Decreasing the Energy Performance Gap and Thermal Loads in Residential Dwellings Using Machine Learning and Building Energy Modeling

The energy demand of the residential sector in Canada represents 12% of the total energy use. About 64% of such demand is assigned to space heating requirements. Adoption of energy-efficient systems is usually driven
by long-term savings. Many passive energy reduction measures haven’t been implemented due to the decreasing credibility of simulation software which lacks the ability to accurately predict energy performance. The deviation
of the actual energy consumption from the simulated one is termed as the energy performance gap (EPG). This research will study multiple residential dwellings categorized based on input variables relevant to a building’s
geometry, envelope, and site parameters. The actual energy consumption of these dwellings will be compared to that of simulated models to calculate the EPG. Artificial neural network will be employed to model the EPG in
correlation with the most dominant independent design parameters. Once EPG is determined using the developed model, an algebraic equation will be developed to better assess energy-efficient strategies by defining their exact
potential and true savings. The model will be validated using real-time thermographic images of a built prototype with a special emphasis on thermal bridging.

Faculty Supervisor:

Mohamad Araji

Student:

Partner:

Canadian Museum of Architecture

Discipline:

Engineering

Sector:

Professional, scientific and technical services

University:

University of Waterloo

Program: