Discrete Adjoint Solver in SU2
Drag Sensitivity of the LM1021 supersonic passenger jet.
Area: Shape Optimization, Computational Fluid Dynamics
AD is used to implement an adjoint solver that enables the robust, accurate and efficient computation of gradients for any of the available state solver included in SU2. Capabilities at the moment include consistent gradient computations of steady and unsteady flows in 2D and 3D as well as for laminar and turbulent simulations. Tight integration into the SU2 framework enables shape optimization using any of these solvers. To validate the results, finite differences, the forward mode of AD or the complex step method are available. Future developments will target coupled problems like heat transfer, fluid-structure interactions and aeroacoustic problems.