This study presents an integrated analysis of Savonius wind turbine (SWT) performance, combining Computational Fluid Dynamics (CFD) and structural simulation with experimental wind tunnel testing. The research focuses on the aerodynamic evaluation of conventional versus optimized blade shapes, specifically examining the implementation of Gurney flaps (GFs) to enhance torque characteristics relative to benchmark data [2]. Numerical simulations utilizing the SST k-omega turbulence model and B-spline shape optimization, building upon previous work [1], predicted significant performance gains for the optimized rotor geometry. However, experimental validation on 3D-printed prototypes revealed critical discrepancies attributed to Fluid-Structure Interaction (FSI). To investigate these deviations, a comprehensive vibration analysis was conducted using ADXL335 accelerometers mounted directly on the rotor blades and bearing housings, alongside high-resolution torque measurements. The vibration study employed two methodologies: static impulse testing to identify natural frequencies and dynamic response monitoring during rotation. The static testing revealed that the rotor's structural natural frequencies---specifically the endplates and shaft---begin as low as 10--15 Hz. Numerical modal analysis confirmed the first three modes lie within the 13--20 Hz range. Simultaneously, the aerodynamic forcing frequency (blade-passing frequency) at the turbine's optimal operational speeds (400--800 RPM) falls exactly within this 13--26 Hz band. This frequency overlap led to severe resonance phenomena, where cyclic aerodynamic loads excited the structural modes of the flexible PLA material. The resulting blade deformation and torsional vibrations significantly penalized the power coefficient (C_P), preventing the prototype from achieving the rigid-body performance predicted by CFD. The study concludes that while additive manufacturing enables rapid prototyping of complex optimized shapes, the structural stiffness must be prioritized to avoid dynamic stall and vibration-induced losses often observed in 3D-printed rotors [3].