 Hi everybody. My name is Brittany VanDorff and we're going to wait about a minute for people to come into the room. So just hang tight. That was exactly a minute. So I'm going to get started. Welcome to our September webinar, Leveraging Energy Source Storage Resources to Improve Combined Cycle Power Plant Operational Efficiency with Dr. Wong from NMSU. I am Brittany VanDorff, the Communication and Outreach Specialist for New Mexico Established Program to Stimulate Competitive Research or New Mexico EPSCOR. EPSCOR is a nationwide program funded by the National Science Foundation. I'll be your host for today's webinar along with Isis Surna, our website administrator who will be working behind the scenes to make everything flow smoothly. A few quick housekeeping items I want to go over before we begin. I want to let you know that I want to let you know that if you have questions at any point, please type them into the Q&A box and Isis will politely interrupt Dr. Wong and read them out loud. I also want to let you know a quick plug for our next and last webinar in the fall series, which is Visible Light Communication and Applications in Smart Grid with Dr. Xiao from New Mexico Tech. Registration info can be found on our website. All right. Okay. I want to get to our speaker because he's got some really great stuff to share with y'all. Anyway, with that, I would like to introduce our presenter for today, Dr. Wong. Dr. Wong received his PhD in electrical engineering from Arizona State University and has significant industrial experience. He was previously employed at Mid-Continent Independent System Operator as a senior market R&D engineer. Much of his work has led to scientific publications and patents, most of which have been adopted into practice. He specializes in power system operation, electricity market design, electric energy policy, renewable energy integration, and energy storage. Thank you so much for being here, Dr. Wong, and please begin whenever you're ready. Thank you so much for Brittany for the introduction, and here I'm going to share my screen of the slides. Can I please confirm if you can see the slides? We can and we can see the list of the slides on the side too. Okay. So let me try again. How about this time? Is the slide coming up? The same thing. Okay. So I'm going to try again and just use the... How about this time? Yeah. Same thing. There we go. Beautiful. Okay. So let's get started. Hello, everyone. So today I'm going to talk about leveraging energy storage resources to improve combined cycle power plant operational efficiency. Okay. The scopes of these topics to optimally schedule the storage resources to minimize the short-term operational cost of combined cycle power plants. And the second scope is optimally size the battery energy storage to maximize the long-term profit of the combined cycle power plants. Okay. So because... So today we have the webinar for one hour, so I have the luxury to go through some background before we get into the model. Okay. So first, I'll talk about the generator cost curves. Okay. So the power is not free. It's not free. So for the thermal generators, they cost the fields. Okay. So cost is natural gas, coal, oil, they're all fuel, and they come at cost. Okay. And so you burn fuel, and then you produce the electricity. Right. So there are two important factors that impact the costs of producing electricity. So first is the fuel cost, how much the fuel cost is. Okay. A second one is the efficiency of the generator. And the generator has two important cost curves. The input output cost curve, which is the total cost curve. And the second cost curve is the incremental cost curve, which can be interpreted as the marginal cost curve. Okay. Here is example of the input output cost curve. Okay. So I borrow a graph from this book. So this is the example cost curve. So basically, this is the total cost if you produce this much power from this generator. That is the total cost. Okay. And the other cost curve is the incremental cost. It is actually the derivative of the total cost curve, the input output cost curve. So this curve is the derivative over P, and then I can get this incremental cost curve. So this can be interpreted as the marginal cost curve. Okay. And usually in practice, we do not use this linear marginal cost curve. Instead, we use the step-wise, the step-wise cost curve. And this is the approximated cost curve is convex and monotonically increasing. So this is usually how we model the terms and costs in practice. So we use the step-wise marginal cost curve instead of linear or even quadratic. The reason for that is RTU can only deal with piece-wise linear cost curve. So that is step-wise incremental cost curve. Due to the computational complexity of the RTU electricity market model. So for instance, a typical MISO using model, so it's a mini-continent independent system operator, the union model has over 50,000 binary variables and 15,000 transmission constants. And more than, it's around 1 million rows and the columns and the four to five million zeros. So if we include the piece-wise linear cost curve, it's also, excuse me, if we use the quadratic cost curve, they'll make this optimization model of the electricity market even more difficult to solve. So as a compromise, we approximate the cost curve using the step-wise incremental cost curve in the market. So this is the background. And so next I'll talk about the cost curve linearization. So this is a quadratic cost curve function for generator and how we can approximate into a linear cost curve. So we actually, we can use the segments to approximate the cost curve. And in this case, we approximate it to three segments. And we can approximate each segment using a linear segment. And we can calculate the slope of the segment. And the slope of the segment is actually the marginal cost for that segment. So we can basically, we can basically use, so the x-axis is the unit output and the x-axis is the cost of that output. And we can calculate the slope for each segment by using the tricky point. So for instance, if we want to calculate the segment's three-slope, we just need to use to Pmax segment three, which is this point. And use this, and this point is P-segment max-segment two. And then we can get the cost for these two. Then the cost difference divided by the megawatt difference. That is the slope of this segment. So that is actually the marginal cost if the unit is producing in this range. And then we can just sum up for each segment. We can sum up the megawatt output from each segment. We can get the total unit output. So we just need to simply add up P-segment one, P-segment two, P-segment three. And we can get, so there's also a minimum. So we need also a minimum. And we can get a total output of the unit. And you can observe that this cost curve is not decreasing. It's not decreasing cost curve. So we don't need to worry about that. So we don't need to worry about that. We clear segment three before we clear segment two, because segment three has a higher cost than the segment two. So the problem we'll always select is cheapest generation to produce power. And so for this case, so the generator will always clear in the segment with a lower cost. And then clear the segment with a higher cost. This is how we can approximate the nine linear cost curve into linear cost curve. And usually the RTO allows, at maximum, we have 10 segments. So the maximum segments you can beat as a generator is 10 segments. Otherwise, it's too competitionally intensive for the RTO to solve the model. Okay. So again, so we can just approximate the caudalic total cost curve by three piece of linear function. And then the slope is actually the stepwise. So the slope for each of the segment is actually the marginal cost for the generator for each step. But not all generators, they have the marginal cost like this. Okay. Some generators, they may have non-convex cost curve. So as the output increase, so the cost may decrease instead of increasing, it may decrease, it's non-convex. Okay. So example for that is the cement turbine generator with four steam or the mission of valves. All the time it's open new valve, the cost decreases. But when you open the valve, the marginal cost drop, but later on as you produce more power, the cost decreases. So this is not non-decreasing. This is actually increase and decrease and decrease. So it's non-convex. Okay. Another example is the comment cycle. Okay. So the comment cycle cost curve is, the incremental cost curve is something, it can be approximate like this. And this one is also not non-decreasing. It is, so it increases and decreases and increases again. So if we want to understand why the comment cycle power plant has such non-convex cost curve, let's get into the model of the comment cycle unit. The comment cycle unit consists of gas turbine and the steam turbine. This is, steam turbine is a gas turbine. Okay. Gas turbine actually burns gas, natural gas, and it consumes natural gas to produce electricity. Okay. And after the gas turbine burns the fuel and the fuel has exhaust. Okay. And then the comment cycle power plant actually reuse that exhaust. So the exhaust uses high temperature and it can be used to produce steam for the steam turbine. So actually the exhaust from the gas turbine actually gets recycled. So the steam turbine can use the exhaust from the gas turbine to produce steam and drop in the steam turbine. And the simple cycle gas turbine efficiency usually is 35 to 45, 35 to 40%. And the comment cycle power plant efficiency is around 50% to 60%. So the high efficiency is because of the steam turbine which recycled exhaust from the gas turbine. Okay. So why it has such non-convex cost curve? Okay. The reason for that is if we want, so if we want the comment cycle to produce power, the first thing we do is we need to turn on the gas turbine. And we will get the gas turbine out. So the gas turbine has a cost curve like this. So it is non-decreasing cost curve something like this. Okay. And once the gas turbine produces at some power level and it produces enough exhaust. So if the gas turbine produces enough exhaust and that exhaust is enough to drive the steam turbine and steam turbine will kick in because steam turbine here is like free power because it uses the exhaust from the gas turbine. So the marginal cost will decrease because the steam turbine starts to kick in and drives down the marginal cost. And then like we would maintain like the low cost for a while. And after that, the steam turbine reaches its maximum output because the steam turbine cannot produce more because of the capacity of the steam turbine. So after this point, if you want to keep increasing the output of the comment cycle unit, then the power has to be from the natural gas turbine because the steam turbine gets saturated. It cannot produce more power. So this is why the comment cycle power plant has such non-command cost curve. Okay. Comment cycle power plants, they are gaining popularity in today's part system setup because the first reason is they have higher efficiency compared to simple cycle units. Second, so they move very fast. So their rep rate is pretty decent. And third reason for that is it has very good operational flexibility. It can be turned on very quickly. Their minimum downtime, minimum uptime is relatively short compared to large coal units. And also another reason is the natural gas price is going down in recent years. So that's actually the reason why we have more and more comment cycle power plant construction in the United States today. So their marginal cost curve, let's go back to this slide, their marginal cost curve is non-convex. But as I stated in the previous slide, the RTO can only process the convex, it's stepwise, that's actually stepwise, non-decreasing cost curve. So basically, if the comment cycle power plant, they want to participate in the market, so they have to approximate this cost curve and make it stepwise non-decreasing, much in the cost curve and bid it to the RTO. So the turns and offer is not exactly match, it's not exactly match their actual cost. I will talk about that later. Before we get into the model, first thing I want to talk about is how the RTO works, the Regional Transmission Organization works. So RTO, they manage the extra market in the United States. Okay, so let's go through this RTO marketing process. The first step they do is called the further the reliability assessment commitment. So they actually turn on some must-run units. And these units usually for rehabilitated region, those units have relatively long leading time, more than 36 hours. For instance, nuclear plant, they have very long time to start up. And the hydropower plants, because they have other purpose, so you probably need to schedule the availability of the hydropower plants multiple days ago. And so this AFRAC has three days, AFRAC and seven days, AFRAC, so you can schedule the units long before the day head market. And once the unit is determined as the online, that means if you turn it on, and it will be labeled as must-run units in the day head market. And the day head market, they're actually, so day head market is just not one single piece. It actually consists at least three components. The first component is called Day Hascock. Day Hascock is a unit commitment problem, and it's 36 hours, multiple interval, and each interval is one hour. And the inputs are generation offers, the amount of bees, the amount of bees include fixed-bead and dispatchable bees. About 85 to 95 percent are the fixed-bead, only five to 15 percent of the bees are dispatchable bees, oppressive sensitive bees. And also you need to do the transactions between RTOs and external control areas, network topology including generation outages, transmission outages, and the consciousness, what consciousness you want to include in the model, what your bees, and the output is the unit commitment schedule. The unit commitment schedule is one from which unit is online, which unit is offline. And then after we solve the Day Hascock, we will pass the commitment decision like the online offline of the unit's decision to the next step that is Day Hascock. The Day Hascock is called Day Hascock Securities Consumption and Economic Dispatch. It's actually a surprising run of the day head market. So the setup is very similar about it, so it only has 24 hours. So it will price, so if you compare it, the commitment has 66 hours, but the economic dispatch only price for the first 24 hours. And the output is actually the other price including the unit dispatch, online peace, so those are the energy electricity price, credit reserve, and the marketing premium price for the reserves. Okay, and then we will go to the SFT, so they'll check if there are any security constraints evaluated. If it's evaluated, we will add those constraints back to the SCAD and run again until there's no violation. But in practice, for instance, you might say only one iteration is allowed because the time to solve it is very long, and as in England, so they can afford to run more iterations because their system is smaller. And after this day head market, and we will get into the real-time operation. In the real-time operation, maybe we have the intraday commitment. And so the intraday commitment usually is that look ahead commitment. So we will look ahead three hours and check if we have enough units committed. If not, we'll turn on new units to have additional capacity for that. Okay, and then we'll get to the real-time market. The real-time market will take the state estimation, the real-time plan output measurements and the RSE, which are the conscious input for the real-time market. The real-time market is 10 minutes. You solve the market 10 minutes ahead of time, and it is energy and a series-series co-incision problem, and a single interval, and each interval is five minutes. And so the only physical elements only problem, and there's no what-to-be is in the real-time market. And the real-time output of the, so the inputs of the real-time market is the real-time output of the units. So that is the initial megawatt of the unit, and the real-time generation offer. So they can update their generation offer every hour. Well, the intermittent resources they have, you know, privilege, they can update all for every five minutes. Okay, and the load is not low, demand be that. So the load in model is the 10 minutes ahead for cost demand. And network topology and the real-time active consciousness which are, which come from the RSE, real-time conditional analysis. And the output is the real-time unit dispatch, cleared reserves and RMPs and MCPs for reserves, right? So basically, so in the real-time market, we solve this market at five minutes, and for every five minutes, you know, the unit will have a new dispatch, and the new dispatch we will set it as the base point. So this is the, we'll update every five minutes. Let's call base points. It's from the non-dispatch. So if the unit is also cleared as regulation, so if the unit is providing regulation reserve, and they will respond to AGC signal, so AGC signal is actually called the set point. So basically, because, you know, the real-time market is solved every five minutes, but the load and renewables, they are moving around with even vaginas feminists. So they are changing all the time, the load and the renewable energies, they are changing all the time. So but you need to balance the load and the generation all the time. So in this, it's called SAP five minutes process model operation. So in that case, we use the, you know, the AGC is called automatic generation control to balance the, you know, the demand on generation. And this AGC signal is actually, you know, it's updated already four seconds. And so you add AGC signal on top of base point, that's the set point. So then, if the unit is cleared as the regulation reserve, and it will respond to the set point. Okay. So with all this background, let's talk about, you know, the model I'm going to talk about today. It's the Kaman cycle power plant and the energy storage co-optimism modeling. So first, suppose that we have the Kaman cycle power plant here, and if we invest in the energy storage at the same location at the Kaman cycle power plant. Okay. Then, so suppose this is the point of the interconnection. Okay. So the RTO does not care about, you know, where the power come from. So it can come from the, you know, the Kaman cycle power plant can come from the energy storage. So they care about, you know, how much power like you give me. So basically, so I can use the energy storage to leverage, you know, the Kaman cycle non-Kamanx cost curve. Okay. So basically, if for instance, if the RTO asks the Kaman cycle power plant to produce 100 megawatts, and so the Kaman cycle can produce 95 and the energy storage can produce five megawatts, but in total it's still 100 megawatts. So then it's still good. Okay. So how we can leverage, you know, the energy storage to improve the operational efficiency. So for instance, for this Kaman cycle power plant. Okay. This is called the economic operational range. So you want, actually you want the Kaman cycle to stay in this range because it's cheaper to produce energy in this range. But sometimes the RTO may ask you to, you know, produce here and produce there. Actually, you can use, so if RTO asks you to produce right here, but you can still use energy storage to, you know, to produce additional megawatts and, you know, still it can produce here, right? Excuse me. So if RTO asks you to produce here, okay. So you can let the Kaman cycle produce inside, but at the meantime, you get the energy storage to charge. So the charge, that means a negative generation, but in total it still gives you, still gives you, you know, the total output of these two assets still give you, you know, this much power. Okay. So actually, we can take the advantage of the energy storage to that the Kaman cycle power plant to stay in its economic operational range. Okay. So next, let's take a look at the energy storage model. So energy storage model, it has several questions. Okay. So first is the state of charge variable. So that is how much energy stored in the battery. Okay. And it is minus the ET minus, ET minus is the discharge rate. And the ET plus it is charging rate. So that is, so that is the, it can be interpreted as the output of the battery storage. And it is equal to the SOC zero, SOC zero is the initial, a state of charge, initial like energy level in the battery. Okay. And so the, the E plus and the E minus, they are the maximum charge and discharge rate. Okay. And the UT is the, it's the binary variable. It is in charge mode, it's zero. And it's discharge mode one. So, so it's battery, excuse me, it's a binary variable. So it's, it's to avoid, you know, the unit is charging discharge at the same time, which is not possible for the, for the, for the, for the battery. And SOC is has a, you know, a maximum capacity and minimum capacity. So, so, so as SOC bar is actually the maximum level of the maximum, maximum level of energy the battery can store in it. Okay. So next is the unique cost in the magician model. Okay. So, so this is used to minimize the cost, the cost of the, you know, the, the general, the chemical or power plant, the cost of the chemical power plant. And so the PT plus E, T minus, minus ET plus is equal to PT. So, so this actually, you know, is the, that means, so the ET minus, minus ET plus is, so this is, this is the output from the, from the, the battery. And then the PT is, PT is the unit output, you know, from the chemical power plant. And the P, the, the capital P, the uppercase, you know, P time, time t is the RTO SOC, the real time dispatch target. So that means, so come, if you combine the output of the, the chemical power plant and the energy storage output, if you still give you the same, you know, the dispatch from the RTO SOC, the dispatch target. Okay. And the PT is equal to the, you know, this, this is, it is actually, you know, the, is from the different segments, from the different segments. So, so this is the, for each segment, you look at here, there's a, for each segment, that determines the, how, you know, how much power like you cleared for each segment. And now here I, I put a binary variable. So it's the U segment KT. So that is commitment of dispatch segment K. And, and I impose this conscience to avoid, so avoid that you, you, so you must clear this segment first, and then this segment, then the sector segment. So you cannot clear, you know, the first, then the, then the third one, even though the third one is tripled in the second one, it still cannot, it still cannot, you know, clear the third segment before the, you know, the second one. So, so this, this conscience is actually enforced, you know, the, once you have the non-convex cost, then you all, you clear, you know, the, you know, the segment in sequence, not based on the cost. Okay. And then I put in, you know, the under-storage conscience. So this is basically, you know, the, the unit to cost to minimize this model. So with, if you have the under-storage, you know, co-located with commercial power plant, basically you can, basically you can minimize the cost, you know, so we can take advantage of the non-convex cost curve. The next, we will talk about, you know, the numerical results from that. Okay. So, so the tax case is, it's a taxes 7000 bar system, and so it's, it's taxed from this, it's a synthetic data, and it's, it carries the geographical footprint of the ERCOT. It's a tax, taxed RTO. And the number of buses and model here is 6,717 buses. There, okay. So number of branches, this is around 9000 branches, I forgot to put in there, the number. And peak load is 74 gigawatts, peak load. And the database here are high quality sensitive electric grid models built from the public information and the statistical analysis of the actual power system. And the software I'm using is the ASA Ames, 4.78, and you can see the, you know, the, the interface is here. And software I use is C-Plex, 0.1. And the CPU I used to solve the problem is Intel S7. The sensitive cutoff is the 0.05. And in input setup, it's considered a renewable incentive, you know, 10 to 40% of incentive for different nodes, a low incentive street to 5%, you know, in real time. Now test days, I stick to four different test days, summer peak day, summer off peak day, winter peak day, winter off peak day. And the, the round trip efficiency for the battery and storage is 85%. Okay. And the, the, the, the Kamazekopopon, you know, I'm looking at is, it has the minimum output is 65.35 megawatt, the maximum output is it's 590 megawatt. And it has two combustion turbine, one steam turbine. And you can see here is the, the cost curve. And, and the, so, so this dot line, you see, this is, is, is, is there submitted a generation curve, which is, you know, this stepwise, non-decreasing cost curve. But, but the, the solid blue line is the actual incremental cost curve for the, for the generator. So you can see there's a large gap, you know, for, for this range. And actually this, this range is the economic operation range for the Kamazekopopon plan. But the RTO does not have that information because what RTO, like the electricity market, market, you know, operator says is, is this cost curve? And the, you know, the RTO does not know, okay, you are more efficient in this range. So they will probably just dispatch you, you know, you know, here or here. So, so we can use under starage to, you know, to that, the unit to stay in this range and to, you know, lower the optional cost and improve the efficiency overall. And here is the, and also another important scope in this project is, you know, how to optimally, you know, set the under starage. And then here, so we can choose, you know, the optimal size of the under starage. So the optimal size of the starage actually goes here. So this is, you know, how much, you know, the energy that you can store in the, you know, in the battery. Okay. And, and then we minimized, you know, the total operational cost for the, for the unit. And the S is, you know, it's different as it's actually for the different scenarios, scenarios that we have, you know, different days, different days, and we have different scenarios for the, for the renewables as well. Okay. Sorry, too much. Okay. And then we can use this program and to solve, you know, what is the optimal size of the under starage. And also the RFI, I first mentioned the RFI is the skilled battery cost is dollars per megawatt per day. And including the investment cost, every gains and costs, you know, depreciation, depreciation costs exactly. Okay. So next we'll talk about is the, so it's a from our reports. Okay. So, so you can see, so this is how much, you know, the battery cost per kilowatt hour. So this particular hour, you see in a high range is, is, so today is like $350 per kilowatt hour for under storage for battery storage with them for hours, the recent storage. And so by, in a high, you know, range is at, you know, by 2030 is $250 per kilowatt hour. And in a mid range is around 200 in lower ranges, about, you know, 150. Okay. And in the future, in the future, it may reach to, you know, $100 per kilowatt hour, excuse me. Okay. Now let's see, you know, so for that particular unit, you know, what is the optimal size of the battery under storage at different costs of battery. So the access is the, you know, the cost of, you know, energy storage. And so, you know, based on, you know, the, you know, the this, this option, you know, programming and different scenarios, I select, I think it's more than several thousands of scenarios from study. So this is actually, you know, the, you know, the optimal size of the, you know, for that power plant. For instance, if, so if the price is $100 per kilowatt hour, the optimal size of the, you know, so what is the best size, you know, the power plants should invest, you know, that is co-located, you know, with the commercial power plant. Okay. It's around, I think it's around, you know, eight or maybe 12, I think it's 12 megawatt hour. And so as the, so you can see, as the, so the cost of, you know, battery under storage increase. And so the optimal size of the battery actually, you know, decrease because it's more expensive. And it's not profitable if the battery costs more than $320 per kilowatt hour. So there, so if the battery costs more than that, it's not worth investing, you know, the battery in the same location as the commercial power plant. Okay. So this graph is for that particular commercial power plant. It's based on, you know, the full sample days and each day has 100 scenarios based on the consideration of renewables and road incentives. Okay. And here is, you know, it's a size of the battery. It's a size of the battery. You can see it's from, from like one to 200 megawatt hour. And this is the average reduce the fuel cost per day. So you can, so the larger the under storage is the more, you know, cost savings you can have, the more efficient that you can operate the commercial power plant. But you can see the marginal benefit is getting smaller and smaller as, you know, the, the, the, the, the size of the under storage increase. Okay. The next is called the net cost benefit. The net cost benefit is the, the net cost is equal to the reduced cost minus the battery cost. So this is actually the reduced cost. This is the reduced cost. So if you have the battery, you know, this, this size of the battery co-located with your commercial power plant. And then so you, so you can actually take advantage, you know, the under storage to make your commercial power plant operate in the, you know, close to your economic range. And, and, and what it will be the, you know, the cost savings. And the, that cost minus the battery cost that your net cost benefit. So if, so for this case, you know, the, if the, the battery cost is $300 per megawatt hour. And so the, the, the, the net benefit is about $100 per day. And so the annual net profits for this other dollars and the 10 year net profit is no around $40,000. Okay. Another case is if, if the battery cost is $200 per kilowatt per kilowatt hour, and the, the maximum, so, so, okay. So this will be optimal battery, optimal battery size. Okay. And at this size, you can get, you know, $320, you know, net profit today on average, you know, I'm considering, you know, different scenarios in different sampling days. And as you can see, $300 per day, but that is the, you know, the, the, the net profit for that is around more than $100,000 per year and no more than $1 million 10 years. Okay. So this is the, you know, the unassisted battery battery cost in the, in the, in the next five years. And another case is if we, we have, you know, you know, very good progress in reducing the cost of battery. If the battery cost, you know, around $100 per kilowatt hour. And so if you invest around like eight, it's like eight megawatt hour, you know, co-located with the Kama cycle power plant, you know, the, the net profit will be around $550 per day. And the return for that is like $200,000 per year and over $2 million for 10 years. Okay. That is pretty good return for that. Okay. So let's take a look at, you know, so, so this is, I, this is a case I also demonstrated, you know, what is the difference of the RTO dispatch and dispatch with the understorage? Okay. So, so the blue, the solid blue line is the RTO dispatch. And so the dispatch base, the understorage is the orange line, other orange line. So, so this dispatch is, so it's just the Kama cycle power plant dispatch. It doesn't include understorage, because if you include understorage, that should be exactly the same as the RTO dispatch, because you still need to meet, you know, the RTO dispatch. But in here, you know, you can see the, so the understorage can, your charge, you know, this is charge and discharge, charge, discharge. And so, so, so you feel some, you know, the understorage schedule with these orange lines give you exactly the RTO dispatch, give you the RTO dispatch exactly. And you can see, you know, so this, this, the two, you know, dotted lines, you know, between this is economic, you know, operational range. And you can see that, you know, the orange lines, you know, they are actually closed to the range. You can see the, so the, for the blue line, the RTO actually asked it to, you know, to produce here. But, you know, if you have understorage, you can, you can actually produce lower, because you can have the understorage produce the rest, you know, of the, the power, but it still meets, you know, the RTO dispatch target. So you can see, so you can see overall, you have the orange line like close to, you know, to the economic, you know, operational range. And in this part, the, so the green line is the real time RMP. And you can see the understorage is actually not, it's not following the press, you know, it actually is pressed. Instead, it tries to keep the generator, it's closed, you know, to the economic option range. And, and today, you know, one difficult challenge to manage understorage is, you know, is you need some like, you know, RMP forecasting to manage the understorage. So you actually understorage can just by low, by low and a sell high, right? I'll be tried in the market. But in, in, in this setup, you don't need to follow the price because our goal is to keep the generator within its, you know, within, you know, its economic operational range. And, and, you know, in the, in the market, it's very difficult to forecast RMP. And in that way, you know, it's still a long way to go. If you want, if you want to manage the understorage, a result, you know, very good RMP forecasting. So you may not be able to, you know, get a good profit. But for this, it's like, you know, some guaranteed profit because you see this. So, so the strategy is very simple. It's just, you know, it just charge and discharge to keep, to keep in the, the, the, the pharmaceutical power plant dispatch, you know, as close as possible to its economic operational range. See, so, so, so this case is if we have to understorage capacities, 8.33 megawatt hour, but if it increase the capacity, you can see we can keep, you know, keep the, the dispatch of the pharmaceutical, you know, even closure to the economic operational range. Okay. And, and it's, and you can see here is the, okay, here is the, the, the charge and discharge of the understorage units. So basically, it's, again, it's not following the, the RMP. You can see there, so there is a RMP spec here, but it's still discharging because it's still discharging because it's trying to, you know, keep the, the units, you know, operate, operate at its most economic range, not, not following the RMP. So that's why, so for this case, the, it's much easier to manage understorage because you just, you just need to, you know, to keep the, you know, the pharmaceutical power plant to operate at, at its economic range, you know, as close as, you know, it can be. Okay. The conclusion is that the understorage can improve the operational efficiency of generators with non-convex incremental costs such as pharmaceutical power plant. And second is gathering of understorage does not need RMP for costs. Instead, the strategy is very simple. It's just to keep the pharmaceutical state as close as possible in its economic operational range. And the battery cost has great impact on the optimal size of the, of the battery. So if, so you basically, if the more expensive the battery is, the smaller size you want to invest. And at some point you, it's just not worth to invest in it. Okay. And the results are dependent on the system and the generator parameters, but the model is generally enough to expand to any generators with non-convex cost curves. So it's, it's kind of, so this is just demonstrated as an example for that particular pharmaceutical power plant that can be, you know, expandable to other power, you know, pharmaceutical power plant or the power plants with non-convex costs. But, you know, because we want to bring the, you know, the gap that the actual cost curve and the, the cost curve, you know, that you procemet it, that you bid, you know, to the, to the electric market. Some future work. So we're going to simulate more testing days. And one of our study, you know, the energy storage impact on greenhouse emission of the CCP. So, so, so in some of the range, you know, it's not only, you know, the, the cost is lower, but also the, the carbon emission is also lower in, in this range, basically. So, so next step, we will, we will study, you know, if we use the knowledge for this purpose and how much greenhouse emission gas we can reduce using this strategy. And impact non-regulation reserve provided by the pharmaceutical plant with co-located energy storage and the resulting wear and tear cost. So if the, if, so the chemical power plant, they usually, they, so they are very flexible units and they provide the regulation reserve. And the regulation reserve, as I talked about in the, in the previous slide, in the, in front of this here, it didn't move up, down, down like, you know, all the time. And it can cause additional wear and tear cost. And the wear and tear cost is actually the maintenance cost. So you, so the generator will probably, you know, need more maintenance if, if, if you move, you know, the, you know, the unit a lot. But if you have inner storage, you can use the inner storage to, to move instead of, you know, that the physical unit to move. Okay. And the first one is the optimal beating strategy of the chemical power plant with co-located energy storage. So if you, if you take a look at, you know, this cost curve, even though the real cost curve is like this, but we can still, you know, make, you know, optimal beating strategy, but it's still stepwise, not decreasing, but, you know, we can maximize our, you know, the journey to profit and, you know, use another. So what is the, you know, optimal beating curve, you know, for the, for the unit, if, you know, it has an inner storage and what is the optimal size for that? Okay. So that would be all the future work. And that, that's all to acknowledge. First off, I'll say thanks to my undergrad students, Ryan and Bob, and Hmerdo for their work. And they worked with me in the summer on this project. And I want to thank you to the New, New, New Mexico, Consultium, and Los Alamos National Lab, and the New Mexico AppSchool NSFRU program for their sponsorship on this project. And also just thanks to Brittany and the SES for the coordination of this webinar. Thank you so much. And thank you all for your attention. And then I'm open to any questions. And if you have any questions, please tap your questions in the chat box. Thank you. Oh, thank you. Oh, that was, that was awesome. Thank you for, for the recognition too. We all really appreciate it. So like you said, please type your questions into the chat or the Q&A box and you would be happy to answer them. While we're waiting for those to come in, I was wondering, you've seen, you've seen a rise in popularity for CCPM, CCPPs, but I'm wondering, are there any like local New Mexico examples you can point to that have really gone on board with this? I saw that you, you cited the Texas. Yeah, so, but yeah, because we are close to taxes. That's one of the reasons why I choose the Texas system as the tested case. Yeah. So this is, so this is the, so I forgot to put the branch in number, branches around 9,000 branches. So, so this is the, I use the Texas, you know, system to produce that the real-time prices, you know, for the generator, and, and to produce like the real-time dispatch for the generator as well. So you can see, so all the, so all, so this is like the archeo-brack, you know, the dispatch is, you know, it's tested based on the Texas system. Well, yeah, that's, but as it looks like, and that's it, that's, that's good to know. I'm hoping that maybe this will be applicable to our other project members as well, or you can use their data. But enough about my questions, you've got two from some people in the chat and it looks like Dr. Leo. It says, it seems the benefit from storage to improve the efficiency of, oh, just CC power plants is RTO system dependent and the CC power plant dependent. What is the challenge for combined cycle power plants to accept the proposed solution? Okay, so that's a good question. So basically you need to perform this, you know, cost-benefit study before you invest in any, you know, energy storage asset. If you make bad, you know, investment decisions, actually you can end up with losing money. So you, so first the thing is you want to even know what, okay, what is the battery cost and what is the aggregation cost and order cost and you include it. Okay, so what is cost for the battery? And second is how much cost savings you can get from, you know, using, leveraging this energy storage, right? So, but this, so that is true. This is system dependent, you know, this is journal dependent. This depends, for instance, depends on, you know, so what is gap between, you know, the, you know, this not, so if this cost is very close to this carbon, there's not much room to improve, right? But if you see this one has, you know, a quite, you know, quite a bit of gap between, you know, the true cost and the cost that the curve that beat into the system, then they have more room, you know, to leverage the energy storage. So, I would say, so each power plant should run such optimal energy storage site information and to decide if it is, you know, worth to investing in energy storage. For instance, if the energy storage, even for this kind of cycle power plant, if the energy storage, you know, costs more than $320 per kilowatt hour, it just doesn't worth to investing any energy storage assets here. Okay. So, we've got one, we've got one more question and then we'll wrap up in a second, but I wanted to get to this question. Weeping says, is it possible to incorporate a data-driven prediction component in this analysis? The data-driven XO PONT can use the historic data component. Can use it, but I am not sure whether those data are available. Okay, I got that. So, I think definitely, yes. So, remember this PTS, so this is the, you know, the RTO dispatch target for different scenarios of the unit. And for independent systems, you know, generators, they actually, they don't have, they don't have the, you know, the system information. So, in this case, we will assume like we have, you know, the perfect information of the system, but the generators, they don't have the perfect information of the system, but they can use their past, you know, historical dispatch targets in the past. And then they can use data-driven, you know, techniques to generate the synthetic dispatch target, like for the next 10 years, you know, stuff like that. And then, so they can put that, those target here, here. Otherwise, you know, they have to run the, you know, the whole system and generate their dispatch, which is not possible. So, the data-driven technique is absolutely, you know, very useful in this case, you know, if the generators, they know their past, their past, you know, the dispatch target and they can use their historical, you know, the dispatch target and generate some future scenarios. Cool. Thank you. Thank you. All right. So, I'm going to take, I'm going to wrap us up really quick. And for you too. There we go. Thank you. So, on behalf of everybody who's attended and the State Office, I want to thank you once more for being so generous with your time. As everyone here likely knows, renewables are expanding rapidly in the energy sector, especially in New Mexico, and CCPPs may be one way to relieve the variability introduced by their intermittent nature. So, once again, thank you. Thank you for being here and presenting with us today. Yeah. Thank you for hosting the webinar. Thank you so much. And thank you for all the people who joined this webinar. Thank you. Yeah. We are all honored. Really quick though, before we sign off, I also want to thank my partner in crime is Isis Surna. And don't forget to join us in October for the final webinar of the fall series, Visible Light Communications and Applications in Smart Grid by Dr. Xiao. Until next time, have a great afternoon, everybody. Thank you. Thank you.