The PhD project will bring transformative change to decision-support tools currently used to balance the Great British (GB) electricity system. The intellectual merits of the project include advancing fundamental understanding of and theory for optimising and controlling a large number of units for balancing the Great British electricity system. The research builds on an existing collaboration with the National Grid ESO’s Electricity National Control Centre (ENCC) where a decision-support tool is collaboratively developed to enable ENCC to use a very large number of units to balance electricity demand and generation in real-time operation of the system. The decision-support tool is based on a mathematical programming model where various constraints of dispatch electricity generators to meet demand are expressed mathematically. While the tool provides theoretical guarantees of converging to an optimal solution, the computational times to reach a solution may be very large. Given the nature of real-time balancing of the British system, there is a desire to reduce the computational time and be able to manage the accuracy versus time trade-off.
The PhD project is motivated by the need to provide good quality solutions to the control room engineers in a short amount of time. The aim will be achieved by the following three strands of research:
· Predictive modelling leveraging available forecasts. This task will leverage the forecasts available to the system operator and will build predictive decision support models. Predictive analytics uses advanced analytics algorithms and provides means for risk assessment i.e. the evolution of the system under various realisations of uncertainty.
· Building credible operational scenarios. Machine learning methods require data – good quality reliable data. Modern data handling capabilities help ensure data quality at the source of origin, however, with the amount and speed of data the quality may be affected. Therefore, it is important that a data-driven approach acknowledges data issues and are robust to input errors. The research will propose robust methods to cleanse, and normalise input data and build a credible set of operational scenarios that represent year around (min by min) operation of the GB electricity system.
Data-driven decision support tools. A data-driven method (based on traditional approaches available within the subject area of machine learning) will be built using the training data, along with the optimal outputs obtained using mathematical models. Historic data on electricity system operation suggests that there are identifiable patterns in system operation. For example, the current operating condition either has existed in the past or has strong similarities to a realised operating condition. In this context, it is useful to do offline analysis using sophisticated tools that take time to return the optimal solution and use a machine learning method to use that information to provide fast decision-support to the control room.
For this exciting and important PhD we are seeking a student with an interest in applying advanced mathematical and computational techniques to electricity system problems. The project will require candidates to have excellent mathematical and/or computational skills and an interest in the energy sector. Prospective candidates should have or are about to receive a good Honours degree (a first class or a 2:1), an MEng or MSc with Distinction or with Merit in Electrical Engineering, Mechanical Engineering, Control Engineering, Computer Science, Mathematics, Physics or other related disciplines.
The PhD position is fully funded (tuition fees plus a stipend for living expenses) for UK home students through the Research Excellence Framework of the University of Strathclyde. Overseas students may apply and make use of the provided scholarship, though we note that they would need to provide their own funding to make up the difference between the UK and overseas tuition fees. The student will be supervised by Dr Waqquas Bukhsh and co-supervised by Prof Keith Bell. Dr Andrei Bejan, Optimisation Delivery Lead for the National Grid ESO Balancing Programme, is industrial supervisor on the project and will provide guidance on application of the research.
We anticipate competition for this studentship to be intense and encourage interested candidates to apply early. Interviews will take place during April and the expected start date is October 2023.
Applicants are requested to email their CV with a statement outlining their interest and suitability for the position to Dr Waqquas Bukhsh ([Email Address Removed]). Applicants from outside the UK should also indicate how they will cover the additional cost of international tuition fees over and above the fees charged to UK students.
Continue with Facebook
