Forecasting Gas Consumption - What to Do about the Weather

September 2010

• For most bundled customers with weather sensitive loads, forecasting their natural gas consumption for the upcoming contract year becomes a bit of a guessing game.
• Some utilities adjust or 'normalize' a customer's recent 12 months worth of consumption for weather, others don't.
• Using non-normalized consumption data as the basis for forecasting consumption can result in increased load balancing requirements throughout the year. However, adjusting the forecast for weather requires an assessment of how much of a customer's load is actually weather sensitive and what portion is baseload.

It's that time again when customers need to settle on a natural gas consumption forecast for their upcoming contract year. For most bundled customers with weather sensitive loads, this becomes a bit of a guessing game. They ask themselves, 'Should I simply use last year's actuals or should I use my $7.99 Nostradamus crystal ball to predict the weather for next winter?'

The gas utilities usually provide customers with their most recent 12 months of consumption. However, simply using the utility's recent consumption data for the customer from the utility may not be the best method for predicting the next year's consumption. While some utilities adjust or 'normalize' the recent 12 months worth of consumption for weather, others do not. Using non-normalized consumption data can result in increased load balancing requirements throughout the year.

Utilities that perform weather normalization for consumption data rely on degree day data from a government body such as Environment Canada. Environment Canada collects temperature data from several locations around the country and uses it to calculate degree days. Degree days are calculated by subtracting the mean daily temperature, in degrees Celsius, from a standard of 18 degrees Celsius. For example, -10 degrees Celsius represents 28 degree days (18 minus -10).

The actual number of degree days for the period September 2009 to June 2010, as collected at Toronto's Pearson Airport, was 3,431. Environment Canada's 30-year (1971 to 2000) average of degree days for the same period and location is 4,039. Thus for the September 2009 to June 2010 period, the Toronto area was 15% warmer than 'normal'.

If customers accept the utilities' non-normalized data as their forecasted consumption for the upcoming year and a normal winter occurs, then these customers would have significantly under forecast their consumption.

Since utilities determine the customer's daily supply requirements (mean daily volume or daily contract quantity) by dividing the annual forecast consumption by 365, experiencing a normal winter would result in a mismatch between supply and requirements. Under these circumstances, customers would need to purchase additional supply over the year in order to balance their requirements.

To avoid or at least reduce the need to perform additional load balancing, customers can gross-up their forecasted consumption to a level that reflects normal weather. As indicated above, the Toronto area experienced 15% fewer degree days last year when compared to the 30-year average. However, this doesn't mean that customers should simply gross up their last year's consumption by the same 15%.

Most weather sensitive loads such as those associated with small to medium bundled customers have a portion of their load that is heat sensitive (e.g., furnace or boiler) and a portion that is a baseload (e.g., water heater load that is essentially constant throughout the year). It is the heat sensitive portion of the consumption forecast that should be grossed up. Grossing up your entire consumption forecast by 8-10% would likely achieve an increase in the heat sensitive portion of the consumption forecast of around 15%.

5 Simple Steps to Effective Energy Buying Read more »