Accurate Modelling of Conjugate Heat Transfer for Density-Based Topology Optimisation

Abstract

In the high-tech engineering sector, industry is always looking for the competitive, technological edge. Through optimisation, in particular the optimisation of component designs, performance gains can often be realised compared to conventional design geometries. Even within existing implementations, this so-called topology optimisation allows for improvements in component performance, simply by exchanging its existing conventional design, and ultimately to higher-level machine assemblies and modules. This low barrier to entry makes topology optimisation a field of engineering that has been gaining traction over the past decades within various fields of high-tech engineering and, of course, research. While much research has been done into the field of topology optimisation, including research into multi-physics optimisation problems, such as conjugate heat transfer problems, more work is still to be done to improve models, improve optimisation schemes and approaches, reduce computational time, increasing results accuracy and much more. This thesis, in particular, is focussed on devising a strategy to derive and implement a thermofluidic model specifically for density-based topology optimisation applications. In doing so, emphasis is placed on the accuracy of the optimisation model when compared to numerical results found in regular thermofluidic analyses of systems. In this thesis, a new technique is introduced to refine existing fluidic topology optimisation models for density-based methods: dubbed IGPP, Implicit Gradient-Parallel Penalisation for fluid velocity fields is devised, implemented and evaluated numerically. Furthermore, a parameterisation is created for the material properties relevant to conjugate heat transfer problems. Finally, the model’s performance is evaluated.