How utilities are using AI to predict energy consumption during grid uncertainty

Challenged by the widening of economic and social shutdowns, the coronavirus pandemic has upended the utility industry’s understanding of energy consumption and highlighted vulnerabilities within utility ecosystems. Residential loads have increased as more people stay home, while demand across the commercial sector dropped drastically, leaving future grid patterns wildly unpredictable. With an urgent need for a more resilient approach to grid management, utilities are fast-tracking the adoption of artificial intelligence (AI) to better understand these changes in grid loads and make smarter business decisions. 

Investments in digital strategies have been underway in the utility industry for years, though progress has been relatively slow. While advanced metering infrastructure (AMI) paved the way for utilities to extract deep layers of energy data from customers, much of it still remained underutilized.

Now, progressive utilities are applying AI to untapped smart meter data to retrieve granular, appliance-level consumption insights for each customer in their territory. These insights are improving business intelligence utility-wide, and especially during the current climate, they are helping them evaluate the impact of extreme load shifts on the grid. With this knowledge, utilities have the power to analyze consumption behavior within individual households and transform it into actionable intelligence. 

AI Powers Deeper Customer Insights

The greater understanding a utility has of each customer’s energy usage, the more accurately it can predict what the new grid will look like to execute upgrades or load shifts accordingly.

Take residential consumption, for example. Unsurprisingly, today people are using more energy at home. But what exactly does that look like? Without diving into the specific ways and times consumers use energy, this information offers limited value to a utility’s ability to manage the grid. 

A comparison of one utility’s residential energy consumption pre-shelter-in-place to post-shelter-in-place highlighted below shows an increase in energy usage post-shelter-in-place, particularly in the morning hours, on both weekdays and weekends. On weekdays, there is a constant increase in load during the morning hours in overall consumption of ~95W/hours from 10am to 5pm after the lockdown was initiated on the 15th of March 2020.

Image 1
Average total consumption by time of day with HVAC load removed to better see the difference in base consumption due to customers staying at home and not due to temperature changes.

Through AI and data analysis, utilities are able to look into the specific factors contributing to this increase at the appliance level for a 360-degree view of a household’s energy behavior. For example, the water heating loads have shifted as people are remaining in their homes as evidenced by the image below.

Disaggregated water heating load for a sample utility compared between the same period in 2019 and 2020.

The combined analysis of leading IOUs shows increased consumption for household cooking, laundry, water heating, and overall base load over the last few months, and this correlated to more customers seeking insights related to high consumption appliances. 

Aggregation of engagement metrics across several utilities showing increase in engagement since start of shelter-in-place guidelines.

Accelerating Digital DSM Opportunities  

The unpredictability of this new grid has also made it difficult for utilities to determine the future of demand side management (DSM) strategies for grid balancing. Deeper, more accurate consumption insights make it possible to not only kickstart DSM programs in this challenging environment, but to accelerate them.

With AI, appliance-level data can be categorized for every customer across 9+ groups, including heating, cooling, lighting, refrigeration, water heating, pool pump, EV, solar and plug load; energy usage by appliance (in kWh, therms, CCF, cost); and time of use by appliance (load profile, peak usage, off-peak usage, duration, etc.). This opens up the potential to cost-effectively find customers who are the best fit for a program, by appliance. This way, a utility that struggles with big peak usage can target customers who are contributing to the peak for a certain appliance and offer an incentive to immediately impact grid management. Or, they can use AI to scan for customers who have smart thermostats and recruit them as ideal candidates for demand response controls programs.

Let’s look again at overall residential consumption. As a result of increased demand, savings are down across the board and more customers are looking to comprehend their higher consumption rates and higher bills.

At a large Midwest IOU, savings are down by almost 45 percent during the month of April 2020 compared to April 2019.

As some utilities place moratoriums on shutting off power to financially distressed customers during this COVID-19 period, they now have an effective way to help these customers conserve energy and reduce their bills. Appliance-level consumption data gives utilities the ability to more accurately identify demographic groups based on appliance usage, region, and customer class.

For low to medium income households, the group most vulnerable to economic hardship, utilities can provide bills broken down by appliance usage, sign them up for budget alerts that notify them if they exceed their budget, and provide a bill projection for what their total bill will be before the end of the bill cycle.

Adjusting to an Accelerated Rate of Change

Load planning and enhanced customer segmentation are just two ways utilities are leveraging AI and data analytics to adjust their business models to keep pace with this accelerating rate of grid changes. The expanded reach of energy consumption insights, made capable by AI, is allowing them to engage with a much broader range of customer segments and advance functionalities across their organizations.

Using AI to build resilience against sudden grid imbalances, like those caused by the coronavirus pandemic, is a timely solution for utilities that will permanently impact how they deploy digital strategies long after the immediacy of the pandemic passes.

Previous articleLearning from the COVID-19 pandemic’s first act
Next articleThe urgent need to modernize the US electric grid and how buildings can play a role
Abhay Gupta founded Bidgely with the mission of leveraging data to transform the utility industry. As CEO, Abhay has led the company from concept to market leadership. Prior to Bidgely, Abhay worked at a combination of energy and technology companies including Grid Net, Echelon and Sun Microsystems. He holds a B. Tech from the Indian Institute of Technology Delhi, M.S. from University of Southern California and M.B.A. from Santa Clara University.

No posts to display