Inverse Design of an Intermediate Compressor Duct

The report at hand investigates the concept of a partially physics driven inversedesign method. Starting point is a literature review as an introduction into thecore concepts in fluid mechanics and optimization theory used to develop thetools used in the inverse design algorithm. The algorithm itself consists of twomain parts: A metaheuristic algorithm which is aimed to generate a targetpressure distribution (TPD), as well as a Modified Garabedian-McFadden(MGM) algorithm to create the respective geometry, that resembles the targetpressure distribution. While this two–step process should be applicable to mostinternal flow cases, this study aims to establish and validate it in the context ofaggressive inter compressor ducts. Furthermore, the influence of turbulencemodels in the generation of the wall pressure distributions is investigated. Inparticular, the Standard k-ω and Standard k-ϵ model’s behavior is beingcompared, when facing different levels of aggressiveness. To further developthe established concept, a more advanced process in the TPD generation isexamined: Including an analysis of the impact of the viscous forces and thenature of the model governing equation, qualitatively commenting on its scope.With the current implementation, the in-house inverse solver requiresapproximately 100 iterations of a 3D k-ε solver with standard wall functions—each taking 25 to 30 seconds—to reach a pressure residual level under oraround 1%. With minor modifications to the model’s governing equations thecurrent 2D inverse design algorithm (IDA) framework shows potential forextension to three-dimensional problems.