Electric Load Forecasting Project Example

Large View

Historically, as an independent load balancing authority, Dominion was responsible for supplying its electric load obligation with its own power generation and purchased energy. If forecasts of customer demand were too low, additional generators could be turned on or more megawatts purchased from outside of the company to supply the load. If forecasts were too high, generators could either be reduced or shut down, or the megawatts sold off-system. The company had a high degree of control for adjusting generation to offset any forecast error.

When Dominion joined PJM Transmission – a regional transmission organization that coordinates the movement of electricity – it became part of a larger market with rules on buying electric load and selling generation. Under this new model, Dominion lost the flexibility to use its own generation to offset inaccurate load forecasts, thus driving the company to refine its forecasting process.

Prior to joining PJM in May 2005, the primary focus was on having a fairly accurate forecast of the daily peak hour demand. The only penalty Dominion incurred from missing that forecast was using more fuel or less fuel (coal, oil, natural gas, etc.) than it should have used. But in PJM, without the flexibility to use its own generation to compensate for any forecasting error, the penalty is much higher.

If Dominion purchases the forecasted megawatts for each hour in the PJM day-ahead market and if the actual load comes in higher or lower, then the company has to buy or sell the megawatts based on real-time pricing. The difference in hourly pricing between the day-ahead market and real-time market can be substantial. From May to December 2005, the load forecasting errors were costing the company somewhere between $3.5 million to $4 million each month.

Management wanted to see an error reduction quickly. An accurate forecast of the amount of megawatts customers will use each hour throughout the entire day was essential. A Six Sigma project team made up of individuals with defined roles in the current load forecasting process was assembled. The project scope encompassed all load forecasting processes and procedures; no area was off limits.