About the project
Concrete properties depend on interactions between multiscale structures and their interfaces. This project will develop a hierarchical GNN by integrating micromechanics theory. Combining material data, numerical simulations and AI, the research will connect mechanical behaviour across scales and evaluate predictions of concrete mechanical properties.
Concrete may appear uniform, but its strength depends on how cement paste, aggregates, and their interfaces interact. Fine aggregate combines with paste to form mortar; coarse aggregate introduces another level of interaction. Thin regions around aggregate particles influence load transfer and where cracks begin to form. Predicting how these connected scales govern strength remains challenging, particularly when measurements of individual constituents are incomplete.
This project will complete and extend an emerging hierarchical graph neural network for concrete. You will develop linked models of paste, mortar and concrete to investigate how constituent properties and interfaces influence concrete mechanical properties. The central research challenge is to integrate micromechanics: the theory linking the behaviour of a composite to its constituents. You will investigate homogenisation to estimate overall stiffness from constituent properties, and model stress transfer and interface damage during loading.
These relationships will guide how the network combines information between scales. Experimental data and numerical simulations will support development and validation, including checks on intermediate material properties where measurements are available. Comparisons with conventional machine learning and mechanics models will assess whether the combined approach improves prediction and physical consistency. The model will help integrating civil engineering, material science and AI technologies.
You will:
- develop expertise in machine learning and computational mechanics through a concrete materials problem
- learn to connect material observations, physical theory and numerical models, and critically assess AI predictions
- build transferable skills for research careers in materials modelling, engineering software and AI-assisted design
You'll receive training in Python, machine learning and computational mechanics. Technical training will cover micromechanical models, numerical simulation and the transfer of material properties between scales. You'll learn how to combine experimental and simulated data, test the accuracy of model predictions, and assess whether they remain consistent with physical behaviour. Professional development will support scientific writing, research presentations and reproducible software development. You'll gain experience communicating across materials science, mechanics and AI, and explaining the strengths and limitations of the models you develop.
The School of Engineering is committed to promoting equality, diversity inclusivity as demonstrated by our Athena SWAN award. We welcome all applicants regardless of their gender, ethnicity, disability, sexual orientation or age, and will give full consideration to applicants seeking flexible working patterns and those who have taken a career break. The University has a generous maternity policy, onsite childcare facilities, and offers a range of benefits to help ensure employees’ well-being and work-life balance. The University of Southampton is committed to sustainability and has been awarded the Platinum EcoAward.
