Synthetic Data for Research and Innovation

What are the challenges?

Access to high-quality data is essential for energy system research and innovation. They can be used as inputs to models to test real-world impacts, can be used to validate our assumptions, or to provide new insights previously unknown. However, often such data is not available or has other challenges such as the following:

Data gaps

Incomplete or limited datasets can make it difficult to produce deeper insights or develop rigorous research.

Privacy Concerns

Data often contains sensitive information that requires additional protection or removal.

Commercial sensitivity

Organisations may be unable to share detailed data due to commercial restrictions.

Coverage limitations

Data may not be representative of all locations, communities, or technologies meaning insights generated from the data could be biased or misleading.

Synthetic data as a solution

There are two main ways that sensitive data can be shared which may reduce some of the issues and concerns outlined above.

One is to apply anonymisation techniques to the data. A simple example of this is to remove identifiable information such as location, names, or other information which may be unique to only a few individuals. There are other more complex approaches such as differential privacy, aggregation or adding noise to data to hide identifiable features.

Although such methods can allow more data to be shared they often have the disadvantage of reducing the usable information. For example, when the energy usage of a household is masked by aggregating with other households. This preserves the average behaviour but means individual actions and patterns are not discernible.

An alternative approach is to share synthetic data.  Synthetic data is artificially generated data that aims to preserve the characteristics and patterns of real-world datasets while protecting sensitive information. With sufficient statistical veracity it can help researchers explore scenarios, test models and develop new tools when access to real data is restricted.

The benefits of synthetic data?

Preserves privacy

Enables wider sharing

Fills data gaps

Supports AI and modelling

Creates data for locations where real data is unavailable


What we did in SENSE

SENSE is focused on two particular applications which have different data challenges and privacy requirements, which therefore mean synthetic data is a useful solution different purposes in supporting research. These are detailed below for each application:

1. Public EV Charging Data

THE CHALLENGE

Real-world public EV charging data is valuable for understanding usage patterns, the potential impact on the network and understanding utilisation. However, typically this data is only sparsely available in a limited number of regions. In our datasets these are for Scotland and Solihull. This restricted the ability to explore charging behaviours across different locations and demographic groups.


WHAT WE DID

We developed a synthetic public EV charging dataset designed to capture realistic charging patterns and extend the usefulness of the original data linked to location information (proximity to urban centre for example) and demographic data.

It therefore supports the generation and exploration of the impact of EV charging in a UK-wide range of locations, for different user types and different charging behaviours.


KEY OUTPUTS

1. Synthetic EV charging dataset

2. Synthetic data generation tool

3. Example datasets for research and testing

4. Generation of charging profiles for different locations and demographics

THE IMPACT

Researchers and innovators can explore EV charging behaviours, develop models and test solutions without needing direct access to the original location-specific data.

2. Hospital Energy Data

THE CHALLENGE

Hospital energy and operational datasets can contain sensitive or safeguarded information, which limits the wider data sharing potential and hence the insights and research which can be built upon it.


WHAT WE DID

In contrast to the synthetic EV data which aimed to augment the public datasets which are not sensitive information, we developed synthetic hospital datasets to preserve important characteristics of the original data while helping protect sensitive information.

The challenge here was to ensure that important features and patterns were retained in the model but no sensitive information was recreated in the synthetic data. This meant we developed a method which didn’t recreate unusual or unique patterns which may be linked to a particular hospital or department.

In addition, these datasets can also support the creation of representative data for hospital sites where real-world data is unavailable.


KEY OUTPUTS

1. Synthetic hospital dataset

2. Dataset generation capability

3. Example datasets for analysis and testing

4. Generation of data for additional hospital sites and scenarios

THE IMPACT

Researchers can work with realistic datasets for analysis, modelling and innovation without requiring direct access to the safeguarded source data used during development.

Why It Matters

Synthetic data can help unlock new opportunities for research and innovation by making high-quality datasets more accessible.

Through SENSE, synthetic data is helping researchers, innovators and organisations explore energy system challenges more effectively. It can:

  1. Reduce barriers to data access

  2. Protect privacy and commercially sensitive information

  3. Develop and test new tools and services

  4. Create representative datasets for future scenarios and locations where real data may not yet exist

By combining domain expertise with advanced data generation techniques, synthetic data can support innovation while maintaining trust and responsible data stewardship.

Resources

Get Involved

We are continuing to explore new applications of synthetic data across the energy system.

Whether you are a researcher, data scientist, technology developer or domain expert, we welcome collaboration and feedback.

Interested in learning more or contributing to future synthetic data activities?

Contact the SENSE team at: sense@es.catapult.org.uk