Energy system design should take into account the hourly, daily and seasonal variations of both the energy demand and the considered utilities, and therefore requires a multi-time-resolution problem formulation. Multi-period / multi-time optimization is needed when a multi-time (e.g. hourly) optimization is performed inside another multi-period (e.g. typical day) optimization. However, optimizations over large temporal or spatial horizons tend to become computationally expensive, due to the large number of variables and constraints indexed over the times and over the periods. Employing typical operating periods (e.g. a number of typical operating days during the year) offers an interesting solution for problem size reduction. A variety of data clustering algorithms have been proposed in literature in order to select the best typical periods for different applications. This work uses a MILP formulation of a k-medoids based algorithm (PAM) in order to obtain typical operating periods which pass energy from one period to another, in view of performing long term energy storage. The algorithm is used coupled with an optimization of a CO2 based district energy network in a typical urban center. The intra-daily resolution allows the exploration of short term energy storage in the form of batteries located in the medium and low voltage grid. Coupled with the seasonal resolution, it offers a better understanding of the impact of daily storage on the long term storage and on the total energy requirement. The results show that implementing short-term energy storage leads to reductions of 2% in the size of the long term storage tank and 7.5 - 7.8% in the size of the main energy providers (PV panels). © 2018 Elsevier B.V.