Improved Accuracy and Uncertainty Management for High-Speed and Unsteady Flows using Optimization Techniques

摘要

The goal of this project was to demonstrate the potential of adjoint methods for improving the accuracy of current simulation capabilities and to enable more effective uncertainty quantification techniques. A modular adjoint approach was developed in this work and has been applied to the following problems: optimized mesh deformation techniques, hypersonic discretization error reduction through h-p adaptive techniques, real-gas hypersonic simulations, unsteady transonic flow simulations, and unsteady fully coupled aeroelastic problems. In addition to formulating and solving the adjoint problem for all these cases, this work has demonstrated the potential of adjoint-produced sensitivities for efficiently solving time-dependent optimization problems, including fully coupled aeroelastic problems, as well as the potential for adjoint methods for providing error estimates which can be used to drive adaptive meshing or time-stepping procedures and/or for performing uncertainty quantification.