ABSTRACT. In this paper a two-layer controller is proposed to tackle the building energy management problem for hybrid systems at different levels of abstraction and different time scales. In the upper layer a relaxed long term energy allocation problem with a large decision time step is defined, taking into account the energy prices, the comfort requirements, and a global power constraint. The discrete decision variables are considered only in the lower layer, where the continuous global solution computed by the first optimization is projected into local mixed-integer programming (MIP) tracking problems with a shorter prediction horizon and a higher sampling rate. To fulfill the building global power constraint each load has a specific priority to access the available power, following a non-iterative priority algorithm.