Considering human behaviour in forecasting energy consumption: insights from the ForNET project

Led by Professor Konstantinos (Kostas) Nikolopoulos, Director of the IHRR Forecasting Laboratory and Co-Director of IHRR, the ForNET (FORecasting Services for Energy NETworks) project addresses a long-standing gap in energy demand modelling — the limited consideration of how people’s behaviour influences consumption patterns.

Moving beyond traditional forecasting models

Conventional energy forecasting has typically relied on fixed cyclical patterns — such as time of day, day of the week, and seasonal changes — to estimate demand. While useful, these approaches overlook the dynamic and often unpredictable ways people respond to real-world events.

For example, a change in household composition or rising energy prices can quickly alter consumption habits. With the widespread adoption of smart meters, these shifts are now visible almost in real time, creating new opportunities for more responsive and accurate forecasting.

ForNET builds on this opportunity by incorporating behavioural insights into predictive models, enabling a more realistic representation of how energy is used in everyday life.

Addressing key challenges in energy demand

The project focuses on two critical and interconnected challenges: the influence of extreme events and the role of human behaviour in shaping energy consumption.

Weather events, policy changes, and socio-economic factors can all drive behavioural changes. For instance, rising energy costs may encourage people to spend more time outside the home — working in offices or public spaces — thereby reducing domestic consumption during the day.

Historically, these behavioural responses have not been captured in forecasting models. By explicitly accounting for them, ForNET is helping to provide a more comprehensive understanding of energy demand.

A data-driven approach to behavioural forecasting

To test this approach, the research team analysed publicly available data from 36 UK households, covering periods in 2021 and 2022. This dataset was combined with weather information and a range of behavioural indicators, including events such as tariff changes and COVID-19 restrictions.

The team explored multiple modelling scenarios, including using past data to predict future consumption and applying insights from one group of households to another.

As Professor Nikolopoulos explains:

“Conceptually, this is like having six new residents arriving in a neighbourhood of 30 people and trying to predict how these six additional residents will behave.”

This innovative methodology significantly improved forecasting accuracy, demonstrating the value of combining behavioural science with quantitative modelling.

Tools for decision-makers and industry

One of the key outcomes of the ForNET project is the development of datasets and modelling tools hosted on the Data & Analytics Facility for National Infrastructure (DAFNI) platform. These resources integrate behavioural, weather and consumption data, and allow users to explore a range of future scenarios.

The platform offers clear, practical benefits for energy providers, policymakers, and local authorities. For example, organisations can assess how new policies, pricing structures, or external events may influence energy demand in specific areas.

By enabling scenario testing and experimentation, the tools support more informed decision-making and planning.

Benefits for society and future applications

Beyond industry applications, the project has the potential to deliver broader societal benefits. A dedicated “microtool” allows analysis at the household level, helping providers to tailor services, improve efficiency, and reduce costs.

These insights can also support efforts to manage energy demand more sustainably, particularly in the context of changing climates and evolving consumption patterns.

Looking ahead, the ForNET team aims to expand the scope and scale of the project by working with external partners, including major energy providers and national organisations. Future developments may incorporate additional variables, such as households generating their own energy through solar power.

Professor Nikolopoulos highlights the wider ambition of the work:

“We wanted an organised environment where data and software for services can be uploaded that people can use in the future. DAFNI gave us a platform that people can access and experiment with easily.”

Collaboration and next steps

The ForNET project, which ran from May 2024 to February 2025, brought together a multidisciplinary team of researchers from Durham University and partner institutions, including York St John University, University College London, and Imperial College London.

With strong links to organisations such as the Office for National Statistics and the National Grid, the team is now focused on building new collaborations and scaling up the platform.

As the energy landscape continues to evolve, projects like ForNET demonstrate the importance of combining technical innovation with human-centred insights — ensuring that forecasting models reflect not just systems, but the people who use them.