An Optimization Framework Based on Domain Decomposition and Model Reduction
Abstract
This collaborative research has developed rigorous mathematical and computational frameworks for reduced order model (ROM) generation and the use of ROMs in real-time, design, control, and probabilistic applications of relevance to the Air Force. This research has provided theoretical analyses and numerical studies for several new/extensions of existing ROM approaches, such as goal-oriented, model-constrained approach, balanced truncation model reduction (BTMR) of descriptor systems, and the integration of domain decomposition and BTMR for systems with localized nonlinearities. Additionally, several important questions related to the design, analysis, efficient computation, and application of ROM were studied. The use of ROMs was demonstrated on example applications, including optimal flow control of linearized Navier-Stokes equations, linearized flow control of a supersonic diffuser, subsonic compressor blade row unsteady aerodynamics and geometric mistuning, a thermal design problem, and nonlinear combustor model dynamics.
Document Details
- Document Type
- Technical Report
- Publication Date
- Feb 27, 2009
- Accession Number
- ADA499216
Entities
People
- Danny C. Sorensen
- Matthias Heinkenschloss
Organizations
- Rice University