Multidisciplinary Optimisation of Additive Manufactured Heat Exchangers for Aeronautical Applications

In this work, we investigate a new methodology for the multi-scale simulation and optimisation of Micro Channel Heat Exchangers with lattice-like topology based on homogenization theory. The lattice matrix is modelled as an equivalent porous medium and is simulated only at the macroscale using standard CFD tools. Micro and meso-scale effects are incorporated in terms of variable permeability tensor and heat transfer coefficients, which are assumed to be non-linear functions of the local flow conditions and lattice geometry. Scale bridging is achieved thanks to a Machine Learning (ML) model which is trained to infer the required closure relationships from a database of Direct Numerical Simulations (DNS). DNS are performed at the scale of a single lattice cell for several geometries at varying operating conditions. The proposed approach is used to solve a benchmark problem provided by Rolls-Royce plc. The test case involves the optimization of a heat exchanger employed in the cooling system of power electronics. In order to assess the soundness of the proposed methodology, numerical results for three different configurations are presented. The effect of under(over) fitting is also investigated in terms of quality and performances of the final optimal geometry.