Thursday, October 11, 2012
Mechanical and Civil Engineering Seminar
Real-Time Predictions for Decision Using Reduced-Order Models
David Amsallem, Aeronautics and Astronautics, Stanford University
This talk will present the real-time performance on a mobile device of an online, computational methodology for the CFD-based flutter analysis of aircraft in the subsonic, transonic and supersonic regimes. This online methodology is based on (1) an offline linearized model reduction method for predicting accurately flutter for a given flight configuration, (2) a database approach for building a compressed knowledge about sampled flight configurations, and (3) an online method of interpolation on a suitable manifold for generating in real-time an aeroelastic reduced-order model at an un-sampled flight configuration. The talk will cover the physical, mathematical, and computer science theories underlying each of these three components. It will also feature live demonstrations on an mobile device of the real-time flutter analysis of a wing-store configuration parameterized by the Mach number, altitude, and fuel fill level. Finally, the proposed approach is applied to a real-time decision framework and embedded in a model predictive control setting for computing the optimal flight path of an aircraft under flutter constraints.