Large-span flexible photovoltaic (PV) arrays exhibit irregular chaotic behavior due to temporal variability and spatial non-uniformity, making the prediction of extreme wind load highly challenging. Current standards and engineering practices also lack effective strategies for wind load mitigation and chaos suppression. To address this, synchronous wind tunnel pressure measurements conducted on large-span flexible PV arrays, with and without four representative flow-altering devices (shaped as , , Γ, and L). The distribution of wind load and chaotic characteristics under 0° and 180° inflow angles were compared and analyzed. Computational fluid dynamics simulations were further employed to reveal the mechanisms of load reduction and chaos suppression. Based on the peak factor, chaos–non-Gaussian joint extreme wind load models are developed for five PV array configurations. Results show that the devices significantly reduce the maximum Lyapunov exponent, sample entropy, and extreme positive and negative pressure coefficients, with average reductions of 47.6%, 21.7%, 16.7%, and 11.1%, respectively. Load reduction is primarily due to vortex separation and reattachment-induced low-pressure zones, while stable large-scale vortices and organized flow structures help suppress chaotic fluctuations. The proposed model improves prediction accuracy by 46.6% and 19.6% compared to conventional peak factor and Hermite polynomial methods. Among all devices, the -type performs best in reducing loads and chaotic intensity, followed by the -type, with the Γ- and L-types being less effective.
Zhao et al. (Fri,) studied this question.