Machine Learning Enhanced Topology Optimisation of Fibre-Reinforced Polymer Composites

Background

Fibre-reinforced polymer composites (FRPC) are increasingly used in high-performance industries such as aerospace, automotive, and renewable energy due to their exceptional strength-to-weight and stiffness-to-weight ratios. However, traditional design methods often fail to fully exploit the potential of these materials. Topology optimisation provides a powerful approach for designing structures with optimal material distribution within a defined space. For FRPC, topology optimisation involves simultaneous optimisation of both material layout and fibre orientation.

However, topology optimisation is computationally demanding, due to the large number of iterations required to reach an optimal design. Each iteration involves finite element analysis to evaluate the structural response along with the calculations of design sensitivities to guide the optimisation process. This process is very time-consuming, particularly when dealing with complex three-dimensional geometries in real-world engineering applications.

This project integrating topology optimisation with machine learning to enhance the computational efficiency. Machine learning algorithms will replace finite element analysis to efficiently calculate design sensitivities, significantly accelerating the optimisation process without compromising accuracy. Additionally, manufacturing constraints will be incorporated directly into the optimisation framework, ensuring that the final designs are ready for direct 3D printing, eliminating the need for manual adjustments and bridging the gap between design and production.

Project Aims and Objectives

This project aims to develop, implement, and experimentally validate a machine learning-enhanced topology optimisation algorithm for FRPC structures. The proposed framework will bridge the gap between design and manufacturing by integrating manufacturing constraints within the optimisation process.

The objectives of the project are as follows:

·    Develop a topology optimisation algorithm capable of optimising both material layout and fibre orientation within FRPC structures to enhance material efficiency and structural performance.

·    Implement the algorithm using two- and three-dimensional finite element methods to ensure accurate and efficient modelling of FRPC structures.

·    Integrate machine learning into the computational framework to adopt a data-driven approach for efficiently calculating design sensitivities, reducing computational demand.

·    Incorporate key manufacturing constraints directly into the topology optimisation process to streamline the design-to-manufacturing workflow, eliminating costly manual post-processing steps and manufacturable designs.

·    Conduct numerical and experimental validation of the optimised two- and three-dimensional FRPC structures to demonstrate their structural performance. 

Requirements:

·      Good MEng or MSc Degree (65% or above) in Engineering or other highly numerate discipline (e.g. Physics, Computer Science, Mathematics)

·      Good understanding of structural mechanics, polymer composites and finite element method

·      Strong programming skills

·      Excellent communication skills (written, spoken)

How to apply

For further details on how to apply, and the link to Durham’s Postgraduate Application portal, see

https://www.durham.ac.uk/study/postgraduate/research-degrees/how-to-apply/

In the first instance (or for informal discussion), contact Dr Zahur Ullah ([email protected]). Application can then be formally submitted through the online postgraduate (research) application system (please highlight that your application should be considered by Dr Zahur Ullah):

https://www.durham.ac.uk/departments/academic/engineering/postgraduate-study/research-degrees/