Modelling and optimization of a large-scale urban energy system with sufficient spatial resolution is a complex challenge. By proper clustering technique, a large-scale problem could possibly be divided into small ones with high spatial resolution and accuracy. Existing literature tends to lower the complexity of large-scale urban energy system problem by accumulating demand profiles on the spatial dimension. This study proposed a new combined clustering approach which considers not only the spatial dimensions, but also the load characteristic of all buildings to solve a large-scale urban energy-water nexus optimization problem. The load complementarity can level off the total demand profile, which is helpful to obtain more economic benefit. The approach can divide district with a large number of buildings into small clusters including fewer buildings. By using complementarity indexes, the load heterogeneity of each cluster can be assessed. And the density of each cluster is used to investigate the distance among buildings within the same cluster. The combined clustering approach consists of two different routes: one is to lower down complementarity index with density as constraints; the other one is evaluating both two criteria simultaneously as a single objective. Through a case study, the proposed combined clustering approach can generate a new clustering map and finally save 4.4% total cost compared to density-based clustering approach.
Keywords urban energy system, combined clustering, large scale, OPTICS, energy-water nexus, demand complementarity