AbstractThe evolution of energy infrastructures towards a more distributed, adaptive, predictive and marketbased paradigm implies an effort on combining communication protocols and energy transmission and distribution systems in a common architecture. This architecture should allow decentralized control in order to be able to manage efficiently distributed generation, storage and exchange of energy between sources and sinks. Dynamic energy management models are a part of this “systems thinking” vision that aims to create a new field of applications that is at the intersection of computing science and energy technology. The broader implications associated with them are related with the possibility of creating communities that integrate energy supply and demand within a given region, in order to limit their impact. In order to push intelligence to the energy networks’ edges, up to individual sources and sinks, scalable and flexible distributed systems will have to be build. In this sense, data mining techniques and multicommodity network flow models can be combined for pattern detection, forecasting and optimization, which are essential features of dynamic energy management.