Updated: Aug 27
COVID-19 has changed our lives at an unprecedented pace and scale. Companies who would never consider working from home are embracing it, retail is online only, all schooling is happening online and delivery services are how more and more people are receiving their essentials. From pig farmers in Nevada whose business relies on Casinos, to Battery Systems for Energy Storage Supply Chains, everything is affected by this pandemic in one way or another. The impact of this pandemic has reached into every corner of our lives and although many of the changes will be short term, their long term effect cannot be ignored.
All of these changes have a massive impact on how electricity is consumed in Ontario - this makes peak predictions more challenging than ever before. Ontario's electricity demand is affected by many factors, and economic activity is one of the most important. 80% of Ontario's exports go to the United States and 50% of imports come from there as well¹. So the impact of the US economy on Ontario electricity demand is massive as well. When making long term demand forecasts for Ontario electricity, the IESO uses a variety of inputs, including: Growth in the Residential, Commercial, Industrial and the Agricultural Sectors, Electric Vehicle growth, and Transit electrification. They also consider Energy Efficiency programs like SaveOnEnergy, Codes and Standards like EnergyStar and the effect of the Industrial Conservation Initiative.
COVID-19 affects every single one of those drivers. Understanding the interplay of these factors and their short term and long term ramifications is a herculean task. Projections of growth during normal times are somewhat reliable as the changes come in slowly and predictably - with COVID-19, historical models are out the window. This will make peak predictions more difficult than ever before.
Change #1 - Overall Reduction in Demand - potential for a strong rebound in late summer
The effect of the pandemic on the Top 5 ICI peaks for this season will be profound. The impact that is currently being seen on the Ontario grid is due to slightly reduced manufacturing demand and significant reductions in commercial, retail and institutional (high schools, universities and libraries) demand. Most of this electricity demand will bounce back relatively quickly once the peak of the pandemic passes. All signs from Italy and Spain point to a strong possibility that rebound will coincide with typical summer peak periods which start in July, making setting peak thresholds difficult.
Change #2 - Flattening of Residential Demand due to Paused Time of Use Pricing
The province has suspended time-of-use rate for 45 days. This means small businesses and farms will be paying the off-peak rate for electricity. This is already having a significant impact on the load profile of the province and flatting peaks.
Change #3 - Soft Economic Demand from the US
This change has short term and long term ramifications. Most anecdotal evidence suggests minimal interruptions to supply chains which feed the US, however the long term impacts of supply chain interruptions remain to be seen. One thing that's for certain is that the demand will continue to change on a daily and weekly basis.
We can already see huge changes in demand
The current impact on the Ontario grid has already been profound. By the IESO's own measure, a 5-17% reduction of peak demand has been observed in all hours.
As various businesses continue to cope and adapt in different ways, historical load projections are being further challenged to give accurate results. For example, see the IESO's Ontario Demand projection for April 1, 2020 below.
A difference of 914MW between projected and actual at Hour Ending 8 on April 1, 2020
The best way to predict peaks has historically been with deep learning AI - this is even more true now as real-time projections become increasingly important. For example, a projection made by Edgecom Energy's pTrack™ model on March 23, 2020 can be compared to the IESO projection for the same time below. The pTrack™ Deep Learning model is clearly adjusting to the new reality quickly and effectively by taking real time factors into account and modifying the projections.
Conclusion: Accurate peak predictions require an AI Deep Learning model that updates regularly with dynamically changing data.
Edgecom Energy's pTrack™ Deep Learning AI model uses economic factors, weather and 15 other data sources to predict the dynamically changing peaks. This model is re-updated every 5 minutes as new data is available. With game theory learning of curtailment efforts of other ICI participants, it also adjusts the threshold at which curtailments begin based on what is happening right now, not what happened in the past.
Predicting peaks will be more difficult than ever this summer. If you're participating in ICI and looking to accurately predict peaks, pTrack™ is your best option.