Incentivising prosocial behaviour in community energy using multi-agent systems

Citation

Brooks NA, Powers ST & Borg JM (2025) Incentivising prosocial behaviour in community energy using multi-agent systems. International Journal of Computational Intelligence Systems, 18, Art. No.: 307. https://doi.org/10.1007/s44196-025-01060-7

Abstract
Community energy systems, where communities own their renewable energy sources, are key to the energy transition. But to effectively exploit renewable energy, communities need to reduce their peak consumption. For households, this involves spreading the use of high-power appliances, like washing machines, throughout the day. Traditional approaches rely on differential pricing set by utility companies, but this has been ineffective and raises issues of fairness and transparency. To address this, we investigate a decentralised agent-based mechanism. Agents, representing households, are initially allocated time-slots for when to run their appliances, and can then exchange these with other agents to try and better meet their own preferences. Previous work found this to be an effective approach to reducing peak load when social capital-the tracking of favours-was introduced to incentivise agents to accept exchanges that do not immediately benefit them. We expand this here by implementing appliance usage data from the UK Household Electricity Survey, to determine conditions under which the mechanism can meet the demands of real households. We also demonstrate how smaller and demographically diverse populations of households, with het-erogeneity in their demand patterns, can optimise more effectively than larger communities, and discuss the implications of this for designing community energy systems.

Keywords
social capital; reciprocity; community energy system; social learning; multi-agent systems

Notes
1

• Incentivising prosocial behaviour in community energy using multi-agent systems