Adjoint Sensitivity Analysis for Hypersonic Flows
Brian Lockwood, University of Wyoming
The simulation of hypersonic flows is a subject of interest in many engineering fields, particularly aerospace engineering. Practical applications of these simulations include designing orbital reentry vehicles and analyzing novel propulsion systems such as scramjets. In addition to the practical applications, the simulation of hypersonic flows presents a number of computational challenges due to the highly nonlinear character of the flows and the large variations in length and time scales. In order to accurately simulate hypersonic flows, the use of complex chemical and thermodynamics models is required. Included in these models are a number of physical phenomena including molecular dissociation, electron excitation, vibrational energy, and ionization. In order to capture these physical processes, the models have a large number of experimentally and empirically determined parameters. The goal of this work is to determine the effect these model parameters have upon engineering quantities of interest in hypersonic flows, such as surface heating. One way to calculate this effect is by using adjoint sensitivity analysis. This type of sensitivity analysis uses the flow adjoint to calculate sensitivity derivatives for a given objective function. The advantage of using adjoint sensitivity is that the sensitivity of an objective to multiple parameters can be determined with a single solve of the adjoint equation. Within this poster, simulation results using a perfect gas model and a real gas model will be presented. In addition to simulation results, the sensitivity of surface heating to parameters within the perfect gas model will be presented.
Abstract Author(s): Brian Lockwood, Dimitri Mavriplis