Continuous monitoring for three years documented membrane temperatures and insulation heat fluxes under ballasted roofs and for roofs with exposed white and black membranes. The overall goal of the project was to evaluate how the thermal mass of three different loadings of stone ballast and a heavy paver, all with relatively low solar reflectance, affects energy performance, especially compared to the highly reflective white roof. This paper summarizes the results of the measurements for all three years. They indicate that thermal mass effects are significant for the low R-value roofs in the climate of East Tennessee. Cooling loads for the heavily ballasted systems and the weathered white roof are nearly the same. The lighter ballasts had cooling loads more than the white roof but less than the black roof. The heating loads for the heaviest stone-ballasted system are slightly less than for the black roof. For the paver and the other stone-ballasted systems, heating loads are nearly the same as for the white roof.

An important goal was to predict energy performance with more typical roof insulation levels and in climates different from the test climate. An effort was made to model the energy performance of all six systems in the test climate with the Simplified Transient Analysis of Roofs (STAR) program. For the black roof relative to the white roof, predicted differences in cooling and heating loads were both slightly higher than measured differences. This is consistent with anomalies in the measurements, including the effect of moisture, which STAR did not model.

For the ballasted systems, effective thermal conductivity and specific heat for use in STAR were estimated by trial-and-error, guided by diurnal behavior of the test roofs. For the ballasted roofs relative to the white roof, differences in cooling loads were very similar to those from the measurements as ballast loading and type were varied. The trends continued with higher roof insulation levels and more severe cooling climates than for the measurements. Using these same properties, differences in heating loads were significantly larger than measurements. STAR is too simple a model to predict heating loads for ballasted roofs.

Presented at Thermal Performance of Exterior Envelopes of Whole Buildings X – December 2007

Units: Dual