Simulation of transient characteristics of thunderstorm out flows through straight-line wind tunnel tests

Wind engineering has well-established methods for designing structures to withstand steady flows, such as synoptic winds, while approaches for addressing non-synoptic wind events (e.g., thunderstorm outflows) remain limited. With the increasing frequency of extreme weather events, potentially linked to climate change, disasters caused by non-synoptic winds are becoming more common. Since 2007, the Giovanni Solari Wind Engineering and Structural Dynamics Research Group (GS-WinDyn) at the University of Genoa has conducted extensive research on thunderstorms, leading to a more precise characterization of three key aspects of thunderstorm outflows: transient angle of attack (AoA), a nose-shaped vertical wind profile, and transient wind velocity. This PhD thesis investigates the transient aerodynamics and aeroelasticity of bluff bodies by independently simulating these three critical aspects in straight-line wind tunnel tests. The first part examines the effects of time-varying AoA on a sharp-edged square cylinder, focusing on differences in aerodynamic coefficients and vortex-shedding behavior between steady and transient AoA conditions. The experimental technique appears innovative compared to existing literature. The use of a CWT-based dynamic filter enables precise refinement of time histories. Aerodynamic coefficients in transient conditions are evaluated using ensemble averaging over a sufficient number of repetitions to ensure result convergence. Variations in rotational speed lead to changes in mean drag and lift of up to 10% and 20%, respectively, and up to 50% for lift RMS compared to steady reference values, with a non-negligible influence from the rotation initial angle. A detailed analysis of vortex shedding behavior reveals frequency discontinuities and delays compared to the steady case, depending on the rotation speed. The second part of the PhD thesis explores the aeroelastic behavior of a sharp-edged square cylinder under accelerating flow conditions, analyzing transverse motion responses under both steady and accelerating flows. The case studies are selected to ensure a clear distinction between VIV and galloping by setting the onset of VIV either at the start of the transient or during an intermediate phase. The steady case is also carefully analyzed to provide reference values. The analysis is performed by ensemble averaging the standard deviations of the transient response, considering two types of accelerating flow conditions: Gaussian-type and Constant-type. The magnitude of the dynamic response is highly dependent on the acceleration level, which can induce a significant delay in the occurrence of the maximum value compared to steady reference cases. Furthermore, Constant-type accelerations result in larger structural responses than Gaussian-type accelerations, with their response characteristics more closely resembling those of the steady reference case. The final part of the PhD thesis develops a convolutional neural network (CNN) algorithm to efficiently and accurately simulate thunderstorm-like vertical wind profiles in a multiple-fan wind tunnel. The findings indicate that the CNN-predicted RPM configurations effectively control mean wind speeds and turbulence integral length scales, an encouraging result, as the latter is traditionally challenging to regulate in conventional wind tunnels. Overall, investigating these transient phenomena, whose effects on structures remain largely unexplored in the literature, this thesis contributes to a deeper understanding of the potential structural impacts caused by non-synoptic winds, providing valuable insights for enhancing the resilience of structures to such events.