Exploring Applications of Additive Manufacturing for Flow Control, Heat Transfer and Mass Transfer

点击翻译按钮获取中文摘要

This project falls within EPSRC Fluid dynamics and aerodynamics research area.This project is jointly funded by EPSRC via the Department of Engineering Science and the Oxford-Ashton Memorial scholarship, there are no companies or collaborators involved.Project Summary: Heat exchangers (HX) are a key component in heating, cooling, and power systems and improving their heat transfer performance has a direct impact on the system efficiencies and global energy sustainability. Likewise, catalytic reactors play an irreplaceable role in chemical synthesis, manufacturing and removal of harmful pollutants. The aim of this PhD is to explore how additive manufacturing (AM), combined with machine learning (ML) can make way for novel geometries that provide a step change in efficiency for both heat transfer and mass transfer applications. This project falls within the EPSRC fluid dynamics and aerodynamics research area.The intention is to numerically and experimentally explore novel geometries using AM, that could not otherwise be manufactured using conventional methods. Given that AM is still an emerging field and its application in heat and mass transfer is just being realised, the work aims to uncover the challenges and opportunities for AM based HX and catalytic surfaces. A primary area of interest are a set of mathematically defined surfaces, known as triply periodic minimal surfaces (TPMS), which can only be manufactured via AM techniques and have already shown success when applied to the area of structures and have promising thermal-hydraulic performance in initial studies. The numerical studies would be conducted using CFD, potentially utilising the in-house cluster to perform some high-fidelity simulations. These simulations would be validated using results, taken from an experimental rig designed during the project to test 3D printed geometries. The goal is to get an understanding of capability of the geometries, including pressure range, temperature range, suitable Reynolds numbers and 3D printing parameters.The design versatility afforded by AM can be exploited using machine learning to develop novel, best performing geometries. With recent advancements in the accessibility of ML tools and increases in computational power, there is a greater argument apply ML in all aspects of design. However, little work has been proposed on an AM oriented ML design workflow that incorporates CFD - an area in which this PhD hopes to explore and contribute. This would likely take the form of developing a workflow that combines density based topology optimisation and a genetic multi-objective optimisation algorithm. Density based topology optimisation has been been successfully demonstrated to improve performance in the design of diffuser and pipe bends, lending itself for use in optimising the envelope for the flow. The genetic algorithm would be used to optimise the parameters which define the internal structure of the device and would complement TPMS based structures. CFD would be integrated within the workflow to simulate the geometries and assess relative performance against specified objective functions.The ultimate aim is to combine research on ML and AM to develop an optimisation design tool, to fully define a candidate optimal geometry based on a given design envelope and boundary conditions. This would have a significant impact on the efficiency and design of heating, cooling, chemical and power systems throughout many industries, in particular reducing energy demand, cost of manufacture and allowing greater space for other components in assemblies with tight packaging requirements.