NISTIR 6228: Analytical Study of Residential Buildings With Reflective Roofs
This report presents an analysis of the effect of roof solar reflectance on the annual heating (cooling) loads, peak heating (cooling) loads, and roof temperatures of residential buildings. The annual heating (cooling) loads, peak heating (cooling) loads, and exterior roof temperatures for a small compact ranch house are computed using the Thermal Analysis Research Program (TARP). The thermal performance requirements for the thermal envelope of the residence are based on prescriptive criteria given in the American Society of Heating, Refrigerating, and Air-Conditioning Engineers, Inc. (ASHRAE) Standard 90.2-1993. The residential models, with minor modifications in the thermal envelope for different locations, are subjected to hourly weather data for one year compiled in the Weather Year for Energy Calculations (WYEC) for in the following locations: Birmingham, Alabama; Bismarck, North Dakota; Miami, Florida; Phoenix, Arizona; Portland, Maine; and, Washington, D.C.
Building loads have been determined for a full factorial experimental design that varies the following parameters of the residential model: solar reflectance of the roof, ceiling thermal resistance, attic ventilation, and attic mass framing area. Attic mass framing area is defined as the exposed surface area of framing within the attic. The values for solar reflectance for the roof are varied from 0.1 to 0.8; ceiling thermal resistance, from "uninsulated" to R-8.6 m2-K/W (R-49 h•ft2-°FBtu); attic ventilation, from 0.5 h- to 9.2 h-1; and, attic mass framing area from 31.0 m2 to 46.5 m2. The computed results for annual heating (cooling) loads and peak heating (cooling) loads are illustrated graphically, both globally for all cities and locally for each geographic location. The effect of each parameter is ranked (highest to lowest) for effect on annual heating and cooling loads, and peak heating and cooling loads. A parametric study plots the building loads as a function of roof solar reflectance for different levels of ceiling thermal resistances and for each geographic location.
An analysis of building loads is presented for the building model conditioned under a transient solar profile and steady ambient conditions of either -1.1 °C (30 °F) for heating, or 26.7 °C (80 °F) for cooling. The time of year is specifically selected to provide equal hours of night and day. The site location for the model, in this case, is arbitrarily selected. Values for roof solar reflectance are varied from 0.1 to 0.8 and the ceiling thermal resistance, from "uninsulated" to R-6.7 m2-K/W (R-38 h-ft2•°F/Btu). Hourly predicted profiles are presented for the cooling load, attic air temperature, and outside roof heat flux. The results are in agreement with the annual loads computed for hourly weather data and indicate that radiative cooling of the structure at night is significant. The presence of thermal insulation in the ceiling reduced the effect.
Additional simulations are conducted for each location to determine the predicted exterior roof temperatures of the residential model for one year of weather data. For these simulations only the roof solar reflectance is varied. The other building parameters are fixed at "base" levels as specified by ASHRAE Standard 90.2-1993. The effect of roof solar reflectance is presented graphically using a box plot to summarize statistically all 8760 exterior roof temperatures. The results indicate a significant reduction in peak roof temperatures is possible for high levels of solar reflectance.
Additional plots for the daily roof temperature profiles for a typical summer day and average monthly temperatures for one year are also presented.
Included in this report is a simple economic analysis that examines cost savings for each geographic location. The estimated annual energy costs for electric and gas heating are plotted versus roof solar reflectance for different levels of ceiling thermal resistance. For a residence without attic insulation in a hot climate, substantial savings are available by making the roof more reflective. At higher levels of ceiling thermal resistance, the savings are less.
ASHRAE; absorptance; attic; building technology; cooling; energy; experimental design; heating; heat transmission; loads; model; radiation; reflectance; residence; roof; solar; TARP; temperature; thermal resistance; ventilation; WYEC
TABLE OF CONTENTS
THERMAL ANALYSIS RESEARCH PROGRAM
WEATHER AND SlTE DATA
Selection of Cities
Environment File Input
Site and Shading
Building Description Input
Factors and Levels
Simulation Runs for Building Loads
Global Ranking of the Factors
Ranking the Factors by City
Effect of Solar Reflectance on Predicted Annual Loads (Heating and Cooling)
Effect of Solar Reflectance on Predicted Hourly Peak Loads (Heating and Cooling)
BUILDING LOADS DISCUSSION
Effect of Roof Solar Reflectance on Building Loads
Effect of Solar Reflectance on Roof Temperature
SUMMARY AND CONCLUSIONS
A reflective coating, when applied to the exterior of a building, is designed to reflect incident shortwave solar radiation, thereby reducing the cooling requirements for a conditioned space. The practice is not new. Buildings in hot climates have long utilized light color construction (i.e., whitewashing) to minimize solar heat gains (Givoni 1969). In recent years, this design philosophy has received renewed attention in the United States, particularly in southern states with hot climates. Recent field studies by Parker et al. (1994) have demonstrated that reflective coatings for residential roofs can significantly reduce the cooling loads for buildings located in hot climates. Further, using reflective coating materials in entire communities has been proposed as a means for mitigating the temperature effect of urban heat islands (Akbari et al. 1995).
These studies indicate that significant reductions in the energy required for space cooling are possible, particularly in hot climates. The reduction in space cooling has generally been found to be more pronounced for residential buildings that are not very well insulated (i.e., low ceiling thermal resistance). In order to gain a better understanding of the subject, the National Institute of Standards and Technology (NIST) conducted an analytical study of a small, compact residential building using the Thermal Analysis Research Program (Walton 1983). The objective of the study was to address the following issues.
- How do the building parameters associated with the attic - roof solar reflectance, ceiling thermal resistance, ventilation, or framing mass area - affect the heating and cooling loads of the building in different climates? Which parameter is the most important?
- What is the impact of reflective roofs for thermally efficient construction, in particular, for newer residential construction with high ceiling thermal resistances?
- Does the presence of a reflective roof cause an undesired increase in the heating load during the winter heating season? If so, what is the net effect on annual loads?
- Finally, how do solar reflective roofs affect overall performance of a residence?
With regards to the last question, this report addresses four performance measures: annual heating (cooling) loads, peak heating (cooling) loads, exterior roof temperatures, and economic cost analysis. Depending on the reader - consumer, energy producer, or roof material manufacturer - these items may be useful for assessing the impact of a solar reflective roof.
The geographic locations selected for this study cover a wide range of climates in the contiguous United States; they are: Birmingham, Alabama; Bismarck, North Dakota; Miami, Florida; Phoenix, Arizona; Portland, Maine; and, Washington, D.C. The associated climatic data for heating and cooling ranged from 152 heating degree days, base
based on architectural guidelines of a small one-story "ranch style" house and the thermal performance requirements for the thermal envelope given in ASHRAE Standard 90.2-1993. The residential buildings were exposed to one year of hourly weather data compiled in the Weather Year for Energy Calculations (WYEC) by ASHRAE (Crow 1981).
For the analysis of building loads, an experimental design was utilized that collectively treated the effect of roof solar reflectance, geographic location, ceiling thermal resistance, attic ventilation, and attic mass framing area. In these simulations, the prime factor of interest was solar reflectance. The other factors were treated statistically as "nuisance" factors. To examine the effect of each factor at each level required 2700 separate computer simulations. Annual heating (cooling) loads and hourly peak heating (cooling) loads were extracted from each TARP output file and subsequently analyzed graphically using a statistical plotting program. An extended analysis of building loads was conducted for the model subjected to constant ambient air temperatures.
Additional computer simulations were executed to examine the effect of the roof solar reflectance on the exterior roof temperatures. In these simulations, the other parameters - ceiling thermal resistance, attic ventilation, and attic mass framing area - were fixed at "base" levels. For each geographic location, the simulations predicted hourly exterior roof temperatures for one year and were subsequently graphed as a function of roof solar reflectance. Additional plots for the daily roof temperature profiles for a typical summer day and average monthly temperatures for one year were also prepared.
Using the predicted building loads above, a simple economic cost analysis was prepared to examine at different levels of ceiling thermal insulation the benefits of increasing roof solar reflectance. Local residential utility rates for summer 1994 and winter 1994-1995 were obtained from the National Association of Regulatory Utility Commissioners and energy costs were computed by assuming local residential performance efficiencies for electric and gas heating equipment. Additional TARP simulations were conducted to determine the latent cooling load for each geographic location. The estimated annual energy cost was graphed versus roof solar reflectance for different levels of ceiling thermal resistance and for each geographic location.
This report presents an overview of the computer program TARP, geographic locations, descriptions of the residential model (including thermal performance criteria), experimental design, results and analysis of the computer simulations. Whenever possible, data from the computer results have been prepared in a graphical format for interpretation by the reader. These plots include predicted annual heating (cooling) loads, peak heating (cooling) loads, exterior roof temperatures, and a simple economic analysis of energy costs associated with the annual building load requirements. Samples of the TARP computer input files used in the analysis are provided in the appendices.
The Thermal Analysis Research Program (TARP) is a computer program capable of computing hour-by-hour temperatures, space heating, and cooling loads for an arbitrary building subjected to dynamic internal loads and external boundary conditions. TARP uses detailed heat balance techniques and an iterative procedure to determine heating and cooling requirements for a conditioned space. Multiple rooms are connected with the appropriate geometric data and coupled thermally with the proper heat balances on the surfaces between the respective rooms and air movements between rooms. TARP can be used to evaluate heating and cooling requirements for a single "design day"or one year of weather data. Further details on the computer algorithms are available in the TARP Reference Manual (Walton 1983).
The TARP program was written in Fortran 77 and compiled into three separate executable programs: 1) an environment file processor (EFP); 2) a building description processor (BDP); and, 3) a thermal simulation processor (TSP). The environment file processor prepares the input weather and site data. The building description processor prepares the input building data (geometries, thermal envelope specifications, zone conditions, etc.) for the simulation. The input data is subsequently processed by the thermal performance simulation processor which computes the building loads and provides report values of the results at time intervals specified by the user. The current release of TARP is portable and has been developed for execution on a personal computer (PC) in the DOS environment (i.e., text-in, text-out format). A single run of all three programs on a PC with a 486 DX CPU operating at 66 MHZ takes about two minutes.
Establishing confidence in the results of a computer simulation program is a difficult task. In general, the program can be checked against other programs, analytical test results, or, ideally, empirically data derived from field test results. TARP simulation results have been compared to empirical data collected from 6 test buildings located near NIST in Gaithersburg, Maryland (Walton and Cavanaugh 1985, Burch et al. 1986). Each test building contained a single room approximately 6.1 m by 6.1 m with a slab-on-grade foundation and pitched roof construction. The buildings had the same floor plan and orientation, but differed in exterior wall constructions. Table 1 compares predicted and measured loads for winter, spring (intermediate), and summer seasons for insulated and uninsulated exterior wall constructions. A negative value indicated that the average predicted load was less than the averaged measured load. In general, the differences for the insulated wood frame building were better than ± 21 %; for the uninsulated wall building, better than ± 9 %.
"WEATHER AND SITE DATA
The selection of geographic locations was based, in part, on the availability of local weather data for specific locations in the contiguous United States. For convenience, weather data were taken from the sets of hourly data compiled in the Weather Year for Energy Calculations (WYEC) tapes developed for ASHRAE (Crow 1981). These hourly values of data represent long-term means for temperature and solar radiation data for a geographic location (i.e., local airport data). Each tape includes 8760 sequential values of weather data such as dry-bulb and wet-bulb air temperatures, wind direction and speed, barometric pressure, and solar radiation. The data represents collectively a "typical" year of weather in the location. Presently, WYEC data has been developed for 46 locations in the United States and five in Canada.
Selection of Cities
The 46 metropolitan locations having WYEC data for the contiguous United States were ranked by the annual heating degree days or cooling degree hours using the climatic data provided in ASHRAE Standard 90.2-1993 (ASHRAE 1993), Figures 1 or 2, respectively. The purpose of ranking was to determine the range of climatic conditions for the experimental design. For heating and cooling, three cities (shown in bold) were selected as the upper value, lower value, and midpoint value. In Figure 1, the cities selected were (lowest to highest) Miami, Florida; Washington, D.C.; and, Bismarck, North Dakota. In Figure 2, the cities were (highest to lowest) Phoenix, Arizona, Birmingham, Alabama, and Portland, Maine. Hereafter, in this report, the metropolitan locations are ranked in order by heating, that is Miami, Phoenix, Birmingham, Washington, D.C., Portland ME, and Bismarck. The geographic location of each city is shown in Figure 3.
Environment File Input
The site and weather data for the TARP environment file processor (EFP) were geographic location data (i.e., latitude, longitude, etc.) for computing solar position, ground temperature, and weather tape parameters (for WYEC files). A separate computer file was prepared for each geographic location. (A sample environmental input file for Miami is provided in Appendix A.) Geographic data for the cities were taken from the ASHRAE Handbook of Fundamentals (1993). In the analysis, a constant annual ground temperature was assumed for each location (Kusuda 1981). Table 2 summarizes the average annual air temperature (ASHRAE 1993), heating degree days (ASHRAE 1993), cooling degree hours (ASHRAE 1993), and average annual ground temperature (Kusuda 1981).