Integrating the UK’s energy systems and climate data

Case Study Overview

As part of an effort to support the decision-making necessary to decarbonise our society, we have extended a digital twin implemented as part of the World Avatar to help understand the impact of different options for how to heat our homes.

What challenges does it solve?

Current energy policy research lacks open data and modelling transparency. This impedes the ability to reproduce results and to adapt and combine existing models.
Our approach enables a dynamic, interpretable, modular, and cross-domain representation of information, in this case describing the UK, allowing domain-specific experts to contribute toward a national-scale digital twin.

We have integrated temporal, geospatial, technical, environmental, and social data. The data can be holistically analysed, providing actionable information to support policymaking.
The digital twin’s design is universal – it can, and will, be extended to include other types of data. This is predicted to become increasingly important to integrate open and transparent data and models to support future decision-making and analysis of different energy scenarios.

How have you integrated data & technology?

The World Avatar uses a knowledge graph to integrate and enable interoperability between datasets from heterogeneous domains that are inherent to decarbonisation and energy supply research.

The information is represented using ontologies expressed as a directed graph, where the graph’s nodes represent concepts and instances, and the edges between nodes represent the respective relations between nodes. Specifying the relationships between data provides more context and makes the information more accessible, making it easier for computational agents to interpret, query, and update the data.

This integration of heterogenous domains can be seen in the instantaneous flow rate data shown on the map of the UK gas grid. Despite referring to the same physical gas terminals, the original source information is not cross-referenced in a consistent way. Therefore, it becomes increasingly difficult to keep track of sources of information as they relate to the same physical entities.

This issue is common when modelling energy systems, where systems such as gas and electricity overlap. Currently, approaches are bespoke and often complex, for instance, developing new management tools to support cross-domain interactions. But this approach makes reproducing the results of previous studies difficult and creates a duplication of effort in new studies, leading to inconsistencies between analyses.

By instantiating each instance of an entity as a node in the knowledge graph, to which each disjoint data set can link, we can unify this information allowing computational agents to identify links between sources of information, therein providing an explicit, coherent representation. Moreover, should additional information come to light, it can be linked to the relevant entity in the knowledge graph without requiring knowledge of existing data.

Do you use a user interface to share information?

The World Avatar project exploits knowledge graphs to enable interoperability between models and data from different domains. The knowledge graph is dynamic – it is operated on by computational agents that read, manipulate, and update data and calculate quantities of interest.

In this case, a set of input agents was developed to incorporate data into the dynamic knowledge graph. The agents demonstrate both the addition of data describing the physical infrastructure of the gas transmissions system and the addition of live feeds of real-time data describing the intake of gas into the transmission system, so the dynamic knowledge graph remains current in time. Agents were developed to process the data to evaluate different decarbonisation options and to allow visualisation of the results. This enables users to analyse the data and policymakers to make informed decisions regarding how to incentivise the adoption of air source heat pumps.

Given a suitable ontology, it is possible to represent anything. Temporal, dynamic, and geospatial data can be integrated, facilitating the complex representation of systems starting from simple sets of rules. The ability of computational agents to input data, simulate the behaviour of systems and provide output has led to the suggestion that a dynamic knowledge graph technology could provide a suitable architecture for implementing a Universal Digital Twin.

What outcomes have you delivered?

We have quantified the temporal and geospatial impact of the transition from using natural gas to heat pumps for domestic heating.
We quantified the performance of heat pumps using historical climate data and estimated the change in household emissions and fuel costs caused by switching from gas heating to heat pumps.

We showed that using heat pumps would reduce emissions throughout the UK, where the change was primarily proportional to the reduction in gas use.
By extending the analysis to consider fuel poverty, we identified areas of high fuel poverty that would experience significant increases in fuel costs, highlighting the tension between the environmental goals of the UK government and the aspirations of its ‘levelling-up’ agenda to reduce inequality, and providing an evidence base to support the development of local policies to support decarbonisation.

Have you delivered any unexpected benefits?

We showed that air source heat pumps would most often cause an increase in fuel costs (2019 energy prices) and that the change would be significant compared to existing energy costs; with household fuel costs predicted to change between +£200 and −£30 per month, depending on the assumptions and location.

However, the geospatial distribution of the change in cost was insensitive to the model assumptions. This is because electricity was significantly more expensive than gas (2019 prices), such that the increase in electricity use (which is more strongly coupled to the prevailing climate after the adoption of heat pumps) had a stronger impact than the reduction in gas use.

We identified specific regions that would experience a disproportionate increase in inequality, confirming existing inequality pictures of the UK. The analysis further elucidated the political dilemmas posed by the ambition to reduce inequality and reach net zero.

Moving towards more sustainable energy use requires the consideration of the practical and socio-economic implications for UK citizens. Thus, politicians must discuss accompanying policy interventions such as place-based strategies to counteract fuel poverty.

What lessons have you learn that apply to future work?

The design of the World Avatar provides a principled framework for creating digital twins that enable the discoverability, reusability and interoperability of data and connected models.
We achieved interoperability leading to substantial results using only simple rules (i.e., simple ontologies) to govern the data represented in the digital twin.