Image courtesy Iberdrola
Synthetic data is in its infancy in the energy sector but offers a breakthrough solution for modelling and planning to support smart energy systems, a policy paper from the Centre for Net Zero reports.
Smart meter data is one of the key datasets for utility operations, capturing granular household-level consumption data that can be used not only for traditional billing but for applications from power system planning to consumer-facing innovation.
However, broad access to such data is limited due to privacy and security concerns and while aggregation and anonymisation techniques have been used, the resultant datasets may not be fully privacy-secure and lack the level of original granularity.
Which is where synthetic data, already pioneered in areas including technology, healthcare and finance, comes in – synthetic data being data created by generative AI replicating the properties and characteristics of the original datasets but without the privacy concerns.
Have you read?
The how, why and benefits of synthetic smart meter data
EU Energy Projects Podcast: Building inclusive data spaces for Europe’s energy communities with project Data Cellar
The Octopus Energy-powered Centre for Net Zero has pioneered the use of smart meter data, developing ‘Faraday’, a synthetic smart meter data generator trained on 1.8 billion data points from 190,000 British households to be nationally representative, and opening up global access with the OpenSynth initiative through LF Energy.
In a new policy paper the Centre reviews five applications for synthetic smart meter data, i.e. consumer products and services, policy and regulatory design, electricity system and network modelling, development of financial products and services and housing development planning.
For example, integration of synthetic data in consumer facing products can help people optimise their energy use, compare tariff rates and save money on bills and it can improve distributional analysis when designing energy policies, ensuring interventions target those most affected.
Unlocking the potential of synthetic smart meter data
With the use of synthetic smart meter data in its infancy, the paper offers five principal actions that must be taken.
These are:
- Integrate synthetic data in data accessibility strategies. Governments should take a proactive and strategic role in promoting synthetic data alongside improving access to real customer smart meter data through consumer consent mechanisms.
- Develop commonly agreed standards for synthetic data quality. Researchers, industry and regulators should collaborate to develop common frameworks for objectively evaluating and quality assuring synthetic data.
- Remove ambiguity through explicit mention of synthetic data in legal frameworks and voluntary codes of conduct. Regulators should be explicit that ‘synthetic data’ is ‘deidentified data’ and therefore suitable for data sharing purposes.
- Fund innovation projects and support regulatory sandboxes to allow regulated industries to test and trial with new, synthetic data sources. Research councils, regulators and governments should prioritise funding for collaborative innovation projects, or establish regulatory sandboxes, to test and trial synthetic data in business operations.
- Collaborate internationally, taking an open source approach, to improve access to data, resources and skills. Through international, open source initiatives such as OpenSynth, researchers, industry stakeholders, and government bodies should share data, algorithms and evaluation techniques for synthetic data.
The paper points out that stakeholders across the sector have a role to play, including governments, the public and private sectors, regulators, researchers and industry.
“With the right policy support and industry collaboration, synthetic smart meter data can drive innovation, accelerate decarbonisation and empower consumers – while ensuring data privacy remains protected,” the paper concludes.
