Forecasting Chilled Water Consumption under Climate Change: Regression Analysis of University Campus Buildings

摘要

This paper discusses the development of a forecasting model for university campus energy use under climate change. University campuses are gradually modifying their building energy and environmental policies towards carbon neutrality. Particularly with the changing climate, a dynamic forecasting model is needed that can be used by the university administrators to forecast the demand of energy consumption, e.g., chilled water use. The model development follows a three-step process: (1) data collection and cleaning, (2) descriptive statistics, and (3) statistical modeling. While the dependent variable is chilled water consumption, the independent variables are U-factor of roof, U-factor of walls, U-factor of windows, window wall ratio, lighting power density, equipment power density, building age, building age after latest year of renovation, temperature, humidity, pressure, and wind speed. For training and testing, historical weather data is used. The validated model is used to estimate energy use with future weather data. Prior to modeling, matrix plots and histograms were used to identify correlations between variables, which is followed by data transformation and normalization. Finally, a non-linear regression model was developed to predict the chilled water consumption under climate change. A key finding of the model is that temperature is the most significant variable in chilled water consumption. In addition, building height, building age, and U-factor of walls were found to be important in descending importance. Based on the regression analysis, it was found that chilled water consumption will increase almost three times in the year 2054.