Modeling energy consumption of a multipurpose university building using multiple linear regression

Erin Leigh Volz

Growing concerns about the increasing amount of energy consumption and carbon dioxide emissions have caused a demand for energy efficiency improvements worldwide. University buildings are getting more attention in the energy efficiency sector because along with consuming a significant amount of energy, most of them contain many opportunities for energy savings. Creating a model that can accurately predict a building's energy consumption can lead to potential ways of managing and improving the building's energy consumption, and possibly lead to cost savings. An energy model is developed for a multipurpose academic building in Vancouver, WA, USA, using stepwise forward selection multiple linear regression. The multipurpose university building houses a variety of office, classroom, and laboratory spaces, causing the energy model development process to be complex. The energy model is based on cooling degree days and heating degree days, and takes into account the building's different energy consumption tendencies during the summer semester months. While the energy model is able to predict the building's energy consumption, it is discovered that there is noise in the energy model causing it to significantly overpredict, or underpredict, energy consumption certain at times. A statistical review of the energy model reveals that there is a low correlation between total building daily energy consumption and the cooling degree day data, as well as the cooling degree day coefficient in the energy model having low significance - creating noise in the energy model. An investigation into the building's energy drivers during the warm months of the year is conducted. Some of the major pieces of equipment that serve the building are metered to get a better understanding of their operations and energy consumption patterns. While this cooling performance investigation leads to important findings about the equipment and how they affect the building's cooling performance, more metering while the building is experiencing high ambient temperatures would be beneficial. Developing a model for this multipurpose university building is a challenging process due to the nature of the building, but it provides good insight into the building's energy consumption trends.

Modeling energy consumption of a multipurpose university building using multiple linear regression

Files and links (1)

Abstract

Metrics

Details