The present work deals with improving the operation of a multi-energy district boiler by adding to the plant an optimally-designed thermal storage tank and developing a model-based predictive controller in order to manage it. The proposed architecture generates optimal command sequences dealing with the amount of thermal energy to be stored or released. In order to implement such a controller, one needs to forecast the power demand. As a consequence, a short-term forecast method, based on both a Multi-Resolution Analysis (MRA) and feedforward (multilayer) Artificial Neural Networks (ANN) is proposed. The present paper mainly focuses on the impact of Thermal Energy Storage (TES) on the functioning of a northeast France multi-energy district boiler selected as a case study. As a result, both the fossil energy consumption and CO 2 emissions are significantly reduced while the economic gain is increased.