Providing energy consumption comparison between occupants through peer network has been proven an effective approach to promote energy efficient behavior. However, an individual’s peer network is not isolated from other networks, which also impact energy saving behaviors. In order to understand how energy saving behavior can spread in heterogeneous networks, we propose a multilayer network system simulation model, which not only analyzes the properties of networks, but also simulates the residents’ energy saving response to energy feedback information. The proposed model is capable of integrating multiple network layers with given arbitrary degree distributions into an equivalent layered network. With different network types, the model is able to provide explanatory powers on energy conservation practice diffusion through various social and physical settings. To validate the simulation results, a multi-layer network system was constructed as a combination of a layer of peer network and a layer of geospatial network by simulating a random scale-free network and a clustered network. The sample multi-layer network system in this paper includes residential peer networks and geospatial building networks, could predict changes in energy saving and simulate feedback perturbation. The validation tests indicate that comparing to other traditional models our proposed multi-layer network system simulation model is accurate and efficient enough to predict and mimic the diffusion pattern of energy saving behavior through multiple layers of networks. Moreover, our model allows us to estimate energy saving at both building level and inter-building level with different network combinations. Therefore, our work provides a new method for predicting energy saving behavior resulting from energy consumption feedback that incorporates with multiple networks.