Continuous Adjoint-Based Aerothermal and Aeroacoustic Optimization of Aero Engine Components Using PUMA

This article addresses the application of the in-house GPU-enabled CFD analysis and shape optimization tool PUMA in Aeroacoustic and Conjugate Heat Transfer problems. The gradients of the objective functions are obtained using the continuous adjoint method that, also, includes the differentiation of the Spalart-Allmaras turbulence model and the Hamilton-Jacobi equation for distance. The aeroacoustic optimization concern the design of the shape of a 3D aero-engine intake generatrix by minimizing the energy contained in the spectrum of the sound pressure, at prescribed receiver locations; the fan manifests its presence through a separately computed time-dependent pressure distribution at the fan inlet and the noise propagation is based on the Ffowcs Williams and Hawkings (FW-H) acoustic analogy. With regard to the aerothermal application, the optimization of the blade shape of an internally cooled linear turbine cascade and the position of cooling channels aims at min. total pressure drop of the hot gas flow and min. highest temperature within the blade. A free-form parameterization tool, based on volumetric NURBS, controls the blade airfoil contour while the cooling channels can be relocated inside the blade, by changing the coordinates of their centers. Geometric and flow constraints are imposed to keep the cooling channels away from the blade contour and retaining the turbine inlet capacity and flow turning.