Skip to main content

Aerodynamic Shape Optimization by Physics-Based Surrogates

Final Report Summary - ASOPBS (Aerodynamic Shape Optimization by Physics-Based Surrogates)

Aerodynamic and hydrodynamic design and optimization is of primary importance in several disciplines, such as in aircraft design and turbine design. For example, the wing shape is designed and optimized to provide maximum efficiency under a variety of takeoff, cruise, maneuver, loiter, and landing conditions; and the shapes of the propeller blades of ships are designed and optimized to increase efficiency. The fundamental design problem, common to all these disciplines, is to design a streamlined wing (or blade) shape that provides the desired performance for a given set of operating conditions, while—at the same time—fulfilling one or multiple design constraints.

The use of optimization methods in the aerodynamic design process, as a design support tool or for automated design, has now become commonplace. In the conceptual design phase, computationally cheap low-fidelity tools are typically used as a number of different design concepts need to be investigated. As the design progresses and the design space is being reduced (in the preliminary and detailed design phases), computationally expensive and accurate higher-fidelity tools are employed. The use of high-fidelity methods, coupled with optimization techniques, has led to improved design efficiency. However, one of the biggest challenges is the computational cost. Furthermore, conventional optimization techniques usually require a large number of simulations. Moreover, a large number of design variables are needed to describe aerodynamic surfaces. Therefore, efficient and reliable computational methods, both for the fluid flow analysis and the optimization, are essential for a rapid design process. The overall objective of the ASOBPS project was to develop efficient and robust aerodynamic shape optimization techniques. To fulfill the objective of ASOBPS, we have developed computationally efficient and robust procedures and algorithms for aerodynamic shape optimization using high-fidelity computational fluid dynamics (CFD) simulators.

The developed methodologies are handling both two-dimensional and three dimensional aerodynamic surfaces. Moreover, the surrogate-based optimization (SBO) technique was adopted that combines the speed of the low-fidelity models with the accuracy of the CPU-intensive high-fidelity ones. A special emphasis was given on finding ways to fully employ the embedded knowledge within the physics-based surrogates to achieve a low computational cost of the overall design optimization. In particular, this was involving the development of new methodologies for generating computationally cheap and robust low-fidelity models, and equally important, reliable and accurate response correction techniques for aerodynamic responses. The design procedures have been implemented in a computational framework by insuring that the data streams between the optimization algorithms and the low-and high-fidelity models are seamless. In addition the aerodynamic design procedures as well as their software implementation were subjected to extensive numerical verification using well known benchmark problems involving 2D and 3D cases for subsonic and transonic conditions. The performance of our optimization algorithms were compared to state-of-the-art methodologies described in the literature. Furthermore, the project has produced research results and subsequently published them in international conferences and journals.

As the result of this project, we developed a software called EOMC-CFD (Engineering Optimization & Modeling Center – Computational Fluid Dynamics) that deals with the design and optimization of aerodynamic and hydrodynamic streamlined shapes. With an informal collaboration with Mechanical Engineering Department of Politecnico di Milano, EOMC-CFD extended to other applications such as Finite Element Method. Hence we are developing a software called EOMC-FEM that deals with the design and optimization of mechanical and structural elements. We have also started collaboration with the prosthetic company, Össur. After the researcher did his outreach activity, the company is interested in having collaboration with our research group since the methodologies are very applicable to the company’s prosthetic products.