A community energy management system for smart microgrids

• Published in: Electric power systems research Vol. 209; p. 107959
• Main Authors: Verba, Nandor, Nixon, Jonathan Daniel, Gaura, Elena, Dias, Leonardo Alves, Halford, Alison
• Format: Journal Article
• Language: English
• Published: Amsterdam Elsevier B.V 01.08.2022; Elsevier Science Ltd
• Subjects: Algorithms; Cloud computing; Controllers; Cyber-physical systems; Decision making; Distributed generation; Electricity distribution; Energy management; Genetic algorithms; Internet of things; Load shedding; Optimisation; Playgrounds; Priorities; Refugee camps; Refugees; Retrofitting; Solar; Testability; Rwanda; Cyber-physical systems; Internet of things; Refugees; Solar; Optimisation; Load shedding
• ISSN: 0378-7796; 1873-2046
• DOI: 10.1016/j.epsr.2022.107959

•Different smart micro-grid energy management control overlays are proposed.•The controllers are deployed and evaluated for a refugee community micro-grid.•Utility functions are modelled to increase battery life, user satisfaction and system efficiency.•A modified GA search-space reduction model outperformed other control methods. Community micro-grid energy projects are needed to drive de-carbonisation and increase equity of energy systems among displaced communities. However, micro-grid solutions are often inflexible and lack functionality to respond to displaced community energy needs and ensure the long-term sustainability of interventions. This paper explores the use of fog-computing retrofit architectures deployed on community micro-grid infrastructures to enable flexible demand management to improve service delivery and longevity. A micro-services solution is proposed that decouples components increasing resilience and testability while allowing hybrid edge-cloud deployments. The architecture is outlined and demonstrated for a micro-grid providing energy to two nurseries and a playground in Kigeme refugee camp, Rwanda. To enact the community priorities within the demand management system, modified Genetic Algorithm (GA) methods are outlined and tested for different use-case scenarios. The performance of the GA methods are then compared with a pre-existing battery protect controller and an alternative deterministic (space-shared) energy manager model. To achieve a high utility function in almost every use-case, a search space GA method was required for GA to outperform both the existing battery protect controller and proposed deterministic method. The results show how simple community priorities can be set and used to enact control on the system in 24h timeframes that are in line with the local decision-making context.