The Exchanges

Q Right, so ERCOT's doing something right. So what is ERCOT doing? Like, walk us through ERCOT's interconnection process.

A The defining feature of ERCOT's interconnection process is that they don't rely on the interconnection process itself to identify and pay for grid upgrades. And this enables them to interconnect projects much more quickly and at much lower cost. And what allows them to do this is that All generators in ERCOT are energy-only resources subject to economic and security curtailment, which means that ERCOT can turn them off for effectively any reason, um, according to market dispatch protocols. And so, so ERCOT manages its grid constraints by curtailing generators as necessary, uh, with, with that market dispatch. And so, these generators can connect to the grid very quickly at relatively low cost, but with the understanding that they may be curtailed to mitigate any grid overloads, and, you know, this, this doesn't mean that ERCOT doesn't pursue grid upgrades, it's just that they do it via separate transmission planning process, and their market structure enables them to really clearly measure These congestion costs that arise because of the system's inability to deliver lower cost, uh, available electricity to load. And so, you know, a simplified example might be helpful here. So say you have a, a city, it's supplied by two large transmission lines. You know, the first line connects to an aging coal plant, you know, with a high cost of electricity that you don't want to run very m…

AI assessment note: “The defining feature of ERCOT's interconnection process is that they don't rely”

Q maybe Tyler, you can comment on this because I've seen you say it publicly. It seems generally, maybe outside of FERC, but in the public domain, this has been pretty well received, broadly speaking, which, I mean, I think does run in contrast to, I remember that previous letter about coal, 90 days of storage, and so on. So your, your sense is that the vibes here are good, Tyler?

A So, so far, you know, let's not discount that There will likely be other perspectives that have not yet been represented, but no, look, I think it was very significant that, um, you know, Commissioner Rosner came out of the gates expressing, you know, an eagerness to, to work on the proposal. You know, on the other side of the aisle, you had, you know, Senator Mike Lee that came out, you know, strongly supporting it, and a variety of different stakeholder groups, at least I've seen, and, and companies that are involved in this space, um, seem to view it generally favorably. I think Just to parse out a key distinction, right? There's this jurisdictional question, and on that one, I think there's obviously going to be a variety of perspectives, and there will be concerns on the part of, especially of some state commissioners, officials, and certainly the investor-owned utilities. But with respect to the substance, that's where I've seen the most excitement, and I know we'll get into it, but in terms of the What, what this would actually do to sort of improve the interconnection process, that's where I think there's been the most positive reception.

AI assessment note: “Senator Mike Lee that came out, you know, strongly supporting it, and a variety”

Q off of, like, typical demand response program frequency. Is there anything different about what you're proposing in terms of the curtailment? Um, not the frequency necessarily, but the type of curtailment versus demand response programs? Like, is there a, is there a shorthand version of the takeaway from this paper, which is basically enroll every data center in a demand response program, and you've Like, solved a lot of problems?

A That would be a, you know, a simplistic way of looking at it, but I, so I think there, there's a one really key distinction, uh, from existing demand response programs, and that is that the vast majority of the loads that participate in demand response, they were planned as firm loads, right? So when the transmission provider ran their interconnection study, and the transmission plan for that jurisdiction, and then the reserve margin planning, uh, They planned for that load as firm, and then at a later date, once that load was online, it decided to participate in demand response for economic reasons, right? But what we're talking about in this case is pulling all this up into the planning realm, and it's sort of like how, you know, we're trying to get a lot more sophisticated with transmission planning. It's like that, but trying to get more sophisticated with our load planning. And so just to articulate, I mean, I think What we're hearing, right, the tip of the spear and what would really present a value proposition for the hyperscalers and the large co-location data centers would be if you can actually interconnect them to the grid more quickly if they're able to offer this flexibility.

AI assessment note: “there's a one really key distinction, uh, from existing demand response programs”

Q now the systems are designed to make sure that you have enough power, enough generation, uh, that is deliverable at the peakiest peak. Of the year. And so this is basically saying if you added a lot of data centers, but they did not contribute to those peakiest peaks, how much could you add? And that's where you get these like astoundingly large numbers. Do I basically have that right?

A That's right, Shale. And just to put some numbers to that utilization rate that you mentioned. So you said it exactly right. We build the entire power system around these occasional extreme peaks that are driven by either Extreme heat or cold snaps, especially these polar vortex events, but, but just to quantify it, right? So in 90% of hours, more than 30% of the power system sits unused. Uh, and we actually found that on average across those 22 balancing authorities that the average load factor, meaning the average consumption over the peak consumption, is 53%. And so what that basically means is that in any given hour, On average, approximately half of all generation and transmission infrastructure is unused on average in the U.S., and it's actually worse than that because you're not counting for all the reserve margin that's on top of those existing peaks.

AI assessment note: “That's right, Shale. And just to put some numbers to that utilization rate”

Q . You know, the utility or, or, or, whoever it is says, okay, we need to shut you off, or we need to shut, close you down to some degree, X number of hours per year, and we're going to give you one day notice when it's going to happen, and we, we have a cap on how much we could do that. Is it that kind of a thing?

A I, I think the program structure is likely to look something like that. I, I mean, obviously, the more controllable the load is on a real-time basis, um, The more valuable it's going to be to the system for flexibility. But, you know, especially in the context when we're talking about a limited number of extreme weather systems that are really driving these extreme peaks, you know, our weather forecasting, fortunately, is getting a lot better, right? So, like, we can see incoming polar vortex events, you know, up to two weeks out, and certainly within a week, and begin to sort of plan around that. But the other finding, and we wanted to look at You know, what, what does this look like in terms of how much of the new load is curtailed for a given number of hours? And cause, cause I think initially we would have assumed, right. That a lot of times we are talking about, you know, a hundred percent curtailment of this new load, but when we ran the numbers, so like at that, that 0.25% load curtailment rate on average, we found that the number of hours in which Curtailment would be required. Some amount of curtailment would be 85 hours, but in 73 of those 85 hours, at least 50% of the new load is retained, and then in 50 of those 85 hours, at least 75% of the load is retained. So we're really talking about partial curtailment here in the vast majority of cases, and, you know, in term…

AI assessment note: “I think the program structure is likely to look something like that.”

Q MISO or CAISO or some, some market outside of Texas, and you could wave a somewhat realistic magic wand, in other words, you can reform the interconnection process, but you're not going to do the fundamental structural market reform, like you can't turn the You know, PJM into an energy-only market all of a sudden. How would you distill, like, what, what could be done within the bounds of reality?

A Yeah. Yeah, so assuming that you're still going to have generators pursuing the interconnection process to receive capacity rights, you have to figure out a way how to study both energy-only resources and these interconnection studies alongside the These capacity resources, and this is tricky, but there really, there should be a, a relatively simplified approach to this, and so there are a number of experts in FERC's rulemaking process that recommended that instead of one big combined, you know, integrated cluster with, with both energy only and capacity resources, that you, you just do this in two steps, right? So you, You start off with a cluster that's energy only, and at the end of that process, the, the interconnection customers will make a go, no go decision on whether they're going to proceed. And, and then, you know, you take all those and then you, you run the second stage, which includes all the capacity resources. In that study, you, you actually, you turn off the energy only resources because again, they're, they're not, um, they're not receiving capacity and they're not contributing to, to overloads because they'll be, Curtailed in real time. So pursuing, ah, a two-step study process with energy only first, followed by capacity resources, I think is, is a relatively simple one that can be, can be adopted quickly. There's also, ah, reforms that can be made to, ah, w…

AI assessment note: “pursuing, ah, a two-step study process with energy only first, followed by capacity resources”

Q you think that Texas is actually a model that we might think about replicating, or at least partially replicating, in the rest of the country. Let's, before we talk about what Texas actually does in terms of interconnection, let's talk about the impacts that it has had. So what are the, you know, how has Texas performed in terms of interconnecting new generation relative to the rest of the country?

A What ERCOT has accomplished here is truly remarkable, and I think maybe it's helpful to start by just capturing the total installed capacity, just to contextualize, right? So, uh, total installed capacity for, for wind to start, you know, this story is better known, but, uh, so Texas has around 40 gigawatts of installed wind. Uh, it's three times more than any other state, so Iowa is the next at about 12 and a half gigs. Um, this is just under 30% of all U.S. wind power capacity. So that, that story is relatively well known. Uh, solar is, uh, is a newer phenomenon, and so ERCOT, uh, is now at about 17 gigawatts of utility scale solar installed. The state at large at around 20 gigawatts of all solar. Um, so it's now number two in the U.S. just behind, uh, California. And the utility scale side now appears effectively equivalent with California. Texas has just blown past every, every other state. You may recall one point that North Carolina was, was number two in solar. Um, you know, even as of 2018 or so, Texas was substantially behind. And just in a few years, they have interconnected a massive volume to, to surpass North Carolina, Florida, and Arizona. And so on an annual capacity additions basis, uh, so in 2020, I think is really where it starts to take off. So Texas interconnects 2.5 gigs of solar in that year. 2021, it explodes to four gigawatts, and then 2022, it's another…

AI assessment note: “Texas has around 40 gigawatts of installed wind. Uh, it's three times more”

Q So it seems very logical to me on its face. Let's contrast it with what happens in the rest of the country. So I'm a, I'm a perspective generator, and I submit an interconnection request outside of Texas. What happens in that case?

A Yeah, so I think the most important thing to understand here for your listeners is what's called the system impact study, and so what this does is it assesses the impact of the grid of the power flow from the proposed generator in interaction with the rest of the system, and so transmission providers use electricity simulation software to conduct this power flow analysis. One of the most common software packages is called Tara, And so what they do is they, they look out several years at a specific fixed point in time. It's called a, to get jargon, it's called a steady state circuit analysis. So it's a single snapshot of the balancing authority in time. They load your project in, and all other generators that are committed to the system, you know, all the existing operating generators, all those that have executed interconnection agreements, And then others that are, ah, ahead of you in queue. And then they run all these power balance equations for power flow simulation to capture the supply, the demand, the transfer of power throughout the system. And then what they do is they run all these contingency analyses where they look at, um, these sort of extreme scenarios where different elements of the system go out, right? So, A transmission line goes out, a transformer fails, uh, and, and other scenarios, and so they test a bunch of these contingencies to see if your generator con…

AI assessment note: “most important thing to understand here for your listeners is what's called the system impact study”

Q not really unified, so this tries to unify them and say, consider them together. Um, but beyond that, you know, these things are kind of intertwined, because generation might be the mechanism, if it's behind the meter, it might be the mechanism to be curtailable. So, as you think about the big picture world of data centers and their energy provision, where does generation fit in, in your mental framework?

A Yeah, maybe it's worth just stating very clearly up front that, like, one of the things that has been revealed over the past year, you know, based on this co-location docket in PJM and other considerations and other jurisdictions is that we really have an antiquated, um, load interconnection study process and study criteria because, as you said, it's really divided from the generation side, and so this has very significant consequences If you are talking about co-located generation and load, because what you might have going on, right, is that a generator can be, um, essentially offsetting the withdrawal from the grid from a given load during the most stress period, right, which is what the utilities are actually studying to determine your network upgrades. And so if they don't consider the ability of that On-site generator to offset your withdrawal, you may be much more likely to trigger the need for, for major network upgrades. And of course that it's more expensive and it can take multiple years. So that's, that's sort of part of the delay. I guess to your sort of broader question there, I mean, I suppose it's no secret that the preferred option, right, for flexibility, or we could say curtailability on the part of these large loads is, you know, either on-site generation or storage. I do think We, we've heard a lot about the, the, the generation option, but, and we hadn't r…

AI assessment note: “the preferred option, right, for flexibility... is, you know, either on-site generation or storage”

Q like the state of interconnection in the U.S. today, particularly in light of the fact that until fairly recently, um, you were involved directly in interconnection by working at a renewable project developer. So you, you probably have seen it evolve to some degree over the past few years. So how would you characterize where, where we are at today and like what the past few years have been like?

A Yeah. Well, Shail, I think it might actually help to start with an analogy here before we get too far into the weeds. So imagine with me that it's 1953, right? We've won the war, the economy's moving, um, automotive technology is advancing really quickly, and there's a demand for a lot more interstate commerce. The interstate highway system hasn't been built yet, right? There's a patchwork of smaller highways that connect each state. So President Eisenhower announces a new program, Right, where trucking companies that want to transport goods across states can apply for the right to do so. Once they apply, they'll go through one or two years of study. These will be really intricate studies, right, where we're, we'll consider detailed forecasts of future traffic, weather patterns, wear and tear on roadways, and we'll consider a bunch of different extreme contingencies. And the output of these studies will identify where each trucking company's fleet is likely to contribute to traffic congestion, And when it does, they'll be assigned a portion of the costs required to expand those highways. And the idea is the trucking companies can't access the roadways until they've paid a hundred percent up front for these future highway expansions. And so President Eisenhower goes out and he announces this, and he says, you know, this is expected to hopefully, maybe, if all goes well and all t…

AI assessment note: “it's not entirely dissimilar to how we've been approaching transmission interconnection”

Q to be able to provide X hours of flexibility with the Y duration events or whatever, and then let the market figure out how much of that is going to be Generation storage and load flexibility? Or does it make sense to actually, um, I don't want to say put your thumb on the scale, but like try to dictate a little bit how much of what gets implemented there?

A Yeah, I know, and actually our, our research lab at Duke University has been, because, you know, we're modeling sort of the bulk power system, and so, ultimately, you want it to be as sort of generalizable as possible, right, so that you can sort of represent any load that is utilizing this type of flexibility for all these purposes, and so you want to kind of parameterize it and make those parameters as generalizable as possible, also because then we can actually create a market, Right. Where a variety of different options can compete, uh, either onsite or even, um, to some extent offsite. I mean, this gets even more interesting and complex when you start to think about, you know, these large loads potentially procuring the flexibility from other loads in the same balancing authority, the kind of simplest version of it's just for capacity, right? Just resource adequacy. But then if you're actually talking about to sort of mitigate transmission congestion, you know, think about the electrical proximity, uh, Of those customers and how you sort of do that study. I think that's sort of like the, the, the, the very advanced version. I hope we can get there, but at minimum, like, let's get in place for, for those that are doing it behind the meter. And yeah, whether it's load shifting, um, you know, actually shifting around the computational workloads or it's, um, you know, it's bat…

AI assessment note: “then we can actually create a market, Right. Where a variety of different options can compete”

Q from the grid operators or from the data center world on, like, how much could this, how much does this suggest we should just move quickly on demand response type programs, or as you said, the type of program where there's a deal in exchange for faster time to power? Like, is there momentum around this, or do you feel like you're, like, fighting an uphill battle for some reason?

A Yeah, great question. So I want to say one thing up front. I don't know where this idea came from of 24 seven. They are not 24 seven loads, and it's critical to clarify this. So the servers may be 24 seven, right, the chips themselves, but there's an enormous amount of cooling infrastructure and other infrastructure around those servers. So, so actually, like, Lawrence Berkeley Lab and their recent data center energy usage report in December that was congression mandated, They put out the number of 50% utilization rate. EIA and E-three have used a number closer to 85% utilization, but I just want to be clear, like, these are, these are not, like, you're not running the data centers that there were a hundred percent potential max draw, uh, at, you know, a hundred percent of the hours, uh, so I'm hopeful that we can just, like, further clear, because I think the regulators in particular have been very confused by this, um, so, you know, good to establish that, but to your question, It's, it's a really interesting dynamic, right? I mean, we're in such a hyper-competitive environment where I think, you know, the hyperscalers and even some of the, you know, these co-location data center developers, uh, you know, they don't necessarily want to be a hundred percent forthcoming about the extent of their capabilities, and to the extent they are going to be forthcoming about them, you kn…

AI assessment note: “to your question, It's, it's a really interesting dynamic, right?”