Check out my git repo for the optimization algorithm: https://github.com/azav123/mae4272-wind-turbine-blade-optimization
In MAE 4272, our team designed and optimized a small wind turbine blade for low-wind conditions, aiming to maximize power output under a Weibull-distributed wind profile with a mean speed of 4.59 m/s. The objective was to achieve a target operating point of 1739 RPM, 3.31 N·cm torque per blade (scaled for three blades), and 6.02 W power, while adhering to structural constraints like a maximum stress of 36.67 MPa. We simplified aerodynamics to steady 2D flow at Re ≈ 50,000, ignoring minor 3D effects and friction for efficient evaluation, and selected the NACA 4412 airfoil for its superior lift-to-drag ratio at low Reynolds numbers compared to alternatives like S1223 and S822.

Figure 1: Assembled Wind Turbine

Figure 2: Rendered CAD model of the optimized blade.

Figure 3: Experimental power curves at tested wind speeds.
The design process featured a parallel workflow: a Python-based differential evolution optimizer refined parameters such as 10 pitch angles, root chord, and RPM, discretizing the blade into 10 segments for aerodynamic and structural analysis using blade element theory and beam bending. Concurrently, CAD modeling in Fusion 360 parameterized geometry for manufacturability, incorporating adjustments like truncating the airfoil trailing edge to strengthen the hub connection without major aerodynamic loss. Key iterations included post-optimization smoothing of pitch angles for monotonic distributions and penalty functions to enforce constraints, linking numerical results directly to a 3D-printable model.
Testing mounted the 3D-printed blades on a turbine in a wind tunnel, evaluating performance at speeds of 9.88 m/s, 11.65 m/s, and 14 m/s, as the design failed to rotate at the targeted lower speeds due to incorrect pitch orientation from a miscommunication. We analyzed power, RPM, and torque data across brake settings until stall, yielding peak powers from 0.091 W to 0.343 W, only 6% of predictions, but confirming structural integrity with a safety factor over 58. This highlighted discrepancies between modeled and experimental results, informing improvements like refining penalties and incorporating real friction data.
My specific contributions included developing the engineering model used for the optimizer and the differential evolution optimizer to iteratively maximize expected power while respecting constraints. I also contributed to wind tunnel testing, data collection, and analysis, where I identified opportunities for enhancement, such as increasing population size for better exploration, tuning hyperparameters to reduce variability, and integrating experimental Cl/Cd data for more accurate modeling.
Sections above included the main points of the paper. For insight into the design process, optimization methods, and further details, you can download the full report.
We use cookies to analyze website traffic and optimize your website experience. By accepting our use of cookies, your data will be aggregated with all other user data.