Power Demand Forecast: Scientific Basis

Home > Scientific Basis for Weather-Based Electricity Demand Forecast

Scientific Basis for Weather-Based Power Demand forecast
The basis for our electricity demand forecast is the fact that variation of electricity demand is closely related to the change in ambient air temperature. If air temperature can be predicted then the demand for electricity can be predicted.

Figure 1: Seasonal variation of electric consumption
Figure 1 shows the weekly national electric output from the beginning of 1992 to present, as reported by the Edison Electric Institute (EEI). The electric output has two peaks in each year (seasonal variation): one in winter and one in summer. The output also shows a trend of year-to-year growth (called longterm variation which is due to factors such as population growth and economic development). One can see that the magnitude of the seasonal variation is overwhelmingly larger than that for the longterm variation.
The peaks in summer and winter are mainly caused by the use of electricity for space cooling and heating. Thus the ambient air temperature is the main factor that dictates the electricity demand. Figure 2, for example, shows the relationship between temperature and the EEI-reported electric output for the Mid Atlantic region.

Figure 2: Sensitivity of electric consumption to temperature
The close relationship between the electric output and temperature is clearly shown in the figure. We can see that electric demand is high when temperature is either high (summer) or low (winter). Minimum output occurs when temperature is moderate (spring and fall).
The actual relationship is potentially better than suggested by Figure 2. This is because in Figure 2 the weekly electric output hasn't been partitioned by season. We expect to see different electric outputs for the same temperature in spring and fall, due to the different heating and cooling patterns. Further more, all weekly data are lumped together in Figure 2 regardless of whether that week contains a holiday or not. It is obvious that a week containing a holiday (Memorial Day, Independence Day, Labor Day, for example) should consume less electricity than a normal week.
Weather forecasting technique has been advancing in recent years. Currently, short range dynamic forecasts (defined as up to 7 days here) are quite accurate, particularly for temperature. Monthly and seasonal forecasts by statistical technique are also making great progress. The accurate temperature forecasts make our electricity demand forecast possible.
At Climaton Research Co., we have conducted extensive research into the issue of electricity demand forecast. Detailed prediction models have been developed for various geographical regions of the country and for different seasons. These models are constantly being refined by ongoing research.
Major steps in our electricity demand forecast include:

Predict daily high and low temperature for each of the next 7 days for every point in a 55 km x 55 km grid covering the entire United States.
For each region where electricity demand is to be predicted, compute the population-weighted temperature using our population database which has the same 55 km x 55 km spatial resolution.
From the many of our proprietary T-E (Temperature to Electricity) models, select the appropriate one depending on which geographical location the forecast area is located and what season the forecast period is in.
The selected model will predict the electricity demand. This procedure applies to anywhere in the continental U.S. so that we can predict for both a small area and the whole country.

References
Climaton Research Co., 1996: Use of climate forecasts in the electric and gas utility industry: opportunities for cost control and profit enhancement through informed decision-making. Technical Report.
Climaton Research Co., 1996: Weather and energy consumption: Models and their application. Technical Report.
Climaton Research Co., 1997: Quantitative relationships between electric consumption and ambient temperature. Technical Report.
LeComte, D. M., and Warren H., 1981: Modeling the impact of summer temperatures on national electricity consumption. Journal of Applied Meteorology, 20, 1415-1419.
Quayle, R. G., and H. F. Diaz, 1980: Heating degree day applied to residential heating energy consumption. Journal of Applied Meteorology, 19, 241-246.
Segal, M., H. Shafir, M. Mandel, P. Alpert, and Y. Balmor, 1992: Climatic-related evaluation of the summer peak-hours' electric load in Israel. Journal of Applied Meteorology, 31, 1492-1498.
Warren H. and S. LeDuc, 1981: Impact of climate on energy sector in economic analysis. Journal of Applied Meteorology, 20, 1431-1439

US | Regional | State | Local | Point | Science | Q&A | Objectives | Tutorial | About | Subscribe