Buildings account for around 40 percent of the primary energy consumption in Europe and in the United States. They also hold tremendous energy savings potential: 15 to 29 percent by 2020 for the European building stock according to a 2009 study from the European Commission.
Verifying and predicting the impact of energy conservation measures in buildings is typically done through energy audits. These audits are costly, time-consuming, and may have high error margins if only limited amounts of data can be collected.
The ongoing large-scale roll-out of smart meters and wireless sensor networks in buildings gives us access to unprecedented amounts of data to track energy consumption, environmental factors and building operation. This Thesis explores the possibility of using this data to verify and predict the impact of energy conservation measures, replacing energy audits with analytical software.
We look at statistical analysis techniques and optimization algorithms suitable for building two regression models: one that maps environmental (e.g.: outdoor temperature) and operational factors (e.g.: opening hours) to energy consumption in a building, the other that maps building characteristics (e.g.: type of heating system) to regression coefficients obtained from the first model (which are used as energy-efficiency indicators) in a building portfolio. Following guidelines provided in the IPMVP, we then introduce methods for verifying and predicting the savings resulting from the implementation of a conservation measure in a building.
Source: KTH
Author: Collard, Sophie