Home | About Me | Research | Publications | Hobbies | Contact
Go to:
<aside> đź’ˇ
Microchannel cooling is often the preferred choice for compact heat sinks. However, widely adopted topology optimisation (TO) techniques, such as density-based and level-set methods, often struggle to generate very thin channel strips without imposing maximum length scale constraints. To address this limitation, multi-scale design methodologies have emerged. We build upon recent advances in de-homogenisation techniques to contribute to the multi-scale design of microchannels for cooling applications. Additionally, imperfections observed in the de-homogenised results serve as benchmarks for future improvements, addressing concerns related to modelling accuracy, manufacturability, and overall performance enhancements.
</aside>
DFF Sub-Project & JSPS PD Sub-Project
<aside> đź’ˇ
Conjugate heat transfer in heat exchangers is at the heart of numerous industrial applications. Topology optimization (TO) is a promising numerical method that allows for the design of high-performance thermo-hydraulic systems from scratch. However, full-scale three-dimensional thermofluidic TO remains largely within the academic sphere and has yet to be easily explored by thermal engineers. To bridge this gap, this paper presents an integrated design workflow tailored for three-dimensional, high-resolution topology optimization of conjugate heat transfer systems, incorporating a mean compliance constraint to ensure structural integrity and load-bearing capability. This is achieved using a dual-mesh approach within the density-based TO framework. We also introduce Tanatloc, a user-friendly graphical user interface developed in JavaScript, which provides versatile functionalities and an interactive experience for thermal engineers. Finally, a 3D printed metal-based prototype is fabricated, and reverse engineering is conducted to reconstruct a CAD model using CT-scan images, paving the way for future experimental investigations.
</aside>
<aside> đź’ˇ
This thesis focuses on topology optimization (TO) of thermal fluid-structure systems, motivated by aeronautic and thermal management industrial applications. Topological design sensitivity of arbitrary cost functionals is derived for a weakly coupled thermal fluid-structure model. A reaction-diffusion equation-based level-set method is then developed for solving generic constrained topology optimization problems, allowing for the design of new holes (or islands) from scratch. This method enables the nucleation of new holes (or islands) during topological evolution. Motivated by the need for real-world applications, two key ingredients are introduced into this workflow. The first is the physics-tailored multigrid preconditioner for distributed finite element analysis. This ensures that the physical computation part of the TO framework can be highly scalable in terms of problem size. The second is the two different unstructured mesh adaptation techniques. More specifically, body-fitted meshes, as one of the surface capturing techniques, allow the disjoint-reunion of a global mesh that involves several (fluid/solid) subdomains. Anisotropic meshes fit high aspect ratio elements (highly stretched elements) along the regions of rapid variation of the solutions like interior or boundary layers. All these ingredients allowed us to solve a variety of two- and three-dimensional multiphysics test cases, from single physics problems in 2D to coupled physics in large-scale 3D settings, including minimal mean compliance, minimal power dissipation, design-dependent and design independent-fluid-structure interaction (FSI), natural/forced convection, lift--drag problems. The final opening chapter sheds light on lattice design. Motivated by the need for porous structure in the design of biodegradable implants, a variational method (PDE-filter) is used to simplify the numerical evaluation of geometric constraints: it enables to computation of “local averaged” characteristic function on an unstructured mesh by solving this PDE without requiring the neighbourhood element spatial information.
</aside>