The following is a viewpoint from Uttara Sivaram, who managed product deployments for Bidgely Corp. with major utilities across Canada and the United States.
This month, Californian electric power utilities are moving forward to better understand how their low-income customers use energy, following a mandate from state regulators. By harnessing the data from nearly ubiquitous smart meters in California and partnering with data science companies, these utilities will actually be able to disaggregate their customers’ electricity consumption down to the level of individual appliances. Doing so could enable utilities to help their customers understand previously inscrutable electric bills and to suggest concrete ways for customers to reduce those bills.
This opportunity isn’t limited to California. I’ve led appliance-level disaggregation projects with utility partners across the United States and Canada at Bidgely, a company that uses data science to help utilities and their customers better manage energy consumption. A consistent theme across all the projects I’ve led has been that utilities are eager to use disaggregation to promote energy efficiency as well as to boost customer satisfaction.
Other approaches struggle to achieve both goals. For example, a utility that simply tells its customers that they used more energy than their neighbors might nudge them to use less energy. But it would also irritate those same customers, who are now motivated to save energy with limited guidance on how to do so. Normative messaging is a powerful tool, but when applied in the absence of effective solutions or support, studies show that this messaging can detract from the overall customer satisfaction with the utility. On the other hand, if customers can see on their bill not just that their energy use is inefficient but also that the culprit may be an outdated pool pump, they then feel empowered to reduce their energy costs and are more satisfied with the utility’s service as a result.
Still, utilities will have to invest attention and resources to successfully roll out disaggregation and realize its full benefits. The most successful deployments have combined data science expertise with a high-level commitment from utilities to provide quality data and treat disaggregation as a core component of their customer engagement strategy. Based on the successes and challenges associated with these deployments, here are three recommendations I’d make to utilities considering end-use disaggregation:
1. Invest in advanced metering infrastructure.
The more granular the data from a customer’s meter, the more accurately data science can break down that customer’s consumption into the various appliance loads. With data from smart meters that remotely report customer electricity usage every 60 minutes, disaggregation can reliably separate the signatures from heating and cooling equipment, pool pumps, and devices such as televisions that stay plugged in 24/7. With 15- to 30-minute resolution data, algorithms can detect additional devices, such as electric water heaters, lighting, and refrigerators. And to complete the “pie”, the system can use regional survey and audit data to estimate the remaining usage (entertainment, laundry, and other small electronics).
To be sure, disaggregating all those appliances is still possible without the highest-resolution smart meters. Rule-based data science models can break out the same appliances in homes that don’t have a smart meter at all. By studying usage patterns in similar regions where smart meters are installed, a deep-learning system can deduce how a customer’s usage breaks out into these individual appliance categories, even based on a single usage measurement every month.
But the fact remains that the most accurate disaggregation is derived from the most granular meter data. So utilities contemplating advanced metering infrastructure should take into account the benefits of high-resolution smart meters when it comes to disaggregation. Regulators should take note as well—smart meters are well worth the upfront investment by enabling customers to pinpoint their savings opportunities beyond an educated guess and lower their monthly bills.
2. Integrate disaggregation with other customer engagement strategies
Rather than treating disaggregation as a standalone service, utilities should incorporate it into a broader strategy to provide seamless customer service. In one recent project, for example, we collaborated with a utility partner to create a mobile app where utility customers could not only check their disaggregated appliance usage but also access other useful services such as a map of electricity outages. In fact, during one widespread power outage, we discovered that the number of customers engaging with the mobile app spiked as customers sought out the outage map. This one-time increase in engagement translated into a sustained increase as customers discovered that they could also track their disaggregated electricity usage every month.
Indeed, disaggregation can enable a utility to enrich its relationship with customers over multiple channels. Traditionally, a utility might send a monthly or bi-monthly paper bill to customers. Customers would then be left to blindly adjust their electricity consumption across all of their devices and cross their fingers that one or two months later, their bill would be lower. But based on insights from appliance disaggregation, utilities can now send out emails to customers in the middle of their billing cycle, projecting their energy use weeks in advance. What’s more, customers can use the utility’s mobile and web applications on any day in the billing cycle for real-time feedback on their efforts to adjust consumption.
3. Invest in data integrity and data quality processes
Utilities are required by regulators to prove the results of their programs. This often involves complex measurement and verification (M&V) projects involving multiple parties. Such projects require running an often elaborate experiment, in which only some customers (the “treatment group”) get to see their usage broken down by appliance to see if they actually use less energy than other customers (the “control group”), who do not get disaggregated bills.
But running a tightly-designed M&V exercise over thousands or millions of customers, many of whom will change addresses, opt for a different rate plan, or occasionally miss their monthly payment, is no simple task. Here, it’s crucial to define the most efficient and secure way to share this information before the program launches, and each entity—the utility, the technology provider, and the measurement and verification (M&V) firm—must do its part to preserve the integrity of the data for which it is responsible.
The utility has full visibility into its customer base, and is best suited to maintain the database that is consistently updated to reflect any changes to customer data. The technology provider typically performs customer engagement via the technology platform, which is a key indicator of who is still eligible to be a part of the savings analysis.
I've seen the best results when the task of merging these data streams—actual energy consumption data, customer billing activity, program participation—and bundling it up to share with the M&V firm is best led by the utility and its considerable IT resources, which are typically tasked with collecting and monitoring customer usage data as part of normal business. What’s more, creating a single home for this data reduces the likelihood of “leaky” or inconsistent data, which can jeopardize the integrity of the analysis and siphon unnecessary time and resources to restore.
Ultimately, these three recommendations all reflect the same overarching lesson:the most successful disaggregation deployment projects are those where utilities are centrally involved in the procurement, preparation, and management of the consumer data that is central to this exciting technology.
Uttara Sivaram has managed product deployments for Bidgely Corp. with major utilities across Canada and the United States.