Full-scale experimental investigation of wind loading on ballasted photovoltaic arrays mounted on flat roofs

Ballasted photovoltaic (PV) systems, in comparison to roof-anchored systems, are gaining notable popularity on commercial flat roofs due to the benefits they provide in evading roof penetration and the associated insulation issues. However, the accurate estimation of the aerodynamic uplift forces and their consequent effects on system responses presents a new design challenge. Moreover, possible dynamic effects, characterized by wind induced vibrations, are not accounted for in the design of PV systems in ASCE 7–22, potentially rendering the code design coefficients unconservative. Additionally, the available literature is based on roof anchored PV systems, while experiments in the literature utilizing ballasted PV systems which have distinct behavior and dynamic properties are very limited. The current study aims for a better evaluation of the behavior of ballasted PV systems and the mitigation efficiency of wind deflectors under simulated extreme wind events. To fill this knowledge gap, a 2 x 2 full-scale ballasted PV array model, equipped with wind deflectors and located on a model flat roofed structure was tested at the Wall of Wind (WOW) Experimental Facility (EF). The experimental campaign consisted of aerodynamic and dynamic tests, which permits pressure measurements on the panels under high Reynolds number flow, realistically influenced by the vibrations of the deflectors, as well as capturing of array's dynamic characteristics. The results show that wind deflectors effectively reduce both net area-averaged and point pressure coefficients, particularly under cornering wind directions. While the top surface pressures remained unchanged with the addition of deflectors, the bottom surface pressures experienced a substantial decrease, and the power spectral densities of pressure fluctuations were significantly reduced. Wind deflectors also proved to be efficient in reducing the correlation of instantaneous aerodynamic pressures occurring at different points, reducing the area-averaged peak pressures on the panels and, consequently, reducing net uplift on the entire array. Moreover, the aerodynamic loads were amplified by up to 30% due to dynamic effects caused by the wind induced vibration of the panels. Finally, from the failure assessment tests, a cascading failure mode was observed where the supports are consequently lifted before the entire system is flipped.