![Turning customer data into grid intelligence [partner content] — Open Circuit cover](https://megaphone.imgix.net/podcasts/7ec68a0c-e24b-11ef-b277-7b3d3c3c73b9/image/796dd136afa5782eae8ad683dd946957.png?ixlib=rails-4.3.1&max-w=3000&max-h=3000&fit=crop&auto=format,compress)
Loading summary
Transcript29 lines
- [00:03]
A
This is partner content from Latitude Studios. Long island is exactly what the name suggests. A very long, narrow island stretching out from New York City. And that geography, narrow weather, exposed and heavily residential, makes it unique for delivering power.
- [00:21]
B
We're out there sticking out into the Atlantic, very prone to weather. So as hurricanes come up the east coast, we are the bullseye. And we often find ourselves in that hurricane prep mode.
- [00:32]
C
It's a beautiful territory. But I will say, between the hurricanes and then snow in the winter and hot, wet summers, I mean, it's, you know, there are a few weeks where it's really paradise. And then, you know, the rest of it, you got to run to the beach. I bet Lou thinks it's paradise all the time.
- [00:47]
B
Most times.
- [00:50]
A
Lou Dabrino has been working in the power business on Long island for more than 30 years. As a VP of Customer Operations for PSCG Long island, he oversees every part of the utility that connects to the customer.
- [01:02]
B
Really encompasses everything from the call center to our metering, our billing, our energy efficiency and renewables marketing.
- [01:09]
A
It's a territory where homes and small businesses matter far more than heavy industry and where customer behavior has a direct impact on the grid.
- [01:17]
B
From that perspective, it certainly has been, you know, just looking to keep up with customer expectations around experiences that they have with other companies.
- [01:26]
A
And as electrification spreads across Long island, the utility is facing the same pressures confronting grids everywhere. With rising demand and rising delivery costs.
- [01:35]
C
I think we're seeing the same kinds of challenges around addressing affordability, handling load growth, and electrification in every single region of the world and every single region in the United States.
- [01:45]
A
Ted Nielsen is the Chief Product Officer at Bijley. Bijely is a vertical AI platform that's worked with over 45 global utilities and energy providers. So as pressure on the system continued to build, Lou and his team started looking for new ways to understand customer behavior and to guide it to help save energy. So they turn to Bitgley.
- [02:05]
C
I think what makes PSAG Long island special as a customer is kind of the willingness to play ball. So in some ways they are perfectly average, and in other ways, they're incredibly, incredibly extraordinary.
- [02:19]
A
Utilities everywhere are searching for new ways to manage rising demand and ease the growing pressure on customers and the grid. And increasingly, they're turning to purpose built AI tools that can pinpoint where the grid is under stress and how customers are using energy and where support is needed most. Through conversations with utility leaders and Bijley experts, this series explores how those tools are being applied in practice and what they reveal about the future of the grid. In this episode, our second in the series, Stephen Lacy talks with Lou Debrino of PSCG Long island and Ted Nielsen of Bidgley. They discuss how PSCG Long island rolled out time of day rates to hundreds of thousands of customers and and how detailed appliance level energy use data gave the utility new tools to manage affordability, load growth and grid stress. Many utilities have increasingly turned to time of day rates as a way to better align electricity prices with the actual cost of producing and delivering power. But the success of those programs depends on more than the rate design itself. Customers need to understand how their behavior affects their bill, and utilities need better tools to guide them. For, for PSEG Long island, that meant turning raw meter data into something customers could actually act on.
- [03:35]
C
So, Lou, this brings us to time of day rates. When did you start exploring variable pricing for customers and why?
- [03:44]
B
So we know during certain times of day, certain times of the year, it does cost us more money to either procure or produce that electricity. So with these rates, it gives us the ability to more accurately reflect the cost of that energy to coincide with when customers are using it. We have had pilot programs with voluntary time of day rates in the past, but this was really focused on moving all 900 plus thousand eligible residential customers onto this rate. What are we going to do to make sure that our customers can be successful on these rates? Really, that gets back to this partnership with bigili in terms of how could we work with them, partner with Biggley, to really collaborate, put information in the hands of our customers that gave them valuable insight around how they were using their energy, where they were using it, when they were using it, and really what were some actionable things that they could do to really benefit or be more successful under these rates? Right. What load could they shift?
- [04:48]
C
Yeah. So Ted, tell me about how bigly came into the picture and what you're doing to identify those opportunities. Yeah, and I'm actually going to steal a great analogy from Lou and his team. I think if you think about a world without Bidgley, sort of like you just get a bill in the mail and someone says, you owe me 100 bucks. And there's no detail on it. With our true machine learning disaggregation, it turns it into a credit card statement so that you can see each of the charges and each of the appliances that are actually driving that $100 bill. And you can see then importantly, when you decide to roll, say, onto a time of the day rate, what specific appliances during specific times have specific consequences. On your bill that allows you to have a much more data driven, informed conversation that's tied directly back to the user behavior. We have close to two dozen different patents in what we would describe as machine learning disaggregation, which allows us to find, detect and accurately describe the appliance signals in the existing usage data so that we can confidently, with very high accuracy, help people understand what their specific usage is. The moment you get there, the moment that you have people understanding what the actual appliances are, all of a sudden you can reach out to the specific customer with specific actions and specific interventions about their specific behavior.
- [06:15]
B
Yeah, absolutely. And just having visibility of this data, time of day rates or not, it means a lot to our customers. As Ted said, just to get a bill and it doesn't give you any appreciation for, you know, where is that money going versus the credit card statement that says, well, $30 is for cooling, $20 is for lighting. You know, what it does is it really gives some perspective to the customer and I think it helps them get a better appreciation for, you know, where their energy dollar is going. What am I getting for this?
- [06:43]
C
And I think this is such a critical point too, because you may not remember, hey, three weeks ago, it was actually really, really hot, so I was running my AC a bunch in order to keep my house comfortable. So all of a sudden you can not only help people understand, well, this much of your bill was attributable to H Vac, but then you can begin to show this longitudinal data to say, and by the way, it's because the weather was actually unusually warm and so you were using it more than you normally do. So Ted Begley's been at this for a long time. What have been the biggest technological changes in disaggregation that has given you even more insight into customer behavior? Yeah, so I think there's really two of them. Bijley actually got its start 15 years ago with clamps. So there was actually physical devices that you had to put around the meter and to get really, really high resolution signal off of the meter. Well, that's really hard to roll out to a million meters in Long island or kind of anywhere. And so we took all of the learning that we had from that really, really high resolution data and applied it to what's called AMI 1.0. And so this turned out used to be kind of monthly meter reads where someone would drive around and look at the little spinner on your meter and write down this is how much energy you used this month. And you'd get the bill into 15 minute increments that sort of go back. You kind of think about it almost like over cell phone, over 3G comes back to the utilities. So they get a feed of every 15 minutes how much energy is being consumed. We found the same signals and took the learning that we had from that really, really high resolution data. We took those same learnings and applied them to the 15 minute AMI data, which didn't require clamps or installs or people to go around and service equipment and be able to roll that out for kind of full territory deployments like we see here at PSEG and at close to 45 utilities around the world. There's another phase that's coming which is actually in sort of classic horseshoe theory all the way back to the beginning, which is really, really high resolution data on the meter. So as compute is getting cheaper, there's going to be increased levels of actual software that you can run on the meter itself. So not just say in the last month you charged your car during peak hours. So when peak hours are coming up, I might send you a reminder to say like, hey, maybe don't plug your car in if you want to save this Money. With AMI 2.0, where you have disaggregation running live on the meter, when a person plugs in their car, you can send a push alert to their phone right now saying like, hey, did you mean to plug your car in? You're on peak hours. So all of a sudden I talked about interventions earlier, this idea of how do you craft a really excellent customer experience with AIMI 2.0 and we have close to 2 million meters that we're working with major meter manufacturers on across the United States. This is going to allow increasingly digital, increasingly real time interactions with the customer base. And so, Lou, you have used this work to expand what you're doing, whether it be alerts to EV charging programs, broader customer engagement. Tell me about the evolution of what those programs are that you built off of.
- [09:56]
B
Yeah, so certainly, you know, I know we mentioned AMI and that was the first step, right. It was having AMI deployed and we did finish that up. By the end of 2020, we had rolled out AMI to all of our residential customers. Right. So it did enable us to now use this technology. So as Ted had said, no longer do we have a meter reader driving around reading a meter, but we were collecting data every 15 minutes. So tremendous amount of data. So we started slow. We started with various pilot programs where we tested really just the receptivity to the information we wanted to Prove its accuracy, customers understanding of the information, the value that they saw in it, kind of draw a little bit of a parallel. Prior to this, we would send out a paper report that basically compared your energy usage to your neighbor and it really didn't give you a lot of information other than in many cases it might anger a customer to say, well, why are they using less? Or in many cases it just wasn't like for like comparisons, right? People had different house sizes, different number of occupants, et cetera, et cetera. So more questions came out of it. With this data, it really allows us to be very customer specific. We're working off their 15 minute interval data, we're seeing their usage patterns, we have demographic data relative to their house size, et cetera. So we can then. And what we've been doing is really finding different ways to use this data to inform and educate our customers. So when they get their bills, they see not only their bill, they see, as we said, we see this breakdown of where they're using their energy. We focus on their top three uses, we give them insights as to what's the next best thing they could do to maybe lower their energy cost and so on and so forth. It also becomes a great tool for our customer representative. So when a customer calls into the call center, if it's a high bill or if they have questions about their bill, our customer reps at the call center now have the same access to that same information that the customer has, and they're able then to have an educated conversation or more educated conversation with that customer. So it does sort of translate them into more of an energy advisor. At that point, they're looking at real data, they're having an honest conversation about their energy use and really trying to work with that customer to come up with a solution to either have them have a better understanding of what's driving their bill, or what they might be able to do from that. We have the ability to couple them with programs that we offer in our efficiency. So whether it's programs where we offer rebates for efficient equipment, or we look to enroll customers or enlist customers in a peak load reduction program where we will pay them an incentive if they're able to curtail usage during certain peak times. This makes it very easy for us to target market those customers. We're not going out to the masses, we're going out to the people that have air conditioners or pool pumps or the types of things that we would look to control during peak times. So it does make us more effective from A marketing perspective. We've used it in cases of identifying electric vehicles to know who has electric vehicles, as Ted had alluded to earlier, and what information can we send to them. Photovoltaic PV customers. Similarly, we're able to identify our PV customers and then again respond with them and give them information that's very specific to them that they can action.
- [13:28]
C
On one hand, you have the problem of spray and pray, which is let's go out mass market and dollar for dollar, you're spending marketing dollars on getting a message in front of people and educating people that it may not apply to. So when you can target not just who has a specific kind of appliance, but we'll take H Vac as a good example because I think weatherization is, especially when you think electric heating is a huge potential demand on the grid year over year. You can compare using Bidgley data. Not only does this person represent, say X percent of the demand on particular grid assets, but you can then see, and it's getting worse, like whether they need a mechanic to come out and fix their H Vac or they need to change their filter. But you can tell from the data for a given house, for a house of that size with say the insulation rating that's on that house, this is actually not an efficient H Vac or it's getting worse year over year. And so that means that when you're trying to spend your marketing dollars, you can actually spend that reaching out to those specific people who are going to have the biggest impact on the demand side management program. And Lou also mentioned solar. This is a tricky one. I think solar on Long island is a very different story than say solar in San Diego, where San Diego can kind of assume that it's probably going to be sunny every single day. And so you can take the capacity of the solar panel and sort of assume that that's going to be a permanent reduction in the grid. But in a place like Long island with hurricanes and snow and everything else, it changes over the course of the year. The amount of energy that gets produced and sent back to the grid changes over the course of the year and how much home demand those solar panels can actually offset. And that's invisible. It's invisible to the utility because it all happens behind the meter. And so with the kind of algorithms that we have where we can detect the presence of solar, estimate the capacity of the panel, back that into the usage of the home, that actually allows then the utility to know and do active balancing with what's available.
- [15:37]
B
And in certain Cases, you can use that data to potentially offset investment in a substation or on a particular feeder, or at least delay it for some period of time to say, hey, there's another way to go about that potentially from a customer side. So we've looked at different ways to do that as well. And as I said, that type of information that's now available really enables us to at least go through the analysis to see where that might make sense.
- [16:04]
C
And I think we also think about the grid and sort of the grid problem on one hand, super easy to diagnose. You can kind of look at it and pretty quickly see like, oh, these assets are under stress or this is when they're under stress. Bijley's data then, because it exists on a premise level, on a meter level, can then help you understand why it's under stress. Is it because of EV growth or is it because of H Vac? Well, if it's H Vac, people kind of want their homes to be comfortable. So you might only be able to move 2% of that load. If it's an EV. Well, you can probably move that around and figure that out. So I think the level of insight about what is driving growth, what behaviors are driving the growth, and then that it's like sort of a water pump without a handle. Without these programs, you don't know exactly how to get the juice out. And so all of a sudden in those places where you do have interventional points where a non wireless alternative is going to be effective, then all of a sudden you have the data, you need to have an intelligent conversation with your regulator to say, hey, we looked at this and in fact there isn't a lot of movable load. I don't think a demand side management strategy is going to be the, the right play here. I think we actually need wires, whether it's batteries or additional assets. On the other side you can say, hey, and by the way, these are areas where we can try demand side management. And we think that this could translate to this kind of deferral. And you can't do that if you're just looking at a pile of electrons. You can only really do that when you understand the appliances that are behind that. So what you've described is a long term partnership where you've grown capabilities together. Ted, what does it mean in practice to grow alongside a utility partner in this way? I think I mentioned this earlier when I said P Sec. Long island in some ways is a very standard utility in terms of composition and size and some problems and I think extraordinary in the sense that the willingness to play ball and I think when we look at both, we've spent a lot of time talking about machine learning and the power of AI in signal detection and signal intelligence around finding appliances and attributes about those appliances and propensities for people to purchase an electric electric vehicle or adopt solar. The level of personalization that we now can get to, especially as Lou was talking about more people adopting say digital channels to interact with the utility. So this is things like using Genai to help even more in the call center around call deferrals and increasing levels of personalization in those alerts and messages so that people can then take specific steps and you can have more of a conversation and turn today what's a message about that credit card statement with you spent $35 at Starbucks. Did you mean to to take that conversation that right now is one way and actually turn it into a dialogue to actually have someone interact with and give and deliver what is effectively white glove service around home energy optimization. What if planning around well what if I switch to a new rate or what if I I want to retire my Toyota Camry and switch over to a Tesla? What's that going to do to my energy bill? That's a conversation you can now have using Genai and then the levels of automation that are possible with agentic AI in terms of not just doing all of the research because again bijely helps put that data to work, put that AMI data and turn it into real signal intelligence. You can flip the problem around instead of well, now I have this incredible resource that I need to dig through and find the insights. You can now have agents go find those insights for you and make recommendations. Hey, you have an opportunity to roll out a weatherization program. You have an opportunity to connect better with your income qualified customers and serve them better because your programs aren't performing quite as well as you thought they would be. So you can kind of almost flip the narrative a little bit and become more goal driven. So I think that level of building a partnership that is based on trust, that's based on good performance, that's based on helping improve the reality of the relationship between the utility and the ratepayer. And made possible by our tools and our tools doing well means that the relationship can continue and can continue to grow. So in this moment of load growth is a moment of massive pressure but unique opportunities for utilities. What have you both learned that can apply to other utilities who are having to get increasingly creative about opening up new grid capacity keeping rates down and using the grid more effectively. So I'll start by saying I think there's two themes that I would kind of pull out from today's conversation. The first one is it's very hard to intervene strategically if you don't actually understand the problems. If you're looking at a giant pile of electrons, it's very difficult to know what you can do next. The only thing you can do is throw more hardware at it and eventually that's going to hit your ratepayer. I think the second thing, and this is from what Lou has been saying, is this is not just a utility problem. This is only going to be solved with the utility and the customer kind of tightly working hand in hand. And that takes real trust. And trust is something that you cannot just assume. It's something that you have to build, it's something that you have to cultivate. It's a lot like a plant. It requires care and watering and feeding and paying attention to and making sure that when you go away on vacation it's going to get watered. And so I think trust through information and transparency and kind of having their back and actually caring about their problems is the only way that you're going to really get that grid.
- [21:55]
B
Consumer partnership I just would reinforce what Ted said. I think that trust factor is very important. We want our customers to trust that the information we're giving them is accurate. And we want them to trust that we're doing everything we can to help from an affordability perspective and trust that we are really taking the steps as we move forward to consider all options and all costs so that we come out with the best solution to serve our customers.
- [22:25]
C
Lou debrino, a pleasure speaking to you. Thanks so much.
- [22:28]
B
Thank you. Appreciate the time.
- [22:30]
C
Ted Nielsen I enjoyed it.
- [22:31]
B
Thanks.
- [22:32]
C
Awesome.
- [22:37]
A
Ted Nielsen is the Chief Product Officer at Bitgley and Lou Debrino is the VP of Customer Operations at PSEG Long Island. This is the second episode in a four part series exploring how utilities are using purpose built AI to tackle some of their most pressing challenges. You can go back in the feed and listen to the first episode published June 16 and stay tuned for episodes three and four. And if you want to learn more about how Bijely gives utilities unique behind the meter insights with its utility AI platform, visit bijely.com or click the link in the show notes.