Time series and data mining techniques have recently become popular for smart grid planning and optimization problems, in applications such as demand forecasting and renewable energy availability prediction. In the future liberalized smart grid with distributed generation and time varying resource pricing and availability, optimizing and sizing centralized and distributed energy resources for profit maximization will become more important. In this paper we investigate the usefulness of time series clustering techniques to reduce the computational complexity of smart grid optimization problems. We focus on the demand side problem of local storage sizing for renewable integration, while highlighting the importance and general applicability of these techniques. We also build and deploy a web-based decision support system to encourage the deployment of rooftop solar. © 2013 IEEE.