Overview
Modern low-carbonenergy systems such as photovoltaic (PV) arrays and battery energy storagesystems (BESS) generate extensive measurement data (electrical, thermal,imaging and diagnostic).
However, there iscurrently no generic, metrology-grounded AI/ML framework that fuses theseheterogeneous data with physics-based models to create trustworthy,asset-specific digital twins with quantified uncertainty.
This project willdevelop a measurement-science-driven digital twin framework for energy assets,initially demonstrated on PV modules/fields and battery systems using existingNPL datasets. The work will integrate suitable physics-based models (for examplePV performance modelling, electro-thermal and thermofluid dynamics) with deeplearning and multi-fidelity modelling.
Bayesianfusion/inference methods will also be integrated for state estimation,uncertainty quantification, anomaly detection, remaining-life prediction andoperational optimisation.
Research aims andindicative work packages: Develop a generalizable, multisensory digital twinmethodology for PV and battery systems that is metrology-guided anduncertainty-aware.
Research Aims and Work Packages
* Create Bayesian datafusion and uncertainty quantification approaches that deliver traceableconfidence intervals for model outputs to aid decision making.
* Validate the frameworkusing calibrated datasets (including ageing, diagnostic, thermal and electricalperformance measurements).
* Demonstrate assethealth assessment capabilities including anomaly detection and remaining-lifeprediction with quantified uncertainty.
* Align outputs withemerging best practice in digital metrology for energy systems and supportdissemination through stakeholder engagement routes.
Training Environment and Collaboration
NPL will provide themeasurement-science foundation, calibrated datasets, specialist support in datascience and uncertainty, and host the student for an extended placement withfacilities and training.
Mansim will provideindustrial supervision, training and access to commercial CFD/AI platforms andrepresentative industrial case studies, supporting rapid translation ofoutcomes into practice.
Eligibility
Applicants shouldhave, or expect to achieve, at least a 2.1 honours degree or a master's (orinternational equivalent) in a relevant science or engineering relateddiscipline.
Essential: Degree in engineering,physical sciences, computer science, or a closely related discipline (typicallyfirst-class or high 2:1, or equivalent; Master's welcome) Strong programmingskills (for example Python, MATLAB, C/C++) Strength in at leasttwo of: machine/deep learning, numerical modelling, statistics, optimisation,scientific computing Ability to work acrossdisciplines and collaborate with academic and industrial teams.
Desirable: Experience in Bayesianinference, probabilistic modelling, or uncertainty quantification. Experience in deeplearning for time-series, imagery, and/or multimodal data. Energy systemsknowledge (PV, batteries) or experience with real measurement datasets. Physics-basedsimulation, surrogate modelling, or multi-fidelity methods.
Funding
This3.5-year PhD project is fully funded and home students, and EU students withsettled status, are eligible to apply. The successful candidate will receive an annual tax-free stipend set at the UKRI rate (£21,805 for 2026/27) and tuition fees will be paid. We expect the stipend to increase each year. The start date is October 2026.
Link to apply: https://www.findaphd.com/phds/project/data-driven-digital-twins-for-measured-energy-systems/?p184389
#J-18808-Ljbffr