Bath Engineering & Design Studentships: Physics-Informed Grey-Box Modelling for Oscillation Diagnosis in Future GB Power Networks

-      This project is one of a number in competition for a studentship from the Faculty of Engineering and Design at the University of Bath. If successful, this studentship is expected to commence 04 October 2027.

An alternative start date may be possible if agreed with your intended supervisors and the Doctoral College.

To support the UK’s statutory target of net-zero greenhouse-gas emissions by 2050 [1], the GB electricity system is increasingly reliant on inverter-based resources (IBRs), including wind, solar, battery storage and HVDC links. Unlike synchronous generators, IBR dynamics are strongly shaped by fast, software-defined controls. Interactions between these controls and the network can produce unexpected or poorly damped oscillations [2], [3]. Recent GB events highlight the operational challenge: mitigation may require generation curtailment or asset disconnection, while inadequately damped oscillations can threaten equipment and security of supply [4].
Diagnosis is complicated by proprietary controls, vendor-specific black-box models and incomplete network information, which limit conventional white-box analysis. Measurement-based methods offer an alternative [5], but purely data-driven approaches may lose accuracy and robustness under noisy or incomplete measurements and unfamiliar operating conditions [6].
This PhD project will develop a physics-informed grey-box framework for oscillation diagnosis in future low-carbon GB networks. Its objectives are to:
•    Establish oscillation mechanisms and prior knowledge. Generate and analyse different oscillation types using simulations and model-based methods, identifying their mechanisms, dynamic signatures and reliable prior knowledge for learning.
•    Develop grey-box diagnostic methods. Embed network topology and parameters, physical constraints and partial device models in learning, aiming to significantly improve accuracy, convergence speed and robustness while reducing data requirements. Estimate modal properties, distinguish principal contributors from affected equipment and support root-cause diagnosis.
•    Validate performance across scenarios. Compare with white-box and purely data-driven approaches using benchmarks, GB-representative simulations and event measurements where available, including noisy or incomplete data, topology changes and unseen operating conditions.
Expected deliverables include validated diagnostic algorithms, benchmark datasets, documented software tools and a reproducible post-event investigation workflow, providing a foundation for future online monitoring and early warning. These outputs will help system operators identify emerging risks, locate principal contributors and select targeted mitigation. By helping reduce avoidable renewable curtailment and enabling secure low-carbon generation integration, the project will support the UK Government’s net-zero target and the University of Bath’s climate-action research ambitions, aligning with the Faculty of Engineering & Design’s sustainability priorities.