Data access matters: New Illinois study reveals six unique patterns of residential energy use
How much electricity do we use at different times of day? How does it vary across residences in a utility service area? Are there any trends or patterns for different demographics or locations?
It’s important we answer these questions to design effective rate structures — and target efficiency and demand-response programs — to reduce costs and facilitate a cleaner grid. And because peak use is a driver of system costs, a better understanding of how peak use varies is also important for designing rates that provide equity.
That’s why the Illinois Citizens Utility Board (CUB) conducted a comprehensive analysis of energy-use data for 2.5 million utility customers — and authors Jeff Zethmayr (the consumer group’s Director of Research) and Ramandeep Makhija (CUB’s Data Scientist) uncovered some intriguing results. As we learned in a recent interview with CUB Executive Director David Kolata, the data revealed six unique summer load shapes for residential customers in the region — suggesting some important takeaways for utility rate and program design.
First, tell us about the data you analyzed — where does it come from and how is privacy protected?
Dave Kolata: The energy-use data we studied is collected through Advanced Metering Infrastructure, or AMI — meters that measure and record usage frequently, such as every 30 minutes. Most utilities don’t allow easy access to customer energy-use data, or even any access at all. That’s a problem, because it makes it difficult for anyone to be a check or an advisor on what a power company says about the customer impacts of a proposal. Here in Illinois, though, our utility regulators approved a plan in 2017 that provides anonymous smart meter data to researchers and other third parties.
The way that privacy is protected is that a customer’s location cannot be included if there are 15 or fewer customers in a given area, or if a customer represents 15 percent or more of the load in that area. This “15/15 rule” ensures no one can ever determine the identity of any customer in the data set. For our study, we looked at hourly or half-hourly energy-use data for 2.5 million customers across the ComEd and Ameren service territories, grouped by zip code (or by municipality in less populated areas). We studied summer usage — June through September 2018.
Why is this detailed data from AMI so important — what’s wrong with just using averages?
Dave Kolata: Many power companies design rates based on an average of utility customers’ energy use. But we know households are not a monolith. There are some that use more or less energy than others, and at different times of day, and those variables in turn correlate with different levels of cost impact on the system. So charging customers as if they were a monolith could potentially produce inequity, such as a situation where customers with a lower cost impact are paying more than their fair share and subsidizing higher cost-impact customers.
We’ve long suspected that certain groups of utility customers — in particular lower income customers — don’t neatly fall into average use patterns. These are households with individuals who may be elderly, underemployed or working third shifts, and are home more during the day. Their homes may be smaller, without air conditioning units, or use less energy than larger homes. But without access to actual real-world data, we can’t test these assumptions, nor will we have solid evidence to back our arguments before utility commissions.
What did you find in your analysis?
Dave Kolata: When we looked at last summer’s energy use by time of day, location and demographics, we found that households fell into one of six different patterns, or “clusters” as we call them in our study. Over the course of the day, some were “peakier” than others. Some were flatter. Some have peaks occurring earlier in the day, some later. And the clusters tend to correlate with factors such as income level, age, education and geographic location.
Cluster 1 on the results graph is the largest group, with about 27 percent of all customers in the study. This group’s load shape last summer was most similar to ComEd’s typical system-wide load shape, with a significant late-afternoon-to-evening peak. Customers in this grouping are more likely to be higher-income and less likely to live in high-density urban areas.
Households in Cluster 2, about 16 percent of the total, use electricity very differently than those in Cluster 1. This grouping used a lot less energy overall last summer, and usage didn’t fluctuate much over time, creating a flat — instead of peaking — pattern overall. There are two overlapping groups in this cluster: younger, urban apartment dwellers and low-income households. The demographics of Cluster 4, about 11 percent of the total, are similar to those of Cluster 2, again low-use overall but with a late peak suggesting residences where everyone is away during the day.
Clusters 3 and 5 are similar to Cluster 1, with a significant peak-use period, but now with differences in when the peak time occurs exactly. The peak for Cluster 3, about 19 percent of customers, occurs in late afternoon. These customers are more likely to be older and living in suburban or exurban areas. The early peak and relatively high use suggests that these are larger homes where residents, such as retirees, are home during the day. For Cluster 5, about 17 percent of customers, peak use occurred several hours later in the early evening, with a small morning peak as well. This grouping likely consists of households where residents are at work during the day.
Cluster 6 is the smallest group, about 8 percent of customers overall. This one stands out for having low overall energy use. Customers in this segment are the most likely to live in very small residences outside of Chicago, to hold an advanced degree and to be higher income. Interestingly, they are also likely to use electricity for heat — a finding that we will examine in future studies. Like Cluster 5, there’s a morning and later evening peak, indicating a population that is away during the day.
What did you conclude about energy use among low-income households?
Dave Kolata: It was really clear: low-income households were more likely to use less energy and to have a flatter load shape. Cluster 2, which is the flattest load shape on the graph, is the single most likely pattern for low-income households. And then Cluster 4, with low use overall, is the next most likely pattern for low-income households. Customers in both these clusters tend to live in densely populated areas of Chicago. It’s particularly important for ComEd, where we found that more than half of the utility’s low-income customers have a flat, low-use pattern.
What implications does this study have for rate and program design in Illinois?
Dave Kolata: There are many interesting possibilities here. You can look at these load shapes and then model and test how customers with those shapes would do on a particular time-of-use or dynamic pricing rate structure. It should be helpful in achieving the desired benefits for customers as well as the system. And then this data may also help with designing and targeting programs. For example, our findings indicate that in addition to the importance of energy efficiency and demand response investment in urban areas, there is also significant potential benefit for expansion of these programs in suburban and exurban areas where we see energy use patterns with large afternoon and evening peaks.
Another important takeaway is that this data should encourage utilities and regulators to explore a wider offering of dynamic rate designs that can accurately reflect customers’ cost of service. Since we’re finding that most low-income households are not contributing to the high peak demand that drives a lot of system costs, it’s important to make sure that there isn’t cross-subsidization taking place — that lower-income residents aren’t paying more than their fair share of system costs. This is especially important when we consider the fact that low-income households already shoulder higher energy burdens, having to put a higher proportion of their income towards paying energy bills. In future studies we will also include marginal emissions data to examine and quantify the expected environmental value from changes in load shape.
What do you hope consumer advocates take away from this study?
Dave Kolata: Part of our work here is to show what can be done and encourage other advocates and states to push for and try to establish similar data access rules like we have in Illinois. The nation could benefit from using real data, not just averages or intuition, to answer the questions about energy use that are so critical to inform rate design and other key policies. We shouldn’t just have to rely on what utilities tell us — we should have the data ourselves. Every utility that has AMI should make the data accessible to others to do this type of research. We would like to see NASUCA, NARUC and EEI team up and come up with a standard plan. No matter what your ultimate position is or who you work for, everyone can agree real data can only help us do our jobs better.
“The nation could benefit from using real data, not just averages or intuition, to answer the questions about energy use that are so critical to inform rate design and other key policies. We shouldn’t just have to rely on what utilities tell us.”
— David Kolata, Illinois Citizens Utility Board
David Kolata and Jeff Zethmayr are presenting the full results of this study at the NASUCA Mid-Year Meeting June 19–22 in Portland, Oregon. Download a copy of the study here.