 Good afternoon and welcome to today's energy seminar. Again, I'm going to have somebody other than me introduce the speaker. So I'm very excited today to welcome our speaker, Jay App, but I'm going to have my colleague in the new department of energy science and engineering, Anash Asavido, who's an old colleague of Jay's that Carnegie Mellon introduced him today. Anash, as you may know, is one of the world's experts in energy system modeling and more recently has led the way into the diversity, equity, and inclusion dimensions of energy system operation. So Anash. It is such a pleasure to introduce Professor Jay App, who is a professor both at the Tepper School of Business as well as having an appointment in the Department of Engineering and Public Policy at Carnegie Mellon University. More than a colleague, Jay was also my professor during the PhD years and I was very frightened of him, very afraid, because his very sharp mind and questioning would make anyone kind of shiver. There is no person that I'm aware of that knows more about the electricity industry. It has been always a pleasure to learn from Jay. He has a PhD in physics from MIT, 1976. He also was an astronaut and worked for NASA for a long time, so you may have some questions about that too. And with that, I welcome you, Jay, and we look forward to your seminar. Anash, thank you so much. It's always terrific when people that you've known as students feel much better than you do as professors, and there's a couple of those in this room. And that's just terrific. So thank you, John. Thank you, Anash, for inviting me here. We're going to talk today a little bit about electricity and gas system as well. So most of you know that the U.S. electric power system is powered by about 20% coal. When I started it was 50%. I didn't do all of that. And now about 40% natural gas, 20% nuclear, and about 20% renewables, although large hydro is often not considered to be a renewable. And I spent about 15 years of my career looking at the mathematical characteristics of renewables to figure out how to reliably integrate those into the electric power system. In the last few years, I've been thinking a lot about the other 80% or 85%. Because renewables we all hope will grow, but any of you who know Vokov-Schmiel's work know that they're not going to grow by next weekend. So we have a lot to do with the reliability of the rest of the power system if we're going to keep the lights on. So what I'm going to cover today are basically a couple of things. One is what we've learned about the reliability of electric power generators and a little bit about the Texas blackout that happened 21 months ago and how the power grid could better provision reliability reserves so the lights don't go out. And second, some available data on the natural gas grid reliability. Spoiler alert, it's not nearly as good as the electric power industry. I'll talk a little bit about what we know about the reliability of gas pipes and gas-fueled generators and then what we've learned about the dependence of the gas pipelines themselves on electricity. Okay, so let's dive right into electric power generation reliability. We've been able to make some big advances because a group called the North American Electric Reliability Corporation, NERC, has required that every generator of any size over about 20 megawatts, that's not wind or solar, report any type of failures that they have to NERC and it turns out that the largest electricity market in the country, PJM, has required that for a bit longer. So we had a fairly short data set from NERC four years initially, then six and then we had almost a quarter century of data from PJM. But the NERC data set covers 85% of all of the generators in North America and it's been critical to understanding some of the reliability issues in a way that just wasn't possible before. Because this is being recorded, I put up the references here so you can play back the tape and see what they are. The initial work on the NERC and PJM data sets was done by our PhD student, Senate Murphy, in cooperation with John Moroff from NERC and a brilliant econometrician, my colleague at Teprphila Soul. We published those after some difficulty, as those of you who tried to do it, in Applied Energy back about four years ago for the initial stuff. So what are these data? Well for all of the eight regions of the reliability areas in the U.S., the North American Electric Reliability Council defines, we have three types of failures for every generator. There's first what are called unscheduled outages, the whole thing breaks. The second is it fails to start, start failure. And the third is a derating, well what's a derating? Well, if you have let's say a coal generator, it's probably supplied by three mills that mill the coal from fist-sized things down to coal dust. If one of those breaks, a third of the plant capacity doesn't work. So that's a derate. Natural gas, even wind turbines and solar rays, everyone, what's a derating in a solar farm? Well, one of the inverters might break, right? So that's a derating. And the relative size of these things varies from place to place in the West, what's called WEC, the Western Electric Coordinated Council where you are, in the lower right of this diagram, about three quarters of all of the megawatt hours that were lost, the whole thing just broke. About 2% they didn't start, and about 20% were derates. Now that derating percentage can go up to 35% in some of the areas. So we have these data for every single generator, any place in North America. For decades, starting in the late 50s, people have been fantasizing about trying to do probabilistic risk analysis and provision of reserves by doing what's called a Markov model. Markov model is not any fancy math at all. It's just you have probabilities of transitions from a working condition to a partially broken condition or a fully broken condition, vice versa. In this simple model, you'd have six transition probabilities, right? Until now, people have had absolutely no way of determining quantitatively what those transition probabilities are. So we now can, and that's very useful. I'll give you an example from our first paper on this. These are histograms, notice the semi-log scale here, vent duration versus count for generator outage duration in hours along the horizontal axis for all of the nerc regions. They're quite different for various reasons. But from that, you can determine, for example, a mean time between failures, and you can break that down to coal generators or natural gas generators or hydro or nuclear, if you like. And then you can have those transition probabilities. We can also get the mean time to recovery of these generators because we know when it fails, we know when the next shows up working, right? So these data are now available. And that's very useful for trying to compute what's going on. We can aggregate those data into time series as well. And so here's one for the area in the mid-Atlantic states. Reliability first corporation is PJM, the same area in the mid-Atlantic. And the blue line at the bottom is the failures to start over the four years of these data. The red one is the partial failures, green is total failures, and the black is just their sum. So you can do that. Now, these data are really high resolution. They're hourly. Well, they're supposed to be minute by minute, but when you look at it, nobody really records it except once an hour. So anyway, they're pretty high resolution, hourly data. And you can see things like this big spike that was a very cold weather event that was turned by the TV weather people, the polar vortex, right? And the polar vortex in January of 2014 knocked out 22% of all of the generation in PJM, big event. Right over into some of the other regions as well. But this is the kind of stuff that you can do. And so here it is for all of the regions. So down in the west, in your region, you can see that an average of 4% or 5% of all the generators are broken, and it never gets much worse than that. But look at one panel above it, and there's RFC. Those are the data I just showed you. It gets to a maximum of 22%. The others regions are in between. And so if you are trying to figure out how much extra stuff you need to keep spinning to keep the lights on, these data are really important to you. And it differs from place to place. They're not the same. So if generators are, failures are time dependent, why? What affects them? Well, people have speculated since the early 60s that environmental conditions can cause generators to break. Seems reasonable. But it turns out it wasn't until this event that I just talked about, the Polar Vortex in 2014, that anybody bothered to plot generation failures against temperature. Or in the case of the PJM report, they did it against windshield. And they got it, you know, that's a pretty decent correlation. And so that goes down to about minus 20 degrees, in this case Fahrenheit. And things sort of break, right? So we had data for PJM that was 23 years long. It turns out that the six-year country-wide NERC data just isn't long enough to get good statistics. But the 23-year PJM one was, and it covers 1845 generators in PJM for, as I say, 23 years. So what I'm going to show you for six different generator types. Let me say what those six are. There's combined cycle natural gas units, the efficient ones. They're combustion turbines. They're basically jet engines that we use for peaker plants and for some other reasons. Then there are diesel plants, hydroelectric plants, nuclear, and steam turbines that are basically coal. So the dashed line here is what happens if you don't take temperature into account. So for gas plants, hydroelectric plants, and nuclear plants, basically 3% of the capacity is broken at any given time. For coal and for diesel, it's more like 10%. Now, when you take temperature into account, this is what you see. This is at median load. Nothing special is happening with the load here. And look at what happens at low and at high temperatures. And so everything, except for nuclear, exhibits this smile. And it should be a frown, right? Because stuff is breaking. Nuclear has the problem at high temperatures, presumably because the cooling water isn't so cold. That's why you have to derate the plant. I will get to one nuclear case. This was in PJM. In the Texas Blackout, one nuclear plant failed because of instrumentation as it turns out. But in PJM, they hadn't seen this. But for all the others, it goes up and up. And it's not a small amount. So here's a combined-cycle gas plant. Average one is broken 3% of all capacity or 3% of the time. But when it's cold out, and that's not very cold, that's minus 20 C, five times that many plants break. So nobody has ever looked at this before. So you've got to, when you come up with new data, be the most skeptical person around of your own data. So we said, all right, what about load? Maybe the load is just higher there. So let's look at it, not just at the median load, but at 90th percentile load. Yeah, the things look the same. Loads not. John, if you want to scoot over here, okay, okay. I'll try and not stand in the way. So load wasn't a problem. What else might be the problem? Natural gas wells freeze up. They did in Texas 21 months ago. And so maybe it was all for the natural gas plants just that they couldn't get any gas. So these data now with uncertainty bars are all of the events that affected gas plants over those 23 years. And these data have the same data only we excluded ones where the plant couldn't get any gas, which they tell us in this big data set. Looks basically the same. It's a little better. Looks basically the same. All right, so what else might you think? Well, how about old plants? The old plants ought to break a lot more than the new plants. Great idea. It doesn't work. So these are decades plant to plant. The lowest ones are the 1980s. And the highest ones are just the decade later. And the 1950s units are hanging in there pretty good. They're the red dots. So it wasn't that. So something out there is really happening. So what? How do we relate this to grid reliability? Well, it turns out that every single grid operator does the math to figure out how much extra generation they need to sit around in case of emergencies by assuming that the plants fail independently. For all of the mathematical optimizers in the room, I'm not going to show any equations. This is IID. That's what they assume. Independent identical distributions. Well, we see correlated outages happening in every single NERC region, even when you exclude big storms. Correlated outages in the longer PJM data set are there even when accounting for peak load, for natural gas availability, and for generator age. So correlated outages ought to be considered in defining resource adequacy requirements. Which brings us to 21 months ago, Texas. Yes, sir? Are we allowed those questions? Why not? Physical equipments or maybe the personnel doesn't want to go out because it's cold. Ah, good question. So it's not so much the personnel. The physical equipment can absolutely cause this. I have been at a natural gas plant that has a big three-story high air filter, where it's sucking all the air that it's going to combust. That thing throws up in the winter at a gas plant that I went to. So some of that happens. Sometimes the natural gas wells as I'll get to freeze up. But there's lots of things. And let me show you in Texas what happened. All right. For those of you who were not paying attention because it was in COVID, this is an outage map. Terrific website, poweradders.us. This is an outage map on the 15th of February 2021. Texas was basically out of power. That wasn't the only place out of power. The cold weather moved east a little bit and then there were some random places. But Texas was basically out of power. So Texas' generation mix in the four year before that was about half natural gas, about a quarter wind, about 20% coal, 10% nuclear, a little bit of solar. So what broke on those three days? Well, 15 gigawatts of natural gas. For those of you who don't speak gigawatts, 1,000 megawatts, a large nuclear plant is about a gigawatt. ERCOT's got 107 gigawatts of capacity or dead at the time of the blackout. So we lost 15 gigawatts of gas, 4 gigawatts of coal, 3 gigawatts of wind. The blades can ice up or the gearbox can freeze up. So a gigawatt of solar and a gigawatt of nuclear, which turned out to be instrumentation. That upper picture is actually instrumentation at a nuclear plant that happened to make it through. So, why? This is a temperature anomaly map on the 15th of February. Apologize, this is in imperial units, degrees Fahrenheit. Right, but it's great website, climatereanalyzer.org. And so, white is like 25 or 30 degrees below normal Fahrenheit temperature. So it was cold throughout much of the greater Midwest, the Great Plains States. And it was pretty cold in Texas. You who were paying attention because you didn't have anything better to do during COVID, to Governor Greg Abbott may have heard him say this was unprecedented cold weather. Don't blame me. Well, so here was 10 years earlier. This was 2011, same map, same scale. And while it didn't affect East Texas so much, New Mexico is not so far from Texas. And it was even colder there in New Mexico. Not only that, here's the map from 1983. Much of the country was fairly cold, including parts of West Texas and much of the Great Plains States and parts of the area near where some of us live or used to live. This was not unprecedented. But Urkot's highly paid, I suppose, consultant looked at this and their report, very comprehensive report that they did in 2015, six years before the blackout, looked at these events and blew them off. So here's a quote from their method section. The weather here was given equal probability except for 2011, which was given a 1% probability, right? One in a century. Okay. They obviously didn't think too much about the 1983 one. And just four weeks before the blackout, they came out with another report and again, they did not consider temperature in assessing the generator failures. So what is a professor with a terrific Ph.D. student do, writes an op-ed for the Washington Post when this happens. So absolutely unbelievable student called Luke Lawvin, who's now at NREL, and I wrote an op-ed for the Washington Post that talked about this. But we went the first. In 2011, the Federal Energy Regulatory Commission, FERC, had done a report and said, you know, this is what a gas wellhead looks like in Texas. It gets cold in Texas. It got cold in 2011. You don't need to leave one protected like that. This is what it looks like up in the upper Midwest, that's silvery stuff is insulation, you're all engineers. You understand that? You inject methanol into the wellhead and you can keep it from freezing. They didn't stop there. They said, how much does it cost? So this table says it basically costs capital expense of something like $30,000 a wellhead and something like $10,000, $13,000 a year and you keep it operating when you do that. So for active wellheads, not a little but not a lot either, and you can do that. You can also do it with conveyor belts for coal plants, a website of a company that sells that stuff. You can do it for wind turbines. It's not only the blades, you can buy heated blades, but it's the gearboxes that can freeze up. You can buy cold temperature packages, but Canada got really surprised a few years ago when it went down to minus 25 and the wind turbines shut themselves off because that was their lowest rating and the software, which I think one of my former students, Steve Rose, wrote, shut them off. They didn't buy the cold weather packages. That was a reliability concern, which they weren't aware of. So here was our op-ed titled something provocative and I'll read you a quote from it. But the math that all grid planners use has the same fatal flaw. It overlooks the tendency of plants to go offline in clusters. When I say fatal, we didn't know but a couple of deaths at the time. There were 200 confirmed deaths due to this outage and if you use what was done during COVID, which is excess deaths, it was 800. So everything's got to be okay nearly two years later, right? Here's the Austin newspaper. You can all read the headline. And it was based on FERC's winter energy market and reliability assessment that basically said they did some paper studies but they really haven't done pretty much. So it happens, you know, three times in 40 years. It's not probably going to happen this year, but it might. So how should we really procure reserves? Well, there's Luke Levin and Senate Murphy and our student Brian Sergei, all three of them now at NREL, National Renewable Energy Lab, and we did a series of papers and we asked basically, hey, is it possible to procure less reserves? Well, less? We should be procuring more. No, PJM has for years way over-procured their reserves because their CEO, Terry Boston, very conservative engineer, they're then CEO, and Terry put his thumb on the scale and basically said, now, how shall I procure more reserves? But they do it basically all year round. And so we did a model of PJM, very simple reduced-form model, five regions, six transmission lines, but every generator there, something that's easy enough to do. And we said, well, suppose you procure reserves week by week or month by month or season by season, which is what we recommend, procuring more reserves when it's cold or when it's hot and less reserves when it's not cold or hot will save you like $17 million a week, cold weather weeks. So we've testified to PJM and so forth about this. Maybe when you're my age, they'll do that. So we'll see, but it's certainly feasible to go do that. All right, so now let's talk about the gas grid in the second half. Let me just ask you, since I wanted to go through this, then there'll be 10 minutes at the end for questions. I loved your last question. I'd love to hear that. All right, so this is what I mean by the gas grid. These are all the pipes in the U.S. that are bigger than 12 inches. That is the gas grid, including a bunch of gathering lines for the wells out in the Gulf there and some out to California. Notice you are not that well supplied with gas, right? And, you know, when this one went out just before the big price spikes after you guys had a market, right, it happens. So that's the gas grid. These are the natural gas plants which are largely located along the big pipes. That's just EIA data for where those are. All right, so what data are out there that people like you and I can use to assess the reliability on the gas side? Well, there is, as Adam knows, some truly awful data out there, okay? FERC requires a bunch of forms, one of which they've recently abandoned Form 588 that should capture all the gas outages but the requirements were lousy and they've since dropped it. They do require reports of service interruptions and what are called force majeure. Force majeure is a legal term that basically means our contract is null and void because God did this. Okay, the thing broke. That's a force majeure event and it usually means something really bad happened. So those are somewhat useful. An organization called PIMSA, the Pipeline and Hazardous Materials Safety Administration does require you to report if you have a release that killed somebody that caused more than $50,000 worth of damage, not released gas, or that released more than 3 million cubic feet of gas. So I don't know what that is in kilograms, Adam, you can probably do that, but it's a lot of gas. A fairly decent sized gas generator, 200 megawatt generator consumes nearly a million cubic feet per hour. So that is a lot of gas that's way too high a limit. These are basically not useful. So what about states? So there are some states specifically Wyoming that requires reporting of everything and puts it out on their website. That's terrific. Some other states do a few other things and that's great. Only three states require the pipeline operators to report pressure drops, which are bad because you have a big gas turbine going around if the pressure drops a little, it's going to automatically cut off. And the thresholds are really variable. All right, so what is useful? What is useful is that pipeline companies maintain websites that you can scrape the data from. Most of them don't archive the data, so you have to do it in real time, but some of them do. And I'll give you an example. This is the transco pipeline that runs, well you see where it runs, and these are the gas plants, the circles denote the size along the pipeline. And we, by which I mean my student, Jerry Freeman, looked at these pipeline websites for what are called operational flow orders. And OFO basically says there's an imbalance between gas being supplied at one end and taken out at the other end, something's wrong with technical definitions. But we found there's basically 10 of those a year and of the 35 and the three years we examined, six of them had coinciding power outages from the big data set, the GADS data that we looked at. Almost 300 power plant failures happened during those six events. So you can buy gas in two ways. You can buy it on the spot market or you can buy it with firm contracts. The interesting thing we found is that 14 events at four power plants failed while holding firm contracts. It's like being on Southwest, you can be A2 in the group, but if the plane doesn't go, you still don't go. Okay, so we wrote that up. Here's the paper by we, I mean Jared, and Michael Dworkin, who was the chair of the Vermont Public Utility Commission and then chair of the Environmental Law Department of Vermont Law. So what do we really need? What we said in our article was we need consistent reporting standards for pipeline events. And we were quantitative about that and these numbers aren't interesting to anybody but a specialist, but we were quantitative about it. And we said these data ought to be collected by somebody like NERC, a federally-chartered organization as opposed to being totally scattershot like now. A National Academy's report, the most recent report on the electric power system chaired by my colleague Granger Morgan with some people on here that are all familiar to many of you, picked up that recommendation and said Congress should build on the example it said in the electric power system when it established an electric liability organization and ought to do that. And last year, just about a year ago, the chair of the House Energy and Commerce Committee's Energy Subcommittee, Bobby Rush, introduced a bill to do just that. Now, no bill ever goes any place the first time it's introduced, but we are hopeful that in the next Congress that may go someplace, especially if there's another blackout. Okay, so despite the limitations, if we work hard we can learn some things and I'll tell you a little bit about that. Here's a paper. John Mora at NERC and Jared Freeman now at the Pacific Northwest National Labs were on this paper. And we had to put together data from a bunch of sources, the GADS database that I told you about, EIA data that tells you where these generators are, there's a form that tells you whether they basically had firm contracts or spot market contracts, and then some very obscure data on pipeline scheduling and pricing data, some of which we had to buy. But we could do things like, here's the four years in PJM and there are something like, I don't know, 17 pipelines and we found the plants on each pipeline that were out of gas because the pipeline didn't supply them as a function of time on those. So that's the kind of work that you can do if you carefully put this together and what this new organization should be doing. So here's what we found. The first thing is the big pipelines are really pretty reliable. Only 5% of all of the times that gas plants went out because of no gas were due to something breaking on the pipeline. No more than 5%, and I'll tell you in a minute that it could be less. Second, gas shortages caused correlated failures, a lot of power plants together with both spot and firm contracts. The spot guys did worse, but firm wasn't firm. And interestingly enough, the folks who played in the spot market in parts of the Midwest and the mid-Atlantic could have kept their power plants going if they had held firm contracts in those days. So I'll show you some more details quickly. These force-major events caused a maximum of 5% of the megawatt hours lost. And I say a maximum because what we did was we took the force-major event on a long pipeline like TransCo, and we said, if it happened any place on the pipeline, the gas plants, even downstream and upstream, were affected. It's probably less than that. My guess is somewhere around 3% of the megawatt hours. The second was here's the firm and the spot market plants. Notice out here in the West, here are you guys, that these are the gigawatts that were lost due to reported fuel shortages, and the firm plants make up the bulk of those. Spot market plants still do it. It's different in different regions, mostly because who buys the spot market contracts. And it's not just the plants that run once in a blue moon, the peaker plants. The horizontal axis here is the capacity factor, the percentage of time they run. And these are base load plants, shoulder plants, peaker plants here. Out here in the West, the red ones are the firm contracts, and so it happened to every kind of plant out there. But there was available pipeline capacity in many places, and we'll get into how we did it, but there was basically space in the pipelines to move the gas. But was the gas to be had? Well, I work in a business school, so I look at prices, and the prices were pretty modest during these times. So they would have been driven up. There's this half-baked law of supply and demand. It's just a theory, kind of like an evolution, right? But they were quite modest prices, so there was gas to be had. And where would we have gotten that gas? Well, it would have come from commercial and industrial customers, which were affected by the power plant outages anyway. So they couldn't actually use the gas in many cases. So where were they? Well, the shaded areas here, not all of the country, but in certain areas there was gas capacity, and it could have worked. What about New England? As Daniel Webster said about his college, it's a small region, but there are some who love it. New England has only two pipelines coming in that still have gas. It had a third one from the Sable Island Field in Nova Scotia, but it's not working anymore. So it's got all of gas comes in from two areas here. And it loses gas quite frequently. So we said, how long are these outages? They're not very long. Most of them are less than a day. There's a few long ones. Given that, you can supply them with, we thought of two things, and Adam told me about a third one today. The two things we thought of were on-site compressed natural gas and dual-fuel oil. Many of the plants already have dual-fuel oil. Adam told me that there are people taking CNG tankers in and hooking them up to the pipeline in there. That could have worked too. But here's the cost. If you wanted to mitigate, let's say, five gigawatts of power, it would only cost you a penny a kilowatt hour for oil and for on-site CNG a little bit more, a few cents. But it's perfectly possible to do that. Finally, and I'm going to go a few minutes over if that's okay, what about the effect of electricity on the compressors that are necessary to overcome the frictional losses in the pipelines? So we have a very good student called Sean Smiley, who's just about to finish up. He came to us from the natural gas industry up in Canada, and so he's been working on this problem. A compressor station has a big compressor. Well, it could be electrically driven, or it could be driven from the gas in the pipeline. Many of them are this way. There's a few of them that are electrically driven for various reasons. First, about 10 years ago gas prices were really high. Second, if you put these in an urban area, you don't want the pollution. Now, there are also auxiliary systems, and these auxiliary instrumentation and so forth do have backup power, usually from natural gas if the electric power fails. This has caused a lot of confusion. Two of the big Department of Energy National Laboratories have screwed this up. They have said that any compressor station that has backup power for the auxiliary loads has electrical power, and so they come up with something like a quarter of the stations have electrical power, that isn't the case. These are way, way too big compressors to have backup power. They're 20 to 80 megawatts. You're not going to get a diesel genset doing that. So what we found when we looked at it closely was that a maximum of 10% of the stations and about 15% of the size of the compressors are electric or stations that have both electric and gas covered. So here's where they are. The yellow are ones that have more than 20% less than 100%, and here's our friend the TransCo pipeline, and here's some others. So some of them are mixed, and some of them have an electric station, then a gas station, then an electric station. These are every 100 miles or so. So what? Well, we did hydraulic modeling of a really simple system, the pipeline going into Florida from the west, and it has a few stations. This one's about half gas, half electric. This one's all gas. This one's mostly electric. This one's about 40% electric. So we said, what happens if you take out the electric stuff and just one of those, two of them, three of them, just one of those stations causes an outage larger than the largest nuclear plant in Florida. And they don't think about this at all. How could these stations lose electricity? There's never hurricanes in Florida, anyway. Okay, so we have reason to believe that that actually caused part of the outages during the polar vortex. Two of the stations that we identified as being electric did report outages at the time of that polar vortex. So this isn't just theoretical. Okay, so let me just summarize what we've learned so far. First, you can now use probabilistic methods to estimate reserves. Second, the grid operators ought to do it seasonally. Third, we don't know squat about the gas grid. The U.S. really needs some good standards and a NERC-like reliability organization. Large pipelines aren't bad. They're fairly reliable. We know absolutely nothing about the gas distribution system. A firm contract isn't firm, but firm contracts could have helped power plants in the Midwest and the Atlantic. New England can probably afford oil or gas or these tankers as backups. Adam tells me they're using the tankers injecting right into the pipelines. And electrical outages can cripple the whole pipeline system in ways that really aren't being accounted for. So, Inej and John, thank you very much and be happy to answer questions. Thank you. Thank you, Jay. That was terrific, educational and entertaining. I'd say on behalf of the audience. Well, it wouldn't be entertaining if the lights didn't stay on. Yeah, I'm surprised you didn't arrange that. But there was a question right here. Any questions from the audience? Got the brown blazer in the red. Yeah, and then there's one back here. Thank you. Thank you for the presentation. I mean, I'm a geek, so I love this stuff. Yeah, me too. My question is about the weekly procurement that you described. I think it's a clarification question. Are you thinking about this as a different type of capacity markets? Ah, interesting. So we don't really propose weekly. We did it so that we could do the net social welfare and weekly increments. We propose seasonal, just for a year. And yes, they'd be not only a capacity market, but ancillary service for spinning reserve and non-spinning reserve mostly. You need some capacity, but it's mostly the spin and non-spin reserves. Hi. So thank you so much. It was a great talk. I would like to get your comments on, in one of your conclusions you say that we need to prepare a little bit more for the reserve when we don't have like the gas outputs with the new regulations and everyone trying to move out from natural gas and try to incorporate renewable energy how you see the reliability of those resources in the energy. So it turns out it was only in the last three years that NERC required wind and solar to report to this big database. And so out of the half a trillion records that our student, Senate Murphy examined, there was nothing about wind and solar in there. That's really fertile ground. We know very little about what the specialists call availability. That is, you go buy a wind farm and three out of the 20 turbines won't be turning at all. Well, they don't get the crane out there just for three turbines because it costs $100,000 to get the crane on site. So what's the real availability look like? Those data are just really hard to get. And I hope that somebody in this room will get them because that's going to be absolutely critical. How often did the inverters fail in the solar fields? Why do they fail? We don't know that. I think it's a great new thing for somebody to do. And we're starting to get those data. Yeah. Can you hear me? Thanks for the talk, Jay. I'm curious if you see any evidence of cascading failure between the two systems. So Texas, for example, was there any evidence that hydrogen compressors were driven off-line thus causing feedback? Because a big fraction of the gigawatt outage was on the gas plants. Do we have a sense of whether there was a bounce back between the systems? We did a FOIA Freedom of Information Act to try and get those data and we were stiff-armed so far. But in PJM and the polar vortex, the next last slide that I showed, we think we've got pretty good evidence that that happened. That a compressor station got killed by electricity and it killed some gas plants downstream. But the next outage that happens, hopefully, you or one of your students will be on top of that and be able to answer it. Thanks, Jay, for the great talk. My question is around you mainly focus on like generation plants and stuff. Do transmission lines also have the failures or are they less significant compared? I didn't see a lot of data on the transmission lines. There was a huge ice storm that took out parts of Quebec for six weeks and parts of downtown Montreal for two weeks. About 20, 22 years ago, why did it happen? Because ice four inches thick coated the transmission lines. And there's beautiful, ugly pictures for your civil engineers of ruined transmission towers on the ground after that ice storm. So absolutely they do. And the data are out there. There is an analogous thing to GADS called TAUS, the Tee for Transmission. Nobody, including people at NERC, has ever examined the TAUS database. You should do it. Last question. I guess based on the Q&A I would add eye-opening to the many adjectives I used to describe your talk. That was just fantastic. Thank you so much for joining us. Thank you. And come visit us often. You're semi-retired, I guess. Yeah, semi. Yeah, lots of time to come out here. And it's much warmer here. Thanks once again. Thank you, John.