TechUK and the Centre for AI and Climate have published a new paper on how machine learning and artificial intelligence (AI) can support the transition of the electrical grid to net-zero.

The ‘AI for Energy’ report is set to be the first in a series of whitepapers techUK, the technology industry trade association, will be releasing in conjunction with its partners over the next year, which will outline how emerging technologies can support the decarbonisation of the energy sector.

The report cited a recent Microsoft and PwC report which estimates that applying AI to the energy sector can help reduce global emissions by up to 4 per cent against business as usual by 2030, while boosting GDP by up to 4.4 per cent.

Use cases outlined in techUK’s report include grid management, where it said AI could optimise the management of existing network operation processes and support the networks’ transition to net zero.

This would include using AI to optimise “grid dispatch”, which is the process where system operators determine how much power controllable generators should produce over a range of timescales.

Another use case highlighted by the report is renewable generation forecasting, where it said AI could be used to reduce the uncertainty grid operators have in either electricity supply or demand forecasts, reducing the back-up power, known as “spinning reserve”, that is needed.

Spinning reserve generally comes from easy to store fossil fuels such as gas.

The report said that balancing the UK electricity grid currently costs end-users about £300 million per year.

The report also covered how machine learning could improve demand-side response, by enabling dynamic pricing and trading, creating market incentives and price signals, and using smart meter signals or learned user preferences to use energy when there is plentiful low carbon energy on the grid.

The report also explores how AI could be used to optimise the efficiency of individual assets, such as windfarms, within the energy system.

AI could be coupled with time series, telemetry data, weather data and maintenance data to uncover the drivers behind wind farm production-decreasing events and allow wind farms to increase productivity by reacting to these negative events before they occur, according to the report.

TechUK cited a case study where Deepmind, Alphabet’s AI Lab, applied their machine learning algorithms to Google’s wind farms, and in doing so claimed to boost the value of the wind energy by roughly 20 per cent compared to baseline.

The report also touched on how AI could help with the electricity-based challenges the increasing number of EVs on the roads will create, suggesting machine learning combined with price signals, could encourage customers to charge vehicles throughout the night and avoid excessive demands on the grid.

TechUK’s report also touched on the potential application of the emerging technologies for domestic building and home management, increasing the energy efficiency of commercial and industrial facilities, and microgrid management.

The report comes after prime minister Boris Johnson promised to mobilise £12 billion in November 2020 to help the UK achieve carbon neutrality by 2050.

“The decentralisation and decarbonisation of the grid means that the complexity of managing the energy system increases exponentially,” said Susanne Baker, techUK’s associate director for climate, environment and sustainability. “AI and machine learning is ideal in helping to manage that complexity.”

She added: “But to do that we need to be really clear where the opportunities are so we can focus our innovation efforts.”

Peter Clutton-Brock, Centre for AI and Climate, said: “AI and machine learning will not be nice-to-haves in managing and optimising a more complex, renewables-dominant grid but fundamental pre-requisites of zero emission electricity systems.”

He added: “This paper highlights some of the ways that AI can make significant contributions towards the transition to zero emission energy, however, ultimately AI will flow into all decision-making processes associated with electricity systems.”

Share Story: