Aerodynamic Shape Optimization of a Gerlach Diffuser (based on Adjoint Method)

Shape optimization has always been an engineers' target in order to improve existing performances of specific components. This process can often be time-consuming, requiring substantial manual inputs and multiple design iterations; furthermore it is able to involve at most about ten design variables, limiting the explorable design space. Adjoint-based optimization, instead, has been acknowledged to be effective for applications characterized by a large number of variables. This kind of shape deformation follows an underlying principle, the Adjoint Method, a specialized mathematical tool that expands the range of a CFD solution by providing detailed sensitivity data, i.e. derivatives of the objective function with respect to the design variables. This feature makes the computational effort independent from the number of these variables. Consequently, this method can be useful to lead smart design shape optimization of any geometric part in the computational domain, deriving the optimal shape from a baseline CFD flow calculation. In order to analyze this method and its features, a baseline computational analysis on a Gerlach diffuser was carried out, using Ansys Fluent software and its Adjoint solver. This particular duct was conceived by Gerlach, indeed, with the aim of decrease total pressure lost along the diffuser. In order to obtain this result, he derived a correction factor to modify the cross-sections, useful to reduce the phenomenon of secondary flow, which generates lateral pressure gradient, without decreasing the velocity of the main flow.