Image: Itron

Load forecasting needs to account for Covid-19 mitigation policies, says Itron’s Frank Monforte.

However complicated load forecasting has been in the past, it is now becoming a whole lot more challenging, thanks to the Covid-19 pandemic.

That was the message from Frank Monforte, director of forecasting solutions, in his session at Itron Utility Week 2020.

Lockdown, or ‘shelter-in-place’ policies have changed usage patterns and will continue to do so as these change and as new patterns of life emerge.

Related articles:
Rise in renewables makes smart utility load forecasting vital
Xcel Energy selects partner for grid load forecasting

“We want to reduce the errors in day ahead load forecasting, which may be consistently wrong with stay-at-home orders and we need to adjust the models to account for that,” says Monforte.

As examples of the impacts of stay-at-home, estimates for North America during April and May are that approximately 7% of the load was lost due primarily to the closure of businesses. By August the load levels appeared to be back to the expected levels.

“Partly this was due to businesses starting to work again but there are weather effects that haven’t been completely accounted for in our expectation,” he says.

“What we are seeing is that with maybe more people at home than previously there is more potential for growth in weather sensitive loads as we are adding in this residential layer, which wasn’t there previously.”

Monforte explains that to account for this effect, a direct modelling approach has been followed – similar to how behind the meter solar PV is addressed – with an interactive term in the base forecast equations to reflect lockdown either as a binary variable or a trend variable.

“The trick to this is in the interaction and to take the trend variable and interact it with the day of the week to allow for any differences such as those seen between weekdays and weekends.”

Other interactions should include the weather and solar PV variables and potentially others.

An alternative approach is to perform day ahead forecast error adjustment.

Looking ahead, Monforte says the issue to watch is the proportion of the workforce that continues to work from home, which could result in the higher than expected weather sensitive loads.

“The potential for a new ‘normal’ could lead to a paradigm shift in load consumption patterns,” he concludes.