Paper Conference

Proceedings of ASim Conference 2016: 3rd Asia conference of IBPSA-China, Japan, Korea

     

Multivariate linear regression model for estimating energy consumption

Elaheh Jalilzadehazhari, Krushna Mahaparta

Abstract: Windows as essential elements of buildings have a significant effect on the energy consumption, including heating and cooling demand in Sweden. A reliable statistical model for estimating the energy consumption associated with various window designs and glazing systems helps architects and designers in the early stage of the design process. Most of the introduced models in literature utilized a mathematic sampling algorithm such as Monte Carlo to develop a simple linear regression model for estimating the energy consumption. A simple linear regression model cannot describe the effect of different groups of a categorical variable. Hence this study considers four variables related to the window characteristics, including window size, design model, orientation, glazing system and develops a categorical multiple linear regression model for estimating the energy consumption. 544 simulations were performed by COMFEN Beta5 software. The results were used as a database for developing categorical multivariate linear regression models. The accuracy of the developed model was studied by the coefficient of determination, Rsquare value (R²). The obtained R² exceeded by 94%. Furthermore, the predicted energy consumptions obtained by the developed regression model were compared with the simulated values by COMFEN software. Results show a strong linear relationship between predicted and simulated values. Developed multivariate linear regression model can be utilized in the early stage of the design process for estimating the energy consumption associated with various window designs and glazing systems.
Keywords: Multiple linear regression, categorical variable, interaction analysis
Paper:
asim2016_276