Shape optimization in adaptive search spaces [electronic resource]
- George R. Anderson.
- Physical description
- 1 online resource.
Also available at
At the library
All items must be viewed on site
Request items at least 2 days before you visit to allow retrieval from off-site storage. You can request at most 5 items per day.
|3781 2015 A||In-library use|
- Anderson, George Robert.
- Jameson, Antony, 1934- primary advisor.
- Aftosmis, Michael J., advisor.
- Alonso, Juan José, 1968- advisor.
- Stanford University. Department of Aeronautics and Astronautics.
- Design space dimensionality is a persistent obstacle to high-fidelity aerodynamic shape optimization. The burden typically falls to the designer to select the shape design variables, which restricts the design space and impacts computational efficiency. In this work, I develop a system that automatically adapts the shape parameterization to efficiently solve the given design problem. Optimization begins with a small set of design variables, which is then progressively refined to enable the discovery of superior designs. The adaptation is goal-oriented and is driven by a novel refinement indicator that focuses shape control resolution on regions with the highest potential to improve the design. This indicator is computed from adjoint sensitivity information and a quasi-Newton Hessian approximation, which are both readily available in many design frameworks. To verify that the system can autonomously discover parameters sufficient to solve a problem, three inverse design problems are examined. For these cases, the procedure robustly converges to the continuous optimum from different starting points and with different refinement strategies. Several airfoil and wing shape optimization problems are then performed. The adaptive approach generally yields smoother design trajectories, at substantially lower cost than under fixed or uniformly-refined parameterizations. Finally, I consider the combinatorial problem of finding an optimal layout for an adaptive, trailing edge flap system. In fewer than 100 simulations, the adaptive procedure discovers a flap layout that outperforms naive layouts. I conclude with a discussion of the tremendous potential this approach offers for many other disciplines and design environments.
- Publication date
- Submitted to the Department of Aeronautics and Astronautics.
- Thesis (Ph.D.)--Stanford University, 2015.
Browse related items
Start at call number: