Density-based topology optimization of a surface cooler in turbulent flow using a continuous adjoint turbulence model

• Published in: Structural and multidisciplinary optimization Vol. 65; no. 8
• Main Authors: Holka, Quentin, Toubiana, Ephraïm, Cortial, Julien, Ghannam, Boutros, Nemer, Maroun
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer Berlin Heidelberg 01.08.2022; Springer Nature B.V; Springer Verlag
• Subjects: Aircraft engines; Circuits; Computational Mathematics and Numerical Analysis; Density; Design optimization; Engineering; Engineering Design; Engineering Sciences; Flow-driven Multiphysics; Fluid dynamics; Heat exchange; Optimization; Pressure drop; Pressure loss; Research Paper; Sensitivity; Spalart-Allmaras turbulence model; Theoretical and Applied Mechanics; Topology optimization; Turbulence models; Turbulent flow; Two dimensional models; Topology optimization; Conjugate heat transfer; Continuous adjoint; Surface coolers; Spalart–Allmaras turbulence model
• ISSN: 1615-147X; 1615-1488
• DOI: 10.1007/s00158-022-03311-z

The present work focuses on the application of density-based topology optimization to the design of a surface cooler. This kind of device is used to cool down the oil circuit in aircraft engines thanks to the cold air in the bypass stream, and is subject to severe heat duty and pressure drop requirements. The optimization is carried out with an in-house implementation of the density method in OpenFOAM. A continuous adjoint strategy is employed to compute the sensitivities with respect to the design variables. Avoiding the so-called “frozen turbulence” assumption, the variations of the turbulent viscosity are taken into account in the computation of the sensitivity. The proposed model also considers the influence of the design variables on the wall distance function occurring in the formulation of the Spalart–Allmaras turbulence model. A simplified two-dimensional model is first employed to tune the optimization and the density model parameters. Then, the methodology is applied to a large-scale three-dimensional case, and the results are compared to a reference straight-fin geometry. The performance is finally evaluated with a reference solver, showing that the density model overestimates both the heat exchange and the total pressure loss, but that the methodology is still able to provide efficient designs in turbulent flow, starting from a very remote initialization.