The adjoint method is the key tool of the Competence Center for Fluid Mechanics and Numerical Methods (CC FNUM) for geometry optimization of hydraulic and aerodynamic devices. Increasing the efficiency or reducing the drag forces are typical objectives, but there is no limit with regard to the objective type, as any objective function can be optimized. The adjoint solver is an extension of the in house developed CFD code. Based on a converged CFD simulation, an adjoint equation is resolved in order to evaluate the sensitivity of the objective function with respect to the geometric parameters. The optimization procedure is fully automated. An algorithm iteratively finds better and better geometries based on the principle of gradient descent until it reaches an optimum. Several automatic parametrization methods allow for a rapid optimization with very little user interaction.
The adjoint method has a wide range of application possibilities. The augmentation of turbulence models is an example. In this case, an existing turbulence model is enhanced with additional parameters which in turn are optimized with the help of the adjoint method. The goal is to align the outcome of a turbulence model with experimental data or simulation results of more complex turbulence models.