49th NIA CFD Seminar
Topic: Efficient Techniques for Optimal Active Flow Control
Date: Thursday, July 17, 2014
Time: 11:00am-noon (EST)
Room: NIA, Rm101
Speaker: Nicolas R. Gauger
Speaker Bio: Prof. Dr. Nicolas R. Gauger received his Master in Mathematics (Dipl.-Math.) from University of Hannover in 1998, and his Ph.D. in Applied Mathematics (Dr.rer.nat.) from TU Braunschweig in 2003. From 1998 to 2010 he was Research Scientist in the field of Numerical Methods for Aerospace Science at German Aerospace Center (DLR) in Braunschweig. Furthermore, he was appointed as Assistant Professor (Jun.-Prof.) for Applied Mathematics at the Department of Mathematics of the Humboldt University Berlin from 2005 to 2010. In Addition, he was Member of the DFG Research Center MATHEON (Mathematics for Key Technologies) in Berlin from 2006 to 2010. Since October 2010 he is now Professor for Computational Mathematics at the Department of Mathematics and the Center for Computational Engineering Science (CCES) at RWTH Aachen University. Furthermore, he is Principal Investigator at the Aachen Institute for Advanced Study in Computational Engineering Science (AICES), which is a Graduate School funded by the German Excellence Initiative.
Abstract: For efficient optimal active control of unsteady flows, the use of adjoint approaches is a first essential ingredient. We compare continuous and discrete adjoint approaches in terms of accuracy, efficiency and robustness. For the generation of discrete adjoint solvers, we discuss the use of Automatic Differentiation (AD) and its combination with checkpointing techniques. Furthermore, we discuss so-called one-shot methods. Here, one achieves simultaneously convergence of the primal state equation, the adjoint state equation as well as the design equation. The direction and size of the one-shot optimization steps are determined by a carefully selected design space preconditioner. The one-shot method has proven to be very efficient in optimization with steady partial differential equations (PDEs). Applications of the one-shot method in the field of aerodynamic shape optimization with steady Navier-Stokes equations have shown, that the computational cost for an optimization, measured in runtime as well as iteration counts, is only 2 to 8 times the cost of a single simulation of the governing PDE. We present a framework for applying the one-shot approach also to optimal control problems with unsteady Navier-Stokes equations. Straight forward applications of the one-shot method to unsteady problems have shown, that its efficiency depends on the resolution of the physical time domain. In order to dissolve this dependency, we consider unsteady model problems and investigate an adaptive time scaling approach.
Additional information, including the webcast link, can be found at the NIA CFD Seminar website, which is temporarily located at