very low background turbulence intensity (*Iu* < 1%). This value is far from the turbulence intensity typically found within a built environment, which is typically higher than 10% [5]. In addition, the characteristic integral length scales (*Lux*, indication of the size of the most energetic eddies) are quite large, with values of *Lux* in the order of 1 m [6]. As one may argue, these values are barely replicable in wind tunnels of limited dimensions: large ones are therefore needed. This latter characteristic is also motivated by the need of limiting the blockage effect inside the tunnel, which could be responsible of creating an interference that may deform the expected wind conditions during turbine operation [18].

As a result, only few studies on the influence of turbulence on VAWT performance in wind tunnels have been carried out so far; in all cases, physical grids were used to increase turbulence in the test section. Regarding power generation, [19] measured a slight increase of the power coefficient for a five-bladed Darrieus turbine, while [20] performed a full set of experiments regarding the interaction between turbulent flows and an H-Darrieus, and despite detecting a very large increase of the power produced, they could not retrieve the complete power curves. The influence of turbulence in the wake was studied by [21] for a five-bladed VAWT, recording better self-starting capabilities and a lower speed deficit in the far-wake; [22] also recorded faster wake recovery for a three3-bladed VAWT. The combination of the two features (turbine performance and wake recovery) was measured by [23,24] for a two-bladed Darrieus, also concluding that turbulence enhanced turbine production and wake recovery. The impact of turbulence was also studied in the past at a structural level [25], to understand

the dynamic response of the rotors to gusts. Numerical studies were also presented, using different levels of fidelity ranging from engineering models [26] to U-RANS CFD [27]. However, very often the studies seem to lack extensive validation or a deep understanding of the involved physics.
