Data-Efficient and Transferable Machine Learning-Based Predictive Models for Catastrophic Risk Assessment in Offshore Wind Infrastructure

About the Project

Offshore wind infrastructure underpins the UK’s Net Zero transition but faces extreme operational challenges. Wind turbines must withstand harsh marine environments where multi-hazard loading from wind, waves, currents, and seismic activities interact with corrosive conditions, accelerating degradation and elevating catastrophic failure risks. Current assessment methods rely heavily on sparse field measurements and computationally intensive simulations, limiting their scalability and responsiveness for risk-informed decision-making across large offshore wind farms.

This project will develop next-generation machine learning (ML) models that are both data-efficient and transferable, enabling more reliable catastrophic risk prediction, defined as the probability of exceeding critical safety thresholds with severe consequence at turbine, farm, and portfolio scales. Three pillars drive the approach: (i) identifying and prioritising the most informative simulations and inspections via intelligent data curation to minimise data demands; (ii) creating ML models with task transfer capabilities that can be adapted and reused across different soil types, geographies, and hazard conditions; and (iii) exploiting unlabelled operational data and self-supervised representation learning strategies to reduce reliance on costly measurements and manual labelling. Multi-fidelity modelling will fuse low- and high-fidelity analyses with observational data to yield robust, uncertainty-aware predictions.

Outcomes include a transparent, open-source toolkit for catastrophic risk and fragility assessment, integration pathways with industrial digital risk workflows used by insurers and asset managers, and validated case studies on representative offshore sites. By reducing downtime, operational expenditure, and uncertainty in financing and insurance, this research will enhance the resilience of offshore wind farms and secure the UK’s leadership in trustworthy AI for renewable infrastructure.

This project is co-funded by Renew Risk Ltd. (https://www.renew-risk.com/), offering opportunities to work with real offshore wind farm models and industrial datasets while addressing real-world challenges in collaboration with industry experts.

Supervisors: Dr Tanmoy Chatterjee and Professor Suby Bhattacharya

Entry requirements

Open to candidates who pay UK/home rate fees. See UKCISA for further information. Starting in October 2026. Later start dates may be possible, please contact Dr Tanmoy Chatterjee once deadline passes.

You will have to meet the minimum entry requirements for our PhD programme.

The successful candidate is expected to be highly motivated and must hold a minimum of a 2:1 Bachelor’s level degree (or equivalent) in AI/ML for Engineering, Structural Engineering, Risk Assessment or a closely related field.

Proficiency in programming languages like Python and MATLAB is essential.

Practical experience in data science, predictive analytics, finite-element modelling, structural analysis, and/or offshore structures is desirable.

The candidate should demonstrate strong analytical and problem-solving abilities, along with excellent written and verbal communication skills. The ability to work independently in research and adapt quickly to new methods and technologies will be highly valued.

How to apply

Applications should be submitted via the Civil and Environmental Engineering PhD programme page. In place of a research proposal you should upload a document stating the title of the project that you wish to apply for and the name of the relevant supervisor.

Funding Notes

UKRI standard stipend: £65,191 for the term of the Project.

Tuition Fees covered: £15,938.50 for the term of the Project.

Research Training Support Grant (RTSG) of £7,500.00 is available for the term of the Project.