Social networks save energy: optimizing energy consumption in an ecovillage via agent-based simulation

Abstract

Energy-conscious communities are continually challenged to optimize electricity usage, maximizing the benefits obtainable from generating systems and minimizing the reliance on the national supply. Achieving an ideal balance is complicated by the fluctuating availability of ‘green’ supplies and the varying patterns of domestic usage. Optimizing a community's net energy balance depends on the degree to which householders can modify their usage patterns. We consider two questions here: given a collection of realistic preferences and constraints on usage patterns of households, what degree of saving is possible by optimizing ‘within’ these preferences and constraints?; and, what amounts of energy saving are possible when the community exploits its social network by sharing electrical appliances? These questions are investigated in the context of Riccarton Ecovillage. A model of the ecovillage was implemented using an agent-based modelling toolkit, then simulated under a range of scenarios, automatically exploring the space of usage patterns to find combinations of usage schedules that minimized dependence on the national supply. Our findings are: evolutionary algorithms perform particularly well at this difficult optimization task; modest savings of 5–10% are achievable under standard assumptions, but savings of 35–40% are achievable when exploiting the underlying social network.

Keywords:

• ABM
• agent-based modelling and simulation
• ecovillage
• energy
• evolutionary algorithm
• optimization