Global Minima

Follow Global Minima
Share on
Copy link to clipboard

Global Minima features interviews targeted at the intersection of bits, watts, and dollars. Hosted by Dr. Jason S. Trager of Sustainabilist, the podcast explore how corporations, governments, and society as a whole can utilize data to minimize consumption while maximizing business returns. With guests across the academic, corporate and government sectors, Global Minima sheds light on the convergence of data, distributed energy resources, and demand-side management.

Sustainabilist

  • Nov 7, 2020 LATEST EPISODE
  • monthly NEW EPISODES
  • 46m AVG DURATION
  • 9 EPISODES


Search for episodes from Global Minima with a specific topic:

Latest episodes from Global Minima

Sandra Kwak on How Renewables Can Improve Energy Access

Play Episode Listen Later Nov 7, 2020 34:36


10Power CEO Sandra Kwak joins Global Minima to talk about how she uses data in her effort to improve energy access in underserved communities across the world. Topics covered include the falling prices of storage, how solar can help provide clean water, and her work with the Foundation for Climate Restoration.

Mia Oppelstrup on the Appeal of EVs

Play Episode Listen Later Oct 5, 2020 36:51


Mia Oppelstrup is really into electric cars. Lucky for her, she works at EV charging station company Volta. In this episode, we delve into trends in EV usage, how Volta decides how many stations to put in and when to add more, and why EV stations are a sound investment for retail centers looking to drive consumer traffic.

Lindsay Baker on Smart Buildings and Sci Fi

Play Episode Listen Later Aug 10, 2020


Lindsay Baker’s passion centers at the intersection of humans and their work environments. As the former VP and Head of Sustainability at We Work and current board member of The Clean Fight, SPUR and Measurabl, she has a lot of thoughts on how workspaces can be both more sustainable and more livable. In this episode, we ask Lindsay about her recent blog on upskilling building managers on efficiency tools, static versus dynamic sustainability ratings for buildings, her work with The Clean Fight, embodied energy in buildings, how the COVID-19 crisis might be the thing that pushes building owners to execute efficiency improvements, and her latest Sci Fi reads.

Danny Wilson on the Global Future of IoT

Play Episode Listen Later Jul 2, 2020 41:48


Matt Golden on the Death of Efficiency and the Rise of Demand Flexibility

Play Episode Listen Later Jun 1, 2020 55:37


Matt Golden:My official title is CEO of recurve. My unofficial title is energy efficiency agitator. I'm been at this quite awhile and what sounds like cynicism is actually positivity wrapped in the belief that we just need to evolve this entire industry. So started out actually as a contractor doing energy efficiency and actual buildings and at a fairly large scale, you know, we had over a hundred people retrofitted four or 5,000 buildings and mostly in the San Francisco Bay area, mostly residential and learned a bunch of things about utility programs and energy efficiency that has informed kind of, or up to that company was called recurve recurve 1.0 so learn that in deemed savings models where customers get rebates in advance and the industry doesn't have any accountability to outcomes, it becomes a race to the bottom.And the more the poor work quality you do, the more money you make. And so learned as a company doing building science, you know, an HVAC and shell and renewables all integrated with engineers and the whole nine that are subcontractors doing crap work. We're actually making better margins than we were. And so this is an existential problem in our industry where there's a complete misalignment of incentives and the utility programs that show up to give rebates to customers actually increase the cost of customer to the, to the marketplace and against them. This reverse signal where the cheaper you can actually deploy your stuff on the front end, the more money you make. So that was one aspect. And the other thing we learned is that everybody talks about energy efficiency and being the first fuel and this great resource, but it does not behave like one and it's definitely not valued like one.So how do we actually extract, you know, if we believe there's all this pent up resource in buildings, how do we come up with a way to pay for it, which has to be either drastically different rate structures, you know, real time rates and they're going to be three or four X what people are used to paying or the alternative, which is direction where we had an is how do you value what we now call demand flexibility, which is like energy efficiencies. New friend. It's a much more, I'll talk more about that, but a, a much more inclusive terminology. But how do we value the flexibility we're creating and all these buildings as an actual grid resource, not just a bunch of white papers or you know, sign on letters. But how do we actually integrate flexibility into the resource mix and get its full true value so that it'll accelerate?Can we create, define grid flexibility a little bit more and dive into charitable job introducing myself? Yeah, we can go back to that. Let me, let me, let me ask you, so, so fast forward to the, to, to the my a new gig and what we're talking about today, which was recurve 2.0, which is a totally separate company actually. But the name sticks because it actually makes a lot more sense in this new iteration, you know, was an attempt to actually build a scalable building, retrofitting and doing energy efficiency in buildings and discovered that all the cards are stacked against us. Iteration 2.0 recurve is attacking those problems and it's attacking them. Really kind of the two ways I described. First of all moving towards market-based approaches, but even in current, you know, traditional cost-effectiveness based programs, how do we use data to align incentives properly to target the right customers to do all sorts of things that yield better results at a lower price.And then secondarily, now that we have AMI and this thing called the duck curve I was actually just doing some math this year and last year, 2019 or like five or six years ahead of the traditional duck curve, 4,300 megawatt hours worse than we thought. So how do we look at flexibility behind the meter? And I will describe what that means in a moment relative to this new problem and value as such. Because when you know, when and where it's happening, which is what we get with AMI data, it's worth a whole heck of a lot more than saving a kilowatt hour because it matters and when and where it's occurring. And in fact the game has changed. And I think before we even started, I led in by saying efficiency is dead and that's true. Kilowatt hours can in many, many parts of this country right now not save any carbon and actually cost ratepayers money. So we are at a point where reducing consumption is not always a net good. So it's time to pivot.Jason S. Trager, Ph.D.:It's like a good t-shirt slogan, "efficiency is dead."Matt Golden:But if you get it followed up by long live efficiency just now flexibility, long live demand flexibility.Jason S. Trager, Ph.D.:Okay. All right. So we've made tee shirts. All right, go on.Matt Golden:'Cause It cause cause the imporant part is the traditional approach is dead, but it's actually more valuable than ever and all we have to do is pivot.Jason S. Trager, Ph.D.:And, and so how does, how do these now differ from kind of traditional ancillary services such as efficiency, right? How, how has this grid flexibility pivot, gone happen?Matt Golden:So the primary kind of pivot for efficiency is recognizing that with AMI data, where it exists we can measure when we are impacting demand and we know where it's located. And those are kind of the primary ingredients for an actual resource, right? Carbon, P & D capacity, energy, all of all of these major grid and climate values are totally locational and temporarily constraint, right? So that's the first big pivot is that we can measure it. And when we talk about demand flexibility, it is inherently on a time and locational basis. But it's just a more inclusive term. So many of the things that we think of as energy efficiency, also do demand response and you know, a storage, you know, a, a smart thermostat can predictably load, shape every day, right? But can also be dispatched. Same with a storage system in the garage or an EV by the way.And then there's other things like insulation, right? There's nothing, there is no such thing as a flat impact, a load shape. It really doesn't exist. Everything creates a shape change. It's just a function of measuring it, right? So it's valuing the fact that insulating an attic in a building with a heat pump will reduce demand at seven in the morning on a winter time in the winter. And because we're now creating a winter peaking grid, and that's incredibly valuable against the avoided cost of winter time, renewables and storage versus an average KWH savings or versus commercial lighting, which peaks at 3:00 PM savings, right? Which is exactly when our over jetting solar and has very little to no value. So it's a much more inclusive approach. And it says we actually don't care what, well, first of all, the premise is that the grid actually goes to the meter.That's where the forecast comes from, the meter. So, and then behind the meter is my house or, or a building. I happen to live in a house. I happen to own it. I know that everything in my house is mine. If I choose to operate in such a way that it helps PG&E, that's my choice. They're not going to reach in and touch it. It's not part of their grid. It's my toaster oven. Right. My wires, my end uses. So the core premise says that the grid extends to the meter and then there's all the stuff behind the meter. So demand flexibility is looking at whatever it is you did behind the meter, whether it's insulation in the attic, IOT on the wall, storage in the garage, Evie being charged at the right times, a behavior campaign, whatever that is. How does it express as flexibility at the meter?That's change in demand. And then the real only question there is, is at that point it can be, it's kind of the unified field theory of flexibility. We don't care what it is, we just care what, how it manifests on the question. The only remaining question then, and this is something else that we can actually measure, is what is, what are you responding to? A longterm predictable signal. For example, with PG&E they're paying one X 19 hours a day and then five times as much during the summer peak ramp period for example. But then they do and they'll pay that, you know, every weekday in the summer or you know, years in these contracts. It's longterm predictable. There also can be events. And so can you break and we can, and we break the can we break what the response is? Is it a longterm capacity, energy efficiency type of a signal or is it responsive to a more real time market event?And that's just a function of who, how you get paid. There's drastically fewer events on the grid, but they're much more valuable. And it's important to have this kind of harmony because right now if you're in a demand response program, you have a distinct and specific incentive not to do energy efficiency at all because it actually reduces or affects your baseline in such a way to reduce your dr payments. So we can create a harmonized signal. So this is really just a macro strategy that says buildings are complicated, customers are complicated, business models are extremely complicated. We're going to create competitive markets that can engage customers and sell them and install and deploy a whole host of solutions that will have benefits that accrue to the customer, which they're going to pay for like lower bills and comfort, but also have benefits that accrued to the T and D utility, the load serving entity, climate change goals, things that are not individual customer benefits. We're going to aggregate those separately and pay for them. Like we're building a distributed power plant. In fact, exactly like we're fulfilling a distributor.Jason S. Trager, Ph.D.:I think it's interesting because few minutes ago you were speaking about the, the location and how these, these resources are different in various locations. And one thing that has in grinding my gears for a few years is that deemed savings, relying on manuals, the, the technical reference manuals that the utilities put out, conflict with each other all over the place. I don't think any, any single technical technical reference manual is the same. And so I'm wondering how, you know, after harmonizing the calculations and harmonizing the, the way that we're pivoting energy efficiency, how would you view us harmonizing how the different regions in the United States interact with these resources? Like what is different?Matt Golden:One is that this is actually a paradigm shift. You know, energy efficiency is unique and different than any other market. So our core, our core kind of philosophy is make efficiency and flexibility work like everything else. So the first thing to isolate floor is that like, we were the different thing, which is a function of like not having any data over the years. And a system that is based on attribution to measure, which is not math, it's magical models, right? So there's a fundamental structural change that does have to happen. This is not just purely iterative. We do need to actually pivot cause right now we're essentially stuck in purgatory between this fallacy that is we're going to model every end use and be able to determine what the impact is going to be. It's incredibly inaccurate but more importantly it's working on averages and the real world does not work on average with some get back to that.But moving from that to an empirical model where we can measure everything all of the time we have AMI data, we have cloud computing. It's an entirely different approach and it says, look, we're not going to, it's not just insulation and 90% efficient furnaces anymore. This is way more complicated than that, which is awesome news, right? Like the list of amazing new technologies that we're dealing with. I don't know which ones work by the way. Nobody does. We have to test. But like the amount of innovation that's going on around grid 2.0 and flexibility on the business model side is amazing. It defies work papers. PUC is actually said that here in California that they realized like, what is it they want to run a smart thermostat program. There is no TRM value for a smart thermostat. They're black boxes on the wall that get updated regularly.A smart thermostat is a thermostat with a wifi chip. How it's controlled is what makes it smart and nobody knows how these things are controlled. And we've seen some thermostats that work really well and some that make the bills go up and they all look the same from the outside. So being able to move to a performance based approach really unlocks the innovation in the marketplace, but it changes the accounting system behind it. So the biggest, there are so many with deemed savings. It's a little hard to even start, but the one that stands out to me, I mean truthfully, they almost always overestimate impacts and they conflict with each other even within the same tier and manual, right? Like we have conflict conflicts where, you know, what we want is aggregators to go into a building and do a whole host of things. Solar plus storage with an Ecobee thermostat and insulation in the attic and maybe a heat pump.And these things all work together to impact the grid in a positive way and you just will never get that with a prescriptive deep approach. And so averages, however, is what's killing us. So, so nothing in the world, I mean this is like, this makes me, this is, this is probably the most, so we can make any program in the country. Just take the traditional thing, forget pay for performance, market based demand flexibility for half a second. If you take any program in the country, first of all, you have to get off in order to make it cost effective. Step one is getting off of deemed because it traps you a deemed a deemed a deemed value in a TRM manual says the average, whatever it is, high efficient heat pump, air smart thermostat. The average customer saves let's say 6% and it's fixed in advance.So therefore nothing you do to improve outcomes can move that number. So it leaves you with one option, which is reduced costs at all costs at all costs. Right? And you've no accountability to this predetermined number. So if you want to get better outcomes, the only thing you can do is cut, cut, cut, cut, cut. Your costs, which results in shitty, shitty results is one of the reasons it always overestimates is there was no incentive to do quality work. So cut costs at all, expense at all costs, and then deploy as much as that crap as you possibly can. If that burger stand problem, right? If you lose money and every hamburger don't open more burger stands, right? That's what we're doing. And that's deemed there is no such thing as the average, right? Like you don't get, you don't get average. You've got a distribution.So if you want to make a program cost-effective, you move off of beam to meter. And what that unlocks is now you have two things you can do, improve your yields and deploy more stuff and reduce your costs at the same time. And by unlocking that, the ability to actually capture the value through increasing yields, you can change the name. So we can make any program in the country cost-effective, pretty much overnight. If you just do like literally three things and you don't even have to do them very well. It's so powerful first, right? You get off of meter. Dean, did you move to metered outcomes? What you actually deliver, not the average from some past study. The first thing you do is you target the customers with a high potential, the safe. So if you had a distribution you're going to find in most cases, and we find specifically that there will be 10% of your customers that you're serving that are saving three or four times X on their bills with drastically fewer negatives in that pool.They're much, the customers are super happy, they're cash flowing and have actually had an HVAC problem. They're more likely to say yes, but they're also worth three or four times as much to the grid and GHGs and we know who they are in advance. So step one, we don't go after everybody that moves, right? We're not going to try to sell an iWatch to my mom who doesn't have an iPhone, right? Like we're going to find the people who actually need the product and are going to benefit and be more likely to say yes. And because there's so much more valuable to the grid and climate, we give them a much better deal targeting high potential first step, second step, optimize your program around actually having a KPI now around performance. Do things like send your leads to the contractors that actually deliver good outcomes.Maybe kick the really bad ones out or give them technical assistance. Just base. If you do those, actually those three things, you can make any cost, any program cost effective pretty much overnight. It's really that simple and it's pretty much my malpractice to go after a market and say we're going to sell the laggards and early adopters at the exact same time. Like that's just not how real markets work. And that's the, that is the biggest trap of deep, right? It's key, you know, there's no way out of that. The only way out is cut costs again. And that's this and that creates this cycle of, Oh shit, we didn't actually hit our deem numbers. I wonder why? Well, maybe it's because we cut her costs.Jason S. Trager, Ph.D.:Yeah. And so recurve I mean it used to be Open EE and then was rebranded. As you've been going into these new programs, right? And part of your lifeblood is this transparency. How have you communicated this transparency to folks who aren't going to read the code but could potentially do so and can look at your models? How, how, how does that come inside?Matt Golden:So we'll talk a little bit about like what, let me, let me talk a little bit about what these things are and then we'll talk about why it's important and why a short will get it. So I'm actually holding and because we're not on video here, but I just ate a weights and measures, which is a 500 gram counterweight, right? Every market. You can't actually have a market you can't trade if you don't have an agreement on what it is you're trading. Is that a frequent problem? So far? Funnily funny. It's actually wasn't on my desk for proper reasons. I just kind of like tossing it around cause it feels good but I'm the problem every single market before you can transact has to have an agreement on what it is and that's doubly triply true with something like demand flexibility, which is a calculated value, right?Energy efficiency is a derivative. You have to calculate a baseline and it turns out that calculation is exceptionally insensitive, insensitive for all sorts of things you're doing. Like the models are one thing, but like my own house I made the advanced home upgrade program here in California. If I pick the nearest weather station based on geographic distance, I ended up with Napa Valley and I get 15% gas savings. It's hot. They're not like Mill Valley where I live, it's close. It's like two valleys over. If I pick the CalTRACK math, this will describe why you might trust this. You pick nearest weather station and climate zone and I get SFO and I get 23% savings. Who's right now that's even a -- So there's hundreds of these knobs. And my joke, honestly you could give me just about any data set you want. How I turn the knobs, I could give you just about any answer you want back.And because there's no standard, everybody's right. And that's a terrible recipe for a transaction. So the U S Congress has been in the constitution, has the power to establish weights and measures. And there was a point in time in America when every state had their own definition of a bushel, of corn. We got up, we got rid of that because you can't trade across state lines and everybody's bushel is different. So at the most fundamental level, and like this is the, this was the kind of aha coming out of Recurve 1.0 saying I want to be paid on performance. Let me figure out the business model. I'll take the performance risk. And then we scratch your head and say, well, what the heck are we actually trading? And what is performance if I don't know how I'm getting paid, that's not risk. That's uncertainty. So this sounds easy.Let's just figure out what it is. Now. 10 years later we have it. So there's two things we've been working on and they started initially in California. Someday they'll get rebranded, but there's, it's like written in legislation all over the world now. So it's very hard to rebrand. It turns out. But we built something called CalTRACK, which was created by the PUC and the CEC and initially all the IOUs in California funded it. And ironically, you know, the irony is that well there's a lot of irony, but they hired me to figure it out and I more or less convinced everybody that if we're going to look at the performance of deemed values or the, or be able to calibrate models, you need a source of truth. And we do happen to have these things on the side of a building. You might've seen them before, they're called meters, right?And so maybe there's truth in the meter. And so that moved us towards getting everyone around this idea that like using meter data, especially as we have interval data we could actually calibrate our models and true everything up to reality. And that was where like the entire notion of pay for performance started. And it actually started around calibrating simulation models, but that was never why I was in. So so we started this effort originally, Michael Blasnik, who's now in Nest, who's kind of a grand Poobah of data analytics and energy efficiency has more experienced than any other 10 people I know. And so, so the, the basic tenants is CalTRACK, which is a set of defined reproducible methods in a process with data science behind the choices. And so as you were saying, these are standard models, right? Like we talk about this as like, this is like the McDonald's at M&V.You can create custom bespoke models with 700 outside variables and machine learning, any other crazy stuff you want. And there's a use case for that at times. But I have no idea what that is. If I'm in a transaction with you and it's totally by very nature, as soon as you start creating custom models, it becomes a custom project.Jason S. Trager, Ph.D.:I mean the, the model for that is onsite generation.Matt Golden:Yeah. So as soon as you become a custom project, everybody in the transaction has to figure out why the heck they trust you and they'll you, and most of these situations is highly conflicted because the people doing M&V are typically the recipients of getting paid. So I can't trust anything they say or should I, because they're writing their own check and you get, and so you get mired in this custom engineering process and it's just so expensive. It falls apart. It's not like it's wrong, it's that it's, you can't afford the due diligence and the whole thing blows up.Jason S. Trager, Ph.D.:And so this kind of leads to the, to the point of the models that you're building are more geared for many buildings and being correct over, over the aggregate than for any specific building.Matt Golden:So the virtual power plant is the aggregate. And so, and this is like hard to explain sometimes to engineering folks whose job it is to be accurate and correct on individual assets. I put air quotes around that y'all can't see because you can overfit any model with enough variables and get a good CVR, messy model fit. And that's baloney. Which is the other side of it, right? You can't just, I mean we can make any model look good if you give us enough variables. Fundamentally there's a way, pretty much, so the idea is that we want to have a consistent standard, reproducible, totally transparent and verifiable method for generating a baseline condition on an hourly basis. Cause that's really the fundamental of energy efficiency and flexibility is what are we measuring against? That's the big question. And it's often not thought about, but you changed the building, right? There is nothing you can't, what do you, what is the, what is the counterfactual that you're using? And so we, the CalTRACK methods, what they essentially do not, what they do do actually not essentially is they use preexisting data from like a year prior to a retrofit to train a model. The model is trained on first identifying occupancy patterns using just AMI data. So some of the table stakes and it has to be reproducible to all parties by data that is verifiable for all parties. Right? So or else you're back in the custom conundrum, right? It's not wrong, it's just too expensive to scale. So we generate occupancy conditions based on the AMI data through a bidding process.And then for each month of the year we look at temperature ranges and generate bins and we generate models before using regression for how that building responds to temperature in those bins for each month in each occupancy state. And those models that are used in the performance per period to predict the baseline. And we know good they are. Cause it sounds like, you know, a bunch of mumbo jumbo, but we can do what's called a sample testing. So, you know, let's use 2017 to predict 2018 and buildings that haven't been treated. And we know, and we can do that on every single asset and an entire state for example. And so the key thing, and this is the big thing from finance, from engineering to insurance finance and like how you actually manage risk on a grid is it's not about being perfectly accurate in the individual asset, which was again, not even possible.It's about knowing your probabilities, how, how, how much do you discount? This is dramatically more important than the having perfectly accurate CV or MSCI is at great expense on each asset that can be verified. And so we can, we can actually figure out, because these are standard measurements, what, how much do you discount them for modeling error? And you can take that right off the top and it's way more efficient than this smart. But at the same time, these models are really good and individual assets also. So when we look at some commercial assets that we are tracking and we're tracking a lot of Rezi and commercial. So these are, you know, there's a big thing we hear a lot from engineers is, yeah, yeah, that'll work in Rezi blah blah blah. You need thousands in a portfolio. But my commercial buildings, they're totally different.You need something way more complicated. I mean, you don't, that's not, that's not, we're finding it all. Again, it's a question. So we have a good example is like there is a commercial building portfolio we're tracking in a West coast state excused thousand projects. They're like everything from gas stations to office buildings, et cetera, are kind of all over the map and without any prescreening, just whoever's coming in the front door 90% of the time we can model the baseline, no problem with CalTRACK. And then there's 10% where there's weird shit going on. Right. And it's not usually subtly weird. It was like most of those are like big strange looks. I don't know. One time we found we were actually wondering what it was and we did what to Google maps and it was a pottery studio for kids and it's a kill.Jason S. Trager, Ph.D.:And, and that's, that's really one of the, one of the pieces of the future of portfolio management is that once you are measuring it and predicting it, then you can say, if I can't predict it, well then maybe it's not behaving well.Matt Golden:Yeah. And so there's a lot of things you do with that. First of all, you don't build the entire, you don't build the entire industry around the three or 4% that you can't model. So first of all do everything you possibly can through a portfolio approach because it's so much more efficient and less expensive and scalable. And then the second piece is like we actually for the bulk of that 10% we can't model like 80% or more. We know in advance that we can't model their baseline. So maybe don't focus out of the gate on targeting that 10% and you won't end up with 10% you'll end up with 2% which is like performance. And then with that two or 3% of problem child projects, then you decide do you hire an engineering firm if it's worth it or you just stipulate it. And discount the whole thing is who cares cause it's just brain damage. So folks, let's start with like what is easy and low cost and efficient, which is most things right? And we go the other way in efficiency, which is, let's say there's a 2% edge case, let's spend 110% of our time and effort burden everybody in the market to address the 2%. And we think that's the wrong way though.Jason S. Trager, Ph.D.:And so I, I guess this leads into this next question is, you know, how is the data that you're gathering how has it influenced your view on the policies in, in ways that maybe you wouldn't have expected in the banding and what, what's really grinding your gears these days that you didn't expect before you got into this?Matt Golden:I mean, I don't even know where to start on the amount of things that you've learned and things I truly believe turned out to be not true at all. So, you know, let the data talk. A couple of things, you know, we, you know, I've been a little personally dogmatic around paper performance as a paradigm and I've come to believe that actually we don't have to be because the data does talk. So I'd say that's the biggest thing is like take any program anywhere, start measuring it, immediately start utilizing that data start actually moving again, move from moving from Dean to actual meter is the folks who are learning this and other States, folks that are not on board with like paper performance markets, but they're trapped by Dean major States, very high up on the AC triple E list are running into this brick wall that says like take lighting out that was propping up 90% of the freaking savings at a very high cost effectiveness and nothing pencils and their only escape is get out of deemed right.They got to get, because they have to turn like there's no way. None of, none of this stuff, pencils on average, they've moved through early adopter customers is getting more expensive. They don't meet cost effectiveness. So the boring thing is like we're just, I've moved my own thinking to be like, actually I don't need to be dogmatic around pay for performance because as soon as you become accountable to outcomes and you have the data, then you start looking for efficiencies and it all just happens because it's obvious that if you, if you win by increasing your yields, you start to optimize your portfolio. And if there are third parties and others who can do that cheaper or you win by working with them and like it all happens. So just that's, that's my biggest overall thing. But frankly, the biggest thing I would say is that like, we're not broken anymore.Like, and by the way, I didn't really finish the CalTRACK thing. So we have CalTRACK, we also have opening meter, which is a Python based implementation of those methods, those incredibly specifically defined reproducible methods. So when you put those two things together, right? A set of methods that are, that have data science behind them that are in a process that everybody can reproduce. Right? And if you have a, you don't like how we're doing it, not, it's not us actually wouldn't even run them. They're part of the JDF-joint development foundation. Combined with an open source implementation, which is part of the Linux foundation, the largest open source foundation in the world. What you put those things together, you have a way to measures, you put in a contract, you can say we're going to, we're going to calculate the grid impacts and payments and customers are based on CalTRACK 2.0 and XX bill to the opening meter.And that can be verified and now I have a way to measure it. And that's really like the fundamental missing link underpinning everything else. And with AMI data, like we still do quite a bit with monthly, but monthly doesn't decarbonize the grid, right? It doesn't balance renewables. In fact can do the opposite. Like I mentioned, you can make monthly TRM programs work better with monthly, but they don't actually solve future problems. But as we get this, like as we're using AMI data to do this, then we know what the, we know what the man flexible ability is, what we call the resource curve. It's understood by all parties. It's something that you can actually move out of repair surcharge based programs, which is the Achilles heel of all of this 8 billion a year. It doesn't decarbonize our economy. Sorry. Right? We need much larger scale capital so now we can actually bring flexibility and part of that being energy efficiency into procurement and actually treat it like we're building a virtual power plant and the counterparty is not a regulator with a rate payer surcharge tax. It's actually the procurement team and or it can be multiple. So folks who are actually utilizing flexibility to reduce their market exposure. So let's take an example. If your East Bay community energy, a CCA here in the Bay area, it's true for all CCAs, like they don't want to do efficiency, efficiency reduces load, everyone wants to load to go up. If you're an electric utility, right? So what would they want to do is flatten their load shape. So they can do two things. One is reduced their resource adequacy requirement to buy more capacity by shaving peak, right? Everything on the beard is built around those peaks. And the other thing is to reduce their exposure to energy market volatility, right?When it's 104 degrees out, they don't have capacity and they end up on the CAISO at great expense. So those are the benefits to the CCA. That's a load serving entity, right? At the same time, if you're attached to the right part of EBC grid on a feeder that runs to the diner G power plant, which is a non wires alternative that PJ is trying to shut down this 1970s turbine and the gas turbine in the middle of Oakland, there's T&D value that accrues to PG&E, right? So if you, and so if your East Bay community energy and energy or PG&E, now you can look at the customers and you realize this is not an abstract issue. You realize that, okay, the, it's clear that the peak is being driven, not an abstract, but by, you know, mostly Rez but also commercial HVAC air conditioning, right?That's the peak and it's not, it's not generally happening. It's happening from specific buildings, right? That are, that have the most volatility and the most consumption of energy in the category of air conditioning in those peak windows. Those are the buildings you go address the customer. Those are the ones that have bad AC where the customer's happy and are worth so much to EBCE and also potentially PG&E that you can afford to give them a drastically better deal, meaning there'll be more likely to adopt. And so it's about finding those counterparties and getting off the Dole. You know, we should be using our rate payer surcharge money to do policy things like take the billion dollars and it's not building any more. When I said efficiency's dead, we're losing 300 million of it. And here in California this year because of cost effectiveness problems that we have to solve for.So like when I say is dead, this is happening all over the country. Like we got to wake up and see that it's not just hyperbole like we're losing major budgets because we're not cost effective and more importantly we're not solving the actual problem we're having anymore. The problem is flexibility, not KWH reductions. So so we want to reserve, we should use that rate. Payer surcharge money to achieve policy objectives like a social justice, right? Equity issues. Let's go help poor people with that money. Let's go do some science projects, some research, some early R & D, some amount market transformation. Basically we take total resource cost test in the resource manual and we're taking the R out of it. We're going to monetize the resource where it belongs over in procurement as an actual resource. And then we can use this policy. The small amount, like 8 billion a year.Really not a lot of money spread out across, across the entire country. When you're doing in California, you look at our 700 million a year, it's a couple of dollars per customer per year, right? It doesn't get you very far. So let's eat. But let's use all of that to achieve policy objectives that are not bound by resource costs and, and how does that reshape the investment markets? Right? So, so who comes in and invests in this load shaping who comes in and invest in the flexibility? Oh, all the usual suspects. That's the beautiful thing about this, right? So if you think about the customer, if you think about aligning the interests to the market and you think about aggregators, which we, which we is a catch all term for us, right? So Google, Nest, EcoBee as an aggravator clear result or ICF can be an aggregator. Packetize or OhmConnect can be aggregators.We don't really care. It's anybody who's engaging customers, delivering some set of technology services, whatever that affects load shape and can aggregate those results into a portfolio, right? And then be able to be large enough that you can manage that risk and then sell it to these counter parties. Right? And there's a lot. And what's beautiful about this, it can be big players, but could also be smaller players, right? Like, we can actually let the innovators in. It's, it doesn't have to be like massive, massive, massive, you know, consultancies are the only players like these contracts. We can have multiple players in the same market. You know, all the people that say it'll confuse customers. I always say like, I dunno, you bought a car on a cell phone and in pants, right? I've never heard of it. I've never heard of having choice being bad for customers ever except in the unfinished enough efficiency industry.So I'm looking at you in a video. I'm not sure you are wearing pants, Jason.Jason S. Trager, Ph.D.:I am wearing pants -- I just checked.Matt Golden:You qualify, ok, qualify.Jason S. Trager, Ph.D.:I'm not sure if you are.Matt Golden:That's a good point. Maybe I'm not. Too complicated, choosing pants. So, so what, so, so you let the customer, the customer gets a bunch of benefits. In this model, you have an aggregator in the middle, right? So let the aggregator figure out how to address me as an individual customer. Sell me something that saves me energy and I don't have to get another good example. I have this fancy zoning system because I have this weird house and I want them to be able to like not heat my office all night long. When I'm heating my bedrooms. It's all about comfort. Those are my benefits. I'm paying for that and I'm happy about it. I paid a lot of money for that.Jason S. Trager, Ph.D.:You've talked about that many times. Actually.Matt Golden:It makes me happy. I spent 20 grand making my house the way I wanted it. It's not saving anybody any, it's not saving PG&E any money. It actually completely screws up their cost. Cause my, my first retrofit was $26,000 because I had all these comfort things I wanted to do, which I am perfectly happy about. So let the aggregator figure out how to sell me a solution that optimizes firm for me. Right? And I'm going to pay for it and they're going to still have to, I'm going to still have to finance my end of things, et cetera, whatever paid for it, but I'm only paying for my personal benefits and then aggregate those mutual benefits. And there's a tertiary cashflow between the aggregator and these counter parties, PG&E or East Bay community energy or whomever to pay for those mutual benefits in aggregate as a virtual power plant over time and align the interest.So they're making money by figuring out how to sell solutions to customers they actually want that are simultaneously optimized for customer benefits and grid and climate benefits and that. So everybody gets the best deal was that value engineering exercise. And then you've created a tertiary cashflow between the aggregator and these markets and you have a portfolio. So you don't have to be right on any, in every single individual asset, right? I'm doing car loans, I know what the default rate is, but I don't know if Jason Trager is going to lose his job tomorrow or what. I'm not trying to, I can't figure that out. Right. And that's custom engineering. Can I predict? No, I can't. Right. And if I tried to, it'd be super expensive and I have to call everybody, you know? Right. And instead, instead you have to assess risk in the aggregate, which is way easier to do.And I'm going to tell it like efficiency and flexibility in general in aggregate is extremely predictable. You get these really nice distribution curves. It looks just like an insurance or financing product and that changes. So target people, it changes everything, right? Exactly. Because now your goal is to maximize that and there's a hell of a lot of arbitrage there for aggregators because everybody thinks the average is how this stuff performs, but I'm telling you there's three or 5% of these customers that are six X the average. Let's go find them. Right? Like it's, there's a ton of arbitrage there. We are drastically underestimating performance and overestimating the price because we're not targeting like there's such a huge opportunity. So w more importantly from a power plant standpoint, we create this tertiary cashflow between whomever, like, I don't know, I won't name names, but it can be any one of these aggregators and that cashflow financing and cash flow is how you build infrastructure.It's called project finance. Financing. A project does not project finance. It's financing a cashflow from a project cause not based on the credit worthiness or asset value of the cusp for the customer. It's based on the cashflow that's created over time. And so you can actually, so when it comes to who's going to pay for this, you finance that cashflow. Like you're building a power plant or a solar farm or a storage system and it's the same characters and the money is plentiful and it's cheap, right? So instead of asking customers to use their credit or asset value, which is the most expensive capital on earth to pay for a distributed power plant, that it doesn't benefit them, it benefits everybody mutually. We're going to aggregate that and pay for it. Exactly like the power plant with the same low cost capital mutualizing the risk off of the individual.So we were talking about Amory, he's going to think of beyond some subsequent Joe sell the customer cold beer and warm showers. Right. And like manage that risk on the backend, which is much easily, much more easily done. And all. Again, aligning incentives. Cause I started out talking about at Recurve 1.0 we learned that doing high quality work is no way to make money in energy efficiency, but in this model it absolutely reverses that. Now my now you know everything you do to actually install things correctly and training your customer and your contractors and all of this is now a creative to actually generating outcomes. So profitability gets up with actual performance, right? I want to maximum, I want to figure out where that value engineering is. In the middle you can finance these cash flows like you're building a power plant and that's the basic model. And get away from thinking about the individual technologies. I want Sunrun when they install a storage system with a solar system to look on the wall and say, we're going to drop. You know, if, if you actually use your air conditioner, we'll, we'll give you a free EcoBee and maybe we'll insulate your attic and like, let's start mining this power plant for this potential because it makes everybody money, not because it's some policy objective.Jason S. Trager, Ph.D.:Anda before they go to that house, the have some sort of information about what they're going to offer people based on the meter data.Matt Golden:Right. Or after they've engaged the customer continue to mine it. Cause again, on average it's really hard to do, but if you have 10,000 customers, you know the 5% of those customers using the most HVAC in those peak windows, again, that's a very different story than the average. Right. Those are customers that have like problems and high bills and they're probably uncomfortable and don't really have data on that. But it's a good guest and again, they're worth so much and give them a great deal. Like there's, it's, it's again, it's kind of business model one-on-one when it comes down to it.Jason S. Trager, Ph.D.:Yeah. And just kinda just to start to wrap things up cause you're getting towards the top of the hour here. As you look at this, right, we, we've talked a lot about people who don't have the don't have the economic incentive to engage in energy efficiency during, during our chat here because they're, they're not going to be getting enough savings, right? But there's still energy efficiency that has to be had there, right? We, we, we have to, we have to increase savings all across the boardMatt Golden:At the right time. So not all across the border. Like, yeah, they're not in that good in the middle of the day when we have more solar than we know what to do with in California.Jason S. Trager, Ph.D.:I agree. But what I was trying to say is that how do we, how do we get the utilities to start to segment differently based on this data, right? How do we get the, the, the aggregators and implementers? Because there, as we gathered this live data and look at the grid flexibility, there's now different ways to sell this value. Like you said, we're going to sell you on comfort cause you have the, the economic means to actually engage in that comfort. But then there's these other market segments and I'm wondering what you've seen there. And how you think of that.Matt Golden:I mean, there's some really new that some new, very positive regulation and in California that's heading us in the right direction. So it's about aligning incentives across the board, right? We have to like, again, this deemed model is like the most fundamental problem. Like if everything is just deemed average, nothing works. Right? And actually for us and I, at some point before we close, I wanna explain what we actually do and all of this. But when we go into States where like the regulator and the utility and every week somebody will get this feedback, we don't know what you're talking about. Efficiency works great. We always get our demon sentence. And if everyone's happy with that total, there's not much to be said. But that's more and more that as we move to actual carbon goals, you know, one out of four Americans buy power from a utility that has a zero carbon by 2050 goal today it might even be higher than that given APS.And like we have this list of stuff that's just happened this year. Like I want to tweet out and having a time. There's so much stuff happening. And so it's like changing the game, right? So you got to go through kind of the whole value stack. So we have to, at the high level, we need to get energy efficiency out of these deemed regulated programs and get and encourage utilities to invest in behind the meter flexibility. One is lower than the marginal cost of the alternatives because it's cheaper for rate payers and it's good for customers and that. And so, and then within that there's a whole bunch of above my pay grade stuff like heck, the way we incentivize utilities was meant to encourage them to put pipes in the ground and wires on poles. And that's not actually what we're doing anymore. So like, come on regulators, let's move into a PBR performance-based regulatory approach and let's move the line from build new infrastructure to decarbonize quickly.Or none of this is gonna work that's above my pay grade. And that's happening, but it's above my, you know, that's, I can't fix that one personally at this point. So once you get the utility focused on, as you say, like this is a resource that, you know, in California we have SB three 50 doubling efficiency, which is a conundrum. SB one Oh 100 focusing on zero carbon by 2045 and AB 32 putting a cap on carbon that's decreasing. And so the utilities are, so what's happening is we're onboarding renewables so fast. I mentioned earlier that the Ducker was 4,300 megawatts, but the belly is 4,200 megawatts lower than we expected this year. So that is real pain, right? And so they need, there's not enough storage, there's nothing else to fill that gap. So they, they're also looking around realizing we need behind the meter flexibility or we're in trouble, right?We are overgenerating solar at the wrong time. We are literally paying Oregon and and, and Arizona to take our nice clean solar electrons. We have to pay them for that privilege. And then we're back to natural gas for the ramp. Yeah. Well, everybody thinks it's pretty funny except for here, but, and we're back in that gas for the ramp, right? I don't know what these major initiatives like electrify everything, put it in heat pumps and we're creating a six, 7:00 AM winter peaking grid. Not easy to decarbonize all those incentives. Yeah. So we align those incentives with the regulators and utility. And then once you move from deemed to metered and you have aggregators who like this is a much better deal for the market, right? Like programs aren't working for anybody managing performance risk through market based approaches. It's just more efficient by a country miles.So, you know, it's, but it's just about once the incentives are aligned, we're not seeing any real barriers there. Like there's a lot of interest and willingness to move in this direction because nobody's scaling in efficiency programs anymore, then those days are over. So you know, we actually think it's this once a couple of, I mean we do the, the biggest barrier is that we are in purgatory where we have all of these rules designed for a deemed savings TRN-based universe that have basically no applicability in a measured pay for performance, meter based measure everything based approach. And so we're kind of stuck in the middle. So we need to, somebody needs to have some leadership and I think that's slowly but surely happening. But you know, options have a price. So keeping all options open, whether it's on the industry side saying like, Oh well he does great but like we kind of like getting T&M values and getting paid in advance.It's kind of good too. But like it's super shortsighted because that is not going to last. And look the way you could not be any more clear. So we need to get on, we need to decide that like before we're left behind entirely, we need to pivot. And again the pivot is pretty easy. We're just saying we should do behind the meter efficiency and flexibility like we already do everything else. How about that? And again, so this model that we're talking about, the way we finance it, the way it's structured, it looks precisely like how you build power plants, grid scale, solar grid scale, storage, storage, everything else works this way. And so, you know, it's a pretty big paradigm shift for people out of the efficiency industry. But like when we talked to bankers and people in procurement, this is all they understand and the missing link was none of that works without being able to measure it on the same dimension as everything else.And that's what, so what we do as a company, we spend a lot of time developing these standard metrics and putting them in the open source and put them in a Linux foundation so everybody can have them. Because nobody can own a weights and measures in a market. It has to be transparent to all parties, can't be a black box. So we built a lot of that stuff with public money and engagement. We don't own any of it. It's all in the public domain. Anybody can run the opening meter to the specifications of the CalTRACK methods. What we do as a business is we deploy that in a very efficient way as a SAS platform. Our first customer are the utilities where we look often and all of their customer data to identify, like I was talking about where is the potential, which exact customers have the right load shapes for a storage system with solar who has, you know, air conditioners that are really bad and causing grid havoc, you know, who are the actual customers?Let's figure out what the virtual power plant potential is and figure out how to form a procurement around that to go address it and then send that signal into the marketplace and then be able to track them. So that's kind of the utility side. And then once they do that so we provide them a platform to adjust the population data to define what is the virtual power plant and how do you target it. And then also to track what we call their fleets, which are when they actually do whatever the retrofits are. Those, let's track the actual buildings that are being retrofitted and how they're performing.Yeah, it is. It is. But we have this special friend, you might know them, they're called Google. And so we ability to parallelize in the cloud at like, except relatively exceptionally low costs. Like we, you know, to, if you want to look at every meter in a state you know, running it, running the running, opening a meter on one meter on a PC, on your little, like a local machine will take about nine minutes. So if you have a million meters, it might take 3,400 days or something. So we just do them on 40 or 60,000 computers at the same time. So there's, again, new tools. The stuff was not possible because of just the computational intensity, not that long ago, but now we can baseline every meter in the state every single night. You can do it every hour if we needed to. Just kind of gets pricey.Jason S. Trager, Ph.D.:And you haven't spun up a a million instances accidentally again, right?Matt Golden:Early on. Jason's joking cause playing with some of these tools is a little bit like experimenting with a loaded gun. So we had some, we had some early early things in our code where we had a couple of instances where we had we had server runaway, which happened kind of quickly, but Google is pretty nice about it actually. And no, we're, we're better at that now. So when you have things that automatically scale servers you have one infinite loop and you can end up on 100,000 servers kind of quickly. So we now have control and that kind of stuff. But it is really power. I mean it is powerful and that's one of the risks like playing with like you can go from zero to 30,000 servers in a blink of an eye, but that's the new world we're in and you need that kind of computational horsepower.So we provide, we provide discreet nodes instances, platform instances to these, each utility. We also provide similar SAAS platform instances to aggregators so they can forecast and track their portfolios and target customers and manage contractors and do all the things that they need. And we will look at the, we will measure the impact of each asset and also join that with all sorts of metadata that they might have. And we have API, so a lot of them will, we'll also build on top of our data, do whatever they want with it. And then, so we call it the utility side is resource planning, the program or aggregator. And it can be either as fleet management and then we, it's like a the ledger, which is the accounting system, which is each, you know, each customer has their own node with all their own data and it's totally a walled garden, you know, special to their, you know, their own instance, not connected to anything else.And ledger is how you actually track the grid values, but also pay for performance contracts. And so it's, we call it, it's a source of truth says here's our contract, here's how you're going to get paid. Here's your 14,392 eligible buildings. Here's all that. They all meet the data and eligibility requirements and here's their savings and here's what, here's how much you're owed this month. And if you'd like, you can audit each one down to the site and confirm that it followed all the rules in the contract so that we're not debating which weather station to use, how many temperature bins we should be using. Right. Yeah. Cause it's agreed to in advance cause what's the right answer? I mean the number of little knobs in these equations, it's hard to explain how sensitive they are so we can now put that on a contract.And so, and that's, you know, that's kinda the, what's meaningful is you can't litigate all these questions you need. You need like an agreement on what a kilowatt hour is and that's what we're providing. And so there is a place for detailed site level engineering. If you're going to give a performance guarantee to one customer, you got to hire a bunch of engineers to get that one project right as right as you can make it and compensate for occupancy and the killing in their basement or whatever the heck's going on. But that does not scale. So we don't try to claim it's for everything, but, but we would even say on the site level is for the most part, there's a lot of ways to provide insurance and whatnot where maybe you don't have to be still be so accurate on individual assets either.Jason S. Trager, Ph.D.:Fair enough. Well do you have any closing thoughts?Matt Golden:Only the is don't be intimidated, you know, on either the aggregator or the program side. Like there's a bunch of no regrets stuff in here, which is, there's absolutely never any regrets to measure what you're doing and start utilizing that data, like aligning, you know, figuring out how you're performing in a near real time basis so that you can do better across the board. So just get started. And I think as an industry like we got to get on board with transitioning quickly cause purgatory is going to suck us all dry.Jason S. Trager, Ph.D.:That's, that's fair. That's good advice, Matt.Matt Golden:And nothing's broken. Let's just go do it already. We've got all the pieces.

Jane Peters on What Motivates Stakeholders to Choose Efficiency

Play Episode Listen Later May 5, 2020 51:15


In this episode we cover:- How Dr. Peters got into environmental psychology- What kind of data was available when she began, and how it was collected- What drives commercial and industrial energy customers’ interest in efficiency measures.- Why customers sometimes signal interest but don’t follow through.- How data utilization has evolved in the utility space- How data played a part in California’s new M&V guidelines.Full Transcription

Amory Lovins on Bending Fewer Pipes and Recalibrating Expectations (GM103)

Play Episode Listen Later Apr 2, 2020 56:23


The thinker, author, and all around advocate for Integrative Design joins us from his blizzard-proof tropical paradise in Colorado. From his latest paper: Recalibrating Climate Expectations to how design software wastes energy by limiting the angles pipes can be connected in buildings, Amory dishes the data on efficiency.

Mary Ann Piette on the Evolution of Data in Demand Response (GM102)

Play Episode Listen Later Mar 30, 2020 38:45


Mary Ann Piette is a Senior Scientist and Division Director at Lawrence Berkeley National Lab. Her work focuses on energy-using technologies and buildings as well as grid integration. Additionally, she runs the Demand Response Research Center, whose goal is to understand “what works,” in the field of Demand Response (DR). Mary Ann relieves the days of capturing data on 3x5 cards stored in a records room and reflects on how streaming services and automated control are revolutionizing the way DR is capturing and analyzing data--and how the technology is still accelerating.

Dan Kammen on Energy, Data, and Canned Air (GM101)

Play Episode Listen Later Mar 30, 2020 57:19


Dr. Dan Kammen, Distinguished Professor of Energy in the Energy Resources Group at UC Berkeley and Former Science Envoy to the State Department is the guest in the premier episode of Global Minima: the podcast at the intersection of bits, watts, and dollars. Dan talks about how data collection on air quality has gone from sealed cans shipped to a lab to air quality sensors on smartphones, and how data can tell us a lot about how to achieve environmental sustainability.

Claim Global Minima

In order to claim this podcast we'll send an email to with a verification link. Simply click the link and you will be able to edit tags, request a refresh, and other features to take control of your podcast page!

Claim Cancel