Master thesis and internship[BR]- Master's thesis : Evaluation of an OpenFOAM-based software for thermal topology optimisation using the continuous adjoint approach[BR]- Integration Internship

Abstract

This thesis deals with density-based topology optimisation applied in the scope of a conjugate heat transfer problem. After an explanation of the methods of topology optimisation, the latter is specified to fluid based problems, and especially to conjugate heat transfer applications for which the density-based method is used. The goal of the optimisation is to maximise the heat transferred to the coolant fluid while limiting the pressure drops as much as possible. A single type of design variable stands for the representation of the material distribution. Through this variable, the presence of solid will be taken into account thanks to a Brinkmann type penalisation term that is included in the flow equations. This term blocks the flow where there should be solid material. The temperature is modelled through the convection-diffusion equation which describes both conduction in the solid and convection in the fluid. As the gradient-based approach is used to perform the optimisation, the design variable can take intermediate values which leads to an unclear topology. To cope with this, a processing of the design field known under the name of the Three Field Topology Optimisation Scheme is used. The analysis of the density-based topology optimisation is conducted on a simple conjugate heat transfer problem using the adjointOptimisationFoam solver of OpenFOAM. The optimisation conducted topologically proves its efficiency by increasing up to almost 200% the heat transferred to the coolant fluid, and a drastically reduced pressure drop compared to the initial configuration of the domain. The method is then applied on the optimisation of a 3D heat exchanger subjected to a highly turbulent flow. Although the design generated by the optimisation is not converged, the method gave interesting topologies for the first optimisation cycles.