City-scale single family residential building energy consumption prediction using Genetic Algorithm-Based Numerical Moment Matching Technique

Abstract Growing energy consumption in urban areas has increased the importance of planning for future energy systems. Thus, improving the modeling abilities for predicting energy consumption at the city scale is critical. In this study, a Genetic Algorithm-Based Numerical Moment Matching (GA-NMM) method is adopted as a primary uncertainty estimation technique to predict the electricity consumption of a large dataset of single family homes by utilizing key features in energy audit and assessors data. This data is used as an input to the GA-NMM to develop a set of index buildings and associated weighting factors that represent statistical characteristics of the dataset. Energy models are then developed for the index buildings using physics-based energy modeling in EnergyPlus. These, in combination, are used to estimate the energy behavior of single family homes of the studied dataset. The proposed method is applied to a large dataset of 8,370 single family homes in Cedar Falls, Iowa, where the expected annual and monthly electricity consumption from the model is calculated and compared with measured data. The expected site electricity consumption for single family buildings in Cedar Falls is estimated as 10,219 kWh/yr, which is within 6% of the measured average annual electricity consumption. At a monthly level, the Coefficient of Variation of Root Mean Square Error and Mean Bias Error are 7.8% and 4.5%, respectively. This method can be used to generate small set of representative homes for demonstrating the energy behavior of a larger set of homes.