Session: Government Agency Student Posters

Paper Number: 174034

Surrogate-Based Optimization of Wind Turbine Airfoil Performance Using Xfoil and Openmdao 

As global demand rises for clean, reliable energy, wind turbines and distributed renewable energy systems are playing an increasingly central role in the transition to sustainable power generation. Improving the aerodynamic efficiency of wind turbine blades is crucial to maximizing energy capture, minimizing material usage, and reducing the levelized cost of energy (LCOE). One promising strategy in wind turbine blade design is to optimize the lift-to-drag (L/D) ratio of airfoils, which directly affects aerodynamic performance and, consequently, the power output of a turbine.

To address this need, our study leverages openMDAO. a flexible, open-source framework for multidisciplinary design analysis and optimization, to develop a scalable, surrogate-based aerodynamic optimization workflow. The goal is to identify airfoil configurations that maximize L/D ratios under given operational constraints. We integrate XFOIL, a widely-used tool for subsonic airfoil performance analysis, with openMDAO to evaluate and optimize key design variables: maximum camber, camber position, maximum thickness, Reynolds number, and angle of attack.

A surrogate model is constructed from XFOIL-generated aerodynamic data across a broad parametric space. This model enables rapid optimization using openMDAO’s built-in drivers. In our implementation, the SciPyOptimizeDriver was selected due to its ease of use and compatibility with smooth, unconstrained optimization problems, making it an ideal choice for the initial single-objective problem. The process flow includes data collection and processing, surrogate model creation, iterative optimization, and convergence checks on the L/D ratio.

To ensure surrogate fidelity, the design space was densely sampled with over 150 simulations spanning the relevant parameter bounds, and regression accuracy was verified using holdout error analysis. Interpolation performance of the surrogate was evaluated using k-fold cross-validation to quantify model generalizability. Optimization convergence was verified by comparing the final solution with direct XFOIL evaluations.

The optimization yielded an airfoil geometry resembling a NACA 2411 profile, with optimal values of 4.0% maximum camber, 12.4% maximum thickness, camber location at 40% chord, and an angle of attack of 5.1 degrees. This configuration achieved a maximum L/D ratio of 81 based on surrogate predictions and was confirmed with direct simulation, showing less than 2% deviation, highlighting the accuracy and robustness of the model.

This work illustrates the viability of combining low-fidelity aerodynamic tools like XFOIL with robust optimization environments such as openMDAO for efficient design space exploration. The workflow is modular and extensible, offering potential for future work involving three-dimensional blade analysis, structural coupling, and multi-objective optimization, including considerations such as fatigue, noise, and manufacturability. The modular nature of the framework also supports integration with higher-fidelity solvers or experimental datasets, creating opportunities for hybrid optimization strategies and adaptive model refinement. Ultimately, this approach can contribute to the advancement of wind energy technologies by enabling more efficient and cost-effective turbine blade designs.

 

Keywords

Airfoil Optimization, Wind Turbine Blade Design, Lift-to-Drag Ratio, XFOIL, openMDAO, Surrogate Modeling, Aerodynamics, Renewable Energy, Parametric Design, Multidisciplinary Optimization.

Presenting Author: Peter Marinelli Kent State University

Presenting Author Biography: Peter is an undergraduate research assistant in the NSF REU in the M3TFluiD Lab at Cleveland State University.

Authors:

Navid Goudarzi Cleveland State University
Peter Marinelli Kent State University