Electricity Demand Forecasting - IFs and Buts

These Days Electricity Forecasting Tools are available which can forecast electricity in the Short term horizon to a tune of 95% Accuracy.

A very simple way of Forecasting electricity demand is to keep a track on the performance of Demand over the past week. Demand profile in the coming monday would be similar as the demand profile for last Monday subject to both of the days are Normal Days and not event days.

 

However above method of Forecasting Demand would be viable if the Historical demand is predictable. Predictability can be evaluated by considering the Profile of Demand for several Mondays of the past. If the profile of last several Mondays looks similar then the profile of a Monday is predictable and can be fruitfully utilized for the Future Mondays.

Having evaluated the Demand Forecast for Normal days, We are sure to encounter Certain dates which will not be normal days but event days. Event Days can either be recurring events as Holi, Diwali, Independence Day or non-recurring as Rainfall, Drought, Strike. Event Days are most likely to encounter fluctuations in Demand / deviations from a Normal Day. Predicting Demand on Recurring event Days is possible if we can track the Profile of Demand during those days in the past. This is a challenge in itself; especially when we have never considered our meter data as a vital source of information, all we retain is the Consumptions of a month, which will help us in billing our consumers.    In cases of Data unavailability the resort is to categorize similar events and hence appropriate assumptions can be made for a specific event.

Above Forecasting Method is a crude approach of Electricity Demand Forecasting wherein we look into the performance of Historical Demand to arrive at Forecasts for the Future. However there are several things which we have ignored such as the impact of Weather Changes, the urbanisation, use of energy intensive technologies, the ever increasing demand supply gap, Government policies to provide uninterrupted power to people. All of these and many others have a significant role to play.

This leads us to another question as to how do we factor in so many predictors or drivers of electricity demand? How do we assign weights to these? Will it suffice to assign 20 % weightage to Weather changes, 10 % weightage to Econometric Changes? What will be the underlying criteria to these weights? Is there a justifiable reason as to why a specific weight is assigned to a specific predictor? How long will these selected weights hold good in predicting Demand Accurately? What should be the revision frequency and am I sufficiently capable to assign weights. These and several other questions will haunt a forecaster.

There have been several researches in the field of Statistics and Econometric Studies which tend to establish a relation between the Predictors and the Predicted and assign appropriate weights.

 

For some more interesting read please refer to:

 

Ghosh, S. 2002. “Electricity consumption and economic growth in India.”Energy Policy 30, 125-129.

 

Saab, S., Badr, E., and Nasr, G. 2001. “ Univariate modeling and forecasting of energy consumption: the case of electricity in Lebanon.” Energy 26, 1-14. 

Add comment


Security code
Refresh

Additional information