Civil Engineering: Fully Funded PhD studentship in AI-Assisted Mesh Generation and Adaptation f[...]

In many engineering simulations, the accuracy and efficiency of the solution depend critically on how the mesh is distributed relative to the underlying physics. Features such as boundary layers, shocks, vortices, thermal gradients and structural stresses often occur in regions that are closely linked to the geometry of the problem. In current industrial workflows, these phenomena are commonly captured either by globally over refining the mesh, which is computationally expensive and environmentally inefficient, or by running multiple successive simulations to iteratively adjust the mesh. Both approaches raise computational cost, energy consumption and turnaround time, placing increasing pressure on sustainability targets.

Understanding how geometric changes influence the flow, thermal or structural response remains a major challenge, and traditional mesh spacing strategies struggle to capture the complex, nonlinear ways that geometry shapes multiphysics behaviour, leading to either unnecessary refinement or a loss of fidelity in critical regions. Machine learning provides a promising route to capture these relationships more systematically by identifying how local geometric features determine the resolution required for reliable prediction.

A central goal of this project is to learn how solution fields and mesh resolution requirements vary with geometric change, enabling sensitivity informed meshes that adapt to both physical behaviour and geometric context. The project will develop a machine learning framework that learns the link between geometric variation, coupled physical responses, solution sensitivity and mesh spacing requirements. The resulting tools will support automated mesh generation and adaptation, reduce manual tuning and improve the reliability of simulations involving geometry driven behaviour across multiple physical models.

As the PhD researcher on this project, you will work at the intersection of machine learning, geometry processing and industrial simulation. You will have the opportunity to explore realistic engineering configurations and gain expertise in areas that are rapidly growing yet still rare across the simulation community, including AI-assisted mesh generation and adaptation for industrial simulation. The skills developed through this work will align strongly with careers in scientific computing, engineering simulation and applied machine learning for design and analysis.

English Language: IELTS 6.5 Overall (5.5+ each comp.) or Swansea University recognised equivalent.

Eligibility criteria: UK & International students EPSRC DLA studentships are available to home and international students. Up to 30% of our cohort can comprise international students. Once the limit has been reached, we are unable to make offers to international students. We are still accepting applications from international applicants. International students will not be charged the fee difference between the UK and international rates. Applicants should satisfy the UKRI eligibility requirements. PhD: Applicants for PhD must hold an undergraduate degree at 2.1 level (or non-UK equivalent as defined by Swansea University) in Engineering or similar relevant science discipline.

This scholarship covers the full cost of tuition fees and an annual stipend at UKRI rate (currently £20,780 for 2025/26).

Selection process: Please see our website for more information.