Empirically Modeling the Energy Implications of Cool Roof Retrofits

Abstract

The energy effects of roofs have primarily been studied using theoretical models. However, empirical methods are needed for validation or when not all drivers of energy change can be observed. This study used mixed empirical techniques to estimate the impact of whitening existing roofs. Two years of hourly site cooling energy use were collected for 114 homes in Austin, TX. Seven properties selected to have their roofs coated white at no charge, imitating incentive policies. The empirical results are mixed, primarily limited by the small treated sample size. Individual household comparisons generally demonstrate statistically significant impacts of whitening, ranging from 14% to 49.2% reductions in daytime site cooling energy use and a 9.7% increase to a 40.3% decrease in nighttime site cooling energy use. Ordinary least squares regression estimates statistically significant mean daytime reductions of 9.7% but no significant nighttime effects. However, clustering the standard errors by household substantially reduces the significance level. Tweedie regression, which better accommodates slightly inflated zeros in the sample, demonstrates significant daytime and nighttime reductions of 14.3% and 5.5%, respectively. A life-cycle costs analysis using a mix of primary and secondary data explored the secondary benefit of delaying roof replacement resulting from the added protection of the coating material. Absent this delay, whitening would pay for itself in only 41% of simulations with a median net cost of $310 and simple payback of 22 years. Including a benefit to delay roof replacement, the median payback period is 1 year and the median net benefit is $310 across a range of assumptions. However, the benefits of whitening are only robust for older roofs and longer service lives for the coating. For roofs older than 10 years of age, most simulations reflect net benefits if the coating lasts at least 5 years.