Data for Digital Decarbonisation (3D): A FAIR approach to energy demand data in buildings

It is a salient truth that the better clarity and structure of data relating to a problem, the easier that problem will be to solve. In the words of Linus Torvalds, the inventor of the Linux operating system upon which the majority of today's internet runs:

"Bad programmers worry about the code. Good programmers worry about data structures and their relationships." - Linus Torvalds

This proposal is fundamental, early stage research into how data for energy demand in buildings is structured and stored. The output will be proof-of-concept of new data structures and techniques to greatly improve our ability to make design and policy decisions for zero-carbon buildings. The proposed work takes a selection of existing, established open-access energy datasets and converts them to highly-structured versions using the FAIR open data guidelines (https://www.go-fair.org/fair-principles/). This is a small but crucial first step in moving the energy data community from 'closed-world' to 'open-world' data models to enhance openness, transparency and collaboration.

Why do this? Reducing the energy demand from buildings is of key national importance in delivering the zero carbon economy. This is largely a policy challenge, as many proven building retrofit technologies already exist and successive governments have found it very difficult to develop long-lasting successful policy initiatives for the sector. Perhaps inspiration can be found in the Government's recent policy response to the Covid-19 crisis. Here the combination of data and modelling from many cooperating epidemiology academic groups added weight to the arguments for action and helped drive the policy of the pandemic response. A similar initiative is needed in the energy demand response to climate change; rather than isolated academic and/or professional groups developing data and tools which only they understand, a fundamental shift in required to transform collaboration and reproducibility within the field so that significant momentum for policy design and implementation can be created.

Epidemiology, and the medical field in general, have been at the forefront of developing new data standards and structures due to the critical importance to avoid misunderstandings in data interpretation. This greatly informed the development of the FAIR open data guidelines which have been widely adopted by the Open Research movement. FAIR stands for Findable, Accessible, Interoperable and Reusable, and the guidelines give a set of broad principles for open research data provision. Although widely discussed, very few datasets in the energy demand field are fully compliant with the FAIR guidelines. Of particular note is the requirement to use Unique Identifiers to represent concepts and information.

Ultimately, meeting the FAIR guidelines require a shift away from simple data structures such as Excel spreadsheets and csv files to more refined, machine-readable and standardised data representations such as node-edge graphs and Resource Description Framework (RDF) structures. These more complex data structures are harder to initially create but contain much more embedded meaning and logic, both in the information and the relationships between the points of information, so that they naturally are better suited for complex analyses and for other users to understand and reuse. The premise of this work is that a sector-wide move to FAIR-compliant energy demand data is an underlying, necessary requirement for a truly collaborative and innovative environment for energy research informing future technology, design and policy developments.