Degree
Doctor of Philosophy (PhD)
Department
Civil & Environmental Engineering
Document Type
Dissertation
Abstract
The wide expansion of utility-scale agrivoltaic and photovoltaic systems has increased the need for reliable solar tracker designs capable of maintaining structural stability under realistic atmospheric wind conditions. Single-axis solar trackers are particularly susceptible to wind-induced aeroelastic instabilities, including torsional galloping, due to their large aspect ratio, low torsional stiffness, and operation over a wide range of tilt angles. This dissertation investigates the aeroelastic, experimental, computational, and data-driven aspects of wind-induced torsional response in solar tracker systems, with emphasis on instability characterization, mitigation, inflow turbulence generation, and damping prediction. The study first develops a computational framework in OpenFOAM to examine the aeroelastic response of solar tracker sections subjected to atmospheric boundary-layer inflow. The numerical model incorporates dynamic mesh motion, rigid-body dynamics, mesh sensitivity analysis, and damping-effect studies to evaluate the onset and development of torsional galloping under different tilt angles and wind speeds. The results provide insight into the role of aerodynamic damping, stiffness effects, edge modification, and reduced velocity in governing tracker stability. Large-scale wind tunnel experiments are then conducted using a sectional solar tracker model under realistic boundary-layer flow conditions. The experimental setup characterizes velocity profiles, turbulence intensity, acceleration response, displacement response, modal frequencies, effective damping, and flutter derivatives. Several damping-identification approaches, including frequency-domain and time-domain methods, are used to quantify the effective damping behavior. In addition, a passive mitigation strategy based on rounded edge modification is evaluated to reduce wind-induced response and improve aeroelastic stability. A separate computational study is performed to generate and verify inflow turbulence in a three-dimensional empty-domain virtual wind tunnel using the vortex method and large-eddy simulation with the WALE subgrid-scale model. Mean velocity, turbulence homogeneity, and power spectral density are assessed to confirm the quality of the simulated atmospheric boundary layer before its application to solar tracker simulations. Finally, a machine-learning framework is developed to predict effective damping from PSD-derived features. Physics-Informed Neural Networks, Random Forest, and XGBoost models are trained and evaluated using split and non-split datasets. The models incorporate damping-related features such as frequency, bandwidth, wind speed, tilt angle, reduced velocity, and half-power bandwidth estimates. The results demonstrate the potential of combining experimental signal processing, physics-guided constraints, and machine learning to improve damping prediction and support safer wind-resistant solar tracker design.
Date
7-8-2026
Recommended Citation
Zahrawi, Amro A., "WIND-INDUCED INSTABILITIES IN SOLAR TRACKERS: A COMBINED CFD, EXPERIMENTAL, AND MACHINE LEARNING FRAMEWORK FOR PREDICTING TORSIONAL GALLOPING" (2026). LSU Doctoral Dissertations. 7147.
https://repository.lsu.edu/gradschool_dissertations/7147
Committee Chair
Aly, Aly-Mousaad
LSU Acknowledgement
1
LSU Accessibility Acknowledgment
1
Included in
Aerodynamics and Fluid Mechanics Commons, Civil Engineering Commons, Structural Engineering Commons