Impact of Dry vs Humid Heat on Residential Air Conditioning Energy Consumption in the United States

Abstract

This thesis quantifies the impact of humid and dry heat on air conditioning (AC) energy consumption under present and future climate projections. It addresses three key questions: 1) Is there a discernible difference in AC energy consumption between humid and dry heat conditions? 2) How is AC energy use projected to change in response to changes in temperature and humidity? 3) What are the implications of future AC demand for energy systems? To answer these questions, we develop a machine learning model that uses dry-bulb (air) and wet-bulb temperatures to infer monthly residential AC energy consumption at the household and state levels. We describe the model’s development, sensitivity testing, and optimization. The model is trained on historical climate data from a reanalysis dataset (ERA5) and residential energy consumption data from the U.S. Energy Information Administration. It is then used to explore the relationship between temperature, humidity, and AC energy demand under current and future climate scenarios. Outputs based on CMIP6 climate projections are used to estimate future energy demand.