This paper explores a new approach to planning and optimization of energy networks, using a mix of global optimization and agent-based modeling tools. This approach takes account of techno-economic, environmental and social criteria, and engages explicitly with inherent network complexity in terms of the autonomous decision-making capability of individual agents within the network, who may choose not to act as economic rationalists. This is an important consideration from the standpoint of meeting sustainable development goals. The approach attempts to set targets for energy planning, by determining preferred network development pathways through multi-objective optimization. The viability Of Such plans is then explored through agent-based models. The combined approach is demonstrated for a case study of regional electricity generation in South Africa, with biomass as feedstock. (C) 2008 Elsevier Ltd. All rights reserved.