Utilities are investing lots of money in software and platforms to analyze the data streaming in from smart meters, grid sensors and other energy assets. But they may not be as tuned in to the data that can help predict how their customers are going to use the energy they deliver -- or how to start offering new products and services that expand that relationship.
Trove Predictive Data Science says its “data fusion” technology can help fill in those gaps between the utility and consumer sides of the equation. Last week, the company announced it has raised $8.4 million from investors including two new strategic partners -- Spokane, Wash.-based utility Avista and smart metering giant Itron -- interested in putting its capabilities to use.
Trove is already working with Avista, along with utilities including Oklahoma Gas & Electric, Southern California Edison, and an unnamed “large Southeastern utility,” on projects ranging from demand response program marketing and outreach to load forecasting based on consumer data.
Trove CEO Isaias Sudit says the company’s specialty is in pulling together hundreds of sources of data on utility customers, including property records, department of motor vehicle records, government information, and consumer and demographic information. It then merges all that data with the data coming from utility smart meters and puts it to use to solve specific utility business challenges.
“We see a lot of companies that say they have analytics that are rendering data, or providing visualization tools for that data,” he said in a Monday interview. “We see ourselves as a data science company.” The difference is, “We go and ask what questions the utility needs to answer, to help them with operations, to help them with products and services they provide, to help them with load forecasting in the future, to help them with financial forecasting.”
“The other part is really understanding the customer,” he said. “We spent about two years working with key data vendors in the industry to provide us with a very robust database of consumer data that comes from the residential and commercial side of the equation.”
These are the same kinds of promises being held out by a long list of data analytics providers, of course, ranging from giants like Oracle, IBM, SAS, Teradata, EMC and SAP, to startups such as C3 Energy and AutoGrid, to name a few. Likewise, most of the big smart meter companies are building analytics capabilities to expand the value of the data they’re collecting, or acquiring startups to help out, as Sensus did with Verdeeco, or as Silver Spring Networks has done with Detectent.
Sudit laid out some key differentiators in the way Trove goes about its business. First, it layers its analytics capabilities on top of existing utility data warehouse systems -- “very robust, very scalable high-throughput systems” from the likes of Teradata, Oracle or EMC Greenplum, he said. That does limit its addressable market to larger utilities that have invested in such systems, as opposed to companies that are promising to provide those core capabilities on their own.
Second, it focuses specifically on customer data, he said. For example, one of Trove’s earliest utility customers, OG&E, used the company’s expertise to help it determine which customers were the best targets for its smart-meter and smart thermostat-based demand response program. But while Trove was able to help OG&E achieve a double-digit percentage improvement in the uptake of this DR offering compared to traditional utility marketing approaches, it isn’t involved in the analytics that help optimize the ongoing operations of that program -- that’s being done by Silver Spring Networks and AutoGrid.
But that consumer data can provide critical insight into core utility operations that energy data or smart meter data alone might not be able to, he said. Take Trove’s ongoing work with Avista on load forecasting, or predicting the future energy consumption of individual customers in aggregate to guide the utility’s purchasing of natural gas and fuel to generate electricity.
Utilities could simply use their smart meter data to predict future load and consumption, he said. But that would leave out a lot of variables, such as whether a household has bought new appliances, brought new children into the home, or changed ownership over the past few months. Weather data obviously has a big impact on energy consumption as well, as does data on whether a household has bought an electric vehicle or rooftopsolarpanels.
Predicting which customers are most likely to make such purchases in the future is also an important part of this equation, he noted. “Every single value that we have, every single response that we bring to the utility, has what is called a 'confidence level' that allows you to understand how confident I am that that answer is correct,” he said. That applies to load forecasting and financial forecasting, as well as some of the emerging efforts among utilities to offer customers new products and services beyond their traditional purview, such as community solar or electric vehicle charging.
“We have an exciting project for Southern California Edison for new product development, identifying what are the best products and what are the best approaches,” he said by way of example. While he wouldn’t discuss details on the project, which is set to be unveiled this spring, he did say that “in a generic way, a lot of these big utilities are starting to offer hundreds of products to end customers, from CFLs and air filters” via traditional utility efficiency incentive programs, to price-responsive smart thermostats or smart-inverter-enabled solar panels.
As for its core technology, Trove, formerly known as GridGlo, has previously raised $2.3 million from lead investor CUBRC, a Buffalo, N.Y.-based nonprofit research organization with roots working with NASA and the aerospace industry. “They have a unique technology for what they call 'data fusion' -- the mashing of different data sources to find patterns, to come up with some kind of pattern recognition you can use to discover things,” he said. “We repurposed a lot of that data science platform to be useful for the utility industry space.”
Trove’s future work with Itron is aimed at marrying the Liberty Lake, Wash.-based company’s expertise in smart meter and grid data with Trove’s consumer data expertise, he said. “They saw there was not a lot of overlap on the stuff they were doing and the stuff we were doing,” he said. While he wouldn’t discuss what specific projects the two may be working on at present, “We have a lot of interesting plans over the next 12 months,” he said.