The building sector is a major contributor to total energy consumption in most countries. Traditionally, researchers have focused on leveraging energy efficiency by improving building materials, in-house facilities and transmission equipment. More recently, however, there has been increased focus on research concerning demand-side energy consumption behavior. Current research suggests that energy efficient behavior of a building's occupants can be extensively enhanced through the sharing of energy consumption information among residents in a peer network. However, most of this research relies on experimental tests and does not theorize concepts related to peer network energy efficiency systematically. My dissertation addresses this research gap on two levels. First, I examined if and how the structure of peer networks can impact residents' conservation behaviors through network analysis by employing agent-based simulation techniques. Following confirmation of the impact that network structure has on user behavior, I created a layered network model to integrate information from various network layers and a block configuration model to reconstruct increasingly reliable random networks.
In contrast to controlled energy efficiency experiments, real-world networks are large in size, heterogeneous in nature and regularly interact with other networks. By utilizing models developed in this dissertation, we are able to estimate the contribution of network structural coefficients to the energy consumption performance of peer networks. By comparing the layered network and block configuration model I developed with other conventional models, I prove the efficiency, accuracy and reliability of these improved models. These findings have implications for assessing network performance, creating accurate complex random networks for large-scale research, and developing strategies for network design to improve building energy efficiency. This research establishes a system to study residents' energy efficient behaviors from the perspective of peer networks and proposes some instructive models for further energy feedback system design.