|
Trends in Heating and Cooling Degree Days: Implications for Energy Demand |
|
Weather-related energy use, in the form of heating, cooling, and ventilation, accounted for more than 40 percent of all delivered energy use in residential and commercial buildings in 2006. Given the relatively large amount of energy affected by ambient temperature in the buildings sector, EIA has reevaluated what it considers “normal” weather for purposes of projecting future energy use for heating, cooling, and ventilation. In AEO2008, estimates of “normal” heating and cooling degree-days are based on the population-weighted average for the 10-year period from 1997 through 2006. In previous AEOs, EIA used the National Oceanic and Atmospheric Administration (NOAA) 30-year average for heating and cooling degree-days as a benchmark for normal weather. Over the past several years, however, many energy analysts have questioned the use of the 30-year average, given the recent trend toward warmer weather relative to the 30-year average. Figure 21 shows percentage differences from the 30-year average in heating and cooling degreedays for the past 15 years. Over the 15-year period, only two winters have been colder, and all but three summers have been warmer, than the 30-year average; and on average, the winters have been 4 percent warmer and the summers 5 percent warmer than the 30-year average. Five of the 15 summers were more than 10 percent warmer than the 30-year average, whereas only 2 of the 15 winters were 10 percent warmer than the average, indicating a larger change for summer than for winter weather over the past 15 years. This suggests that the 30-year average is heavily weighted by years before 1993 and is less representative of heating and cooling degree-days in more recent years. The recent changes in average heating and cooling degree-days have not only affected the accuracy of AEO projections for heating and cooling demand. Underestimating summer demand for cooling—particularly, peak demand—can undermine the plans made by electricity producers for wholesale power purchases and capacity additions. Overestimating winter demand for heating can affect plans for natural gas storage and supply. Consequently, many energy analysts have suggested that shorter time periods provide a more appropriate basis for projecting “normal” weather. For example, Cambridge Energy Research Associates, Inc., now uses a 15-year period (1991-2005) to estimate normal weather in its projections for heating and cooling degree-days [63], and NOAA, responding to customer feedback, has undertaken a process to revise its traditional 30-year average by creating “optimal climate normals” that will be more representative of current weather trends [64]. EIA decided to use the 10-year average to provide a better match with recent trends in heating and cooling degree-days. Heating and Cooling Degree-Days in AEO2008 All the AEO2008 projections use the 1997-2006 average as a proxy for normal weather from 2009 through 2030. The 10-year average is based on heating and cooling degree-day data by State, provided by NOAA, and State population weights provided by the U.S. Census Bureau. The State population projections allow for dynamic estimates of heating and cooling degree-days at the Census Division level. Where State populations are expected to shift within and across Census Divisions, the projections for average heating and cooling degree-days at the national level can vary from year to year. Figure 22 shows differences in heating and cooling degree-days in the AEO2008 projection for 2010-2030 from the 1971-2000 30-year average published by NOAA. (It should be noted that the projection is not based on any assumption about global warming. Rather, expected U.S. population shifts cause the numbers of average heating and cooling degree-days to change over the projection period.) In 2010, the number of U.S. cooling degree-days in the AEO2008 reference case is about 10 percent greater than the NOAA 30-year average with fixed population weights, and the number of heating degree-days is 8 percent less [65]. Accordingly, electricity providers are projected to see more peak summer demand, and direct fuel use for heating in buildings is projected to decline through 2030 as a result of State population shifts, all else being equal. Impacts on the AEO2008 Projections Fuel Use in Buildings and for Electricity Generation Because space heating accounts for more direct energy use in buildings than does cooling, use of the 10-year averages for heating and cooling degree-days results in a 2.4-percent net decrease (about 0.6 quadrillion Btu) in buildings sector energy consumption in 2030, as compared with the same projection based on 30-year average heating and cooling degree-days (Figure 23). For electricity providers, on the other hand, the increase in electricity use for cooling is more than the decrease in electricity use for heating, and the result is a 0.7-percent net increase (about 0.4 quadrillion Btu) in fuel use for electricity generation. The effect on total net energy consumption in the reference case is small, amounting to a 0.4-percent decrease (about 0.4 quadrillion Btu) in 2030. As a result, expenditures for energy purchases in residential and commercial buildings are 0.4 percent lower in 2030 ($1.8 billion in 2006 dollars), and total CO2 emissions in 2030 are reduced by 0.1 percent (10 million metric tons). Electricity Prices As expected, the additional summer demand for cooling that results from using the 10-year average for cooling degree-days shifts more electricity demand into the summer peak period (Figure 24). In 2030, demand in the summer peak period increases by 4.4 percent, whereas winter demand is reduced by 0.8 percent. The increase in summer peak demand leads to higher real electricity prices, with average increases of 2.3 percent for residential customers and 0.3 percent for commercial customers.
63. CERA Advisory Service, “Monthly Natural Gas Briefing” (April 20, 2007). 64. NOAA Webcast, “Improving Climate Normals” (September 26, 2007). 65. A small amount of the difference is due to the use of dynamic population weights in AEO2008.
Contact: John Cymbalsky |
