Loading summary
Transcript102 lines
- [00:00]
A
I'm your host, Ed Porter. Welcome back to Transmission. People think the secret to making money from a battery is predicting the price. It isn't. Ask where the value is and you'll hear forecasting. But that's not everything. The real edge is positioning an asset so it profits whichever way prices move. Brian Lon is head of flexibility at Statcraft, which runs around 4.5 gigawatts of flexible capacity in GB. This is optimization from the inside. Here's how the money actually gets made. Before we start. If you want to see how GB battery revenues are stacking up across the day ahead Intraday and balancing mechanism, ask co Moto Energy's AI analyst trained on GB power market data and built to answer exactly these questions. Link in the description. Let's jump in. Hello Brian. Welcome to Transmission.
- [01:00]
B
Hi Ed, thank you very much for having me. I'm really excited to be here.
- [01:03]
A
Our pleasure. I'm really looking forward to this one. So let's get straight into it. What is one thing that everyone gets wrong about? Asset optimization and trading.
- [01:10]
B
Yeah, I think when speaking to people who are less familiar with asset optimization and trading, there's a lot of focus on forecasting and forecasting driving a lot of the value creation. While forecasting does have an important role to play, we think it's not just about forecasting a specific outcome, but rather understanding the potential outcome space and how to position an asset so that no matter what the outcome is, value will be captured and then value can be captured by prices changing. So for example, at the day ahead stage, we're not necessarily trying to predict what an intraday price is going to be, but rather what the intraday prices could be and how we will position the asset day ahead so that we can capture value from prices moving up and down and continuing to evolve through to delivery.
- [02:01]
A
Okay. And for people listening to this, some people kind of when they hear the word forecasting, right, their mind will think to like 2050, 2060 as a long term forecast. But you don't mean that. You mean more like you're looking over the next 24, 48 hours like a rolling view of prices.
- [02:16]
B
Exactly, exactly. So we use the prices that are available in the market to inform our optimization decisions. And we're continuously optimizing the batteries every five minutes across the market prices that are available. But further away from delivery, there's more uncertainty on where the market prices are going to be. So for example, the intraday market, that only becomes liquid a few hours away from delivery. So we need to make estimates on where the intraday price could evolve for those periods until there's really a market price that we can optimize against.
- [02:49]
A
So, so I think that's making sense to me. There's also kind of, there's more to it. Right, so you've also said the forecasting for a price, but it's, it's kind of the intraday price. It's also maybe the day ahead price. Maybe you're thinking about that, maybe you're thinking about the balancing mechanism, maybe also thinking about the frequency response prices at the same time. So yeah, I imagine you sort of in your office, you're not spinning like one plate on the forecast. You're, you're kind of imagining five or six different themes that you're thinking about.
- [03:14]
B
Exactly. There's, there's five or six key markets, all of which have uncertainty and where the price is going to land. And then we run dozens of optimization scenarios in order to pick schedules that on average are looking to create the most value.
- [03:32]
A
Yeah.
- [03:32]
B
And then we've developed optimization techniques and tools which will allow us to capture value or be positioned for uncertainty. So for example, when we're bidding for ancillary services, the volume is not all bid at one specific price, but we can put in parent bid for a first chunk of capacity and then children bids where we can commit incremental capacity if the clearing price goes high and then it becomes optimal to do so.
- [04:00]
A
Yeah. And so your view about the thing that people get wrong is that they think that sort of all of the secret sauce is in the forecasting. Whereas what you're saying is actually the forecasting is really important, but what we have to make sure is that we keep ourselves buffers on these things or we put in conditional bids or offers, parent child bids, as you said, that give us flexibility. So if, okay, we think X might happen, that's our sort of expected, but if Y happens, we're still in a place where we can get good value from this.
- [04:27]
B
Yeah, that's it. It's, it's about also having the tools that allow for that value to be captured. So, for example, we've put a lot of effort into algorithmic intraday trading to have the assets in a position where we always have volumes in the market so that we can have those offers accepted as the prices move upwards, as opposed to reacting to prices moving upwards and later trying to put in our volumes. So it's about being positioned so that irrespective of the outcome, assets can profit from changes in prices as much as specific outcomes.
- [05:06]
A
I love this. I love that we're into parent child bids inside the opening five minutes. That's where I hope we're going to be. Well, let's just make sure that we do the broader piece. Right. Just for Statcraft more broadly over the. For the GB Flex portfolio, which is only sort of one part of what is within statcraft. From around 2018 to today, you've grown to about 4.5 gigawatts under contract. What does that change or what does that trajectory tell you about where the optimization market is heading? And I think one thing that's quite interesting for lots of people to think about, is there a natural ceiling to how much capacity an optimizer could hold?
- [05:43]
B
Yeah. So we launched, as you say, in about 2018 and then grew with the market. Now we have about four and a half gigawatts under contract across various technologies. Battery storage is the largest component of the portfolio by now. We also optimize reciprocating engines. We have hydro in our portfolio and we have some really exciting and innovative power to heat as well. And looking towards hydrogen electrolysis. So we have a broad range of technologies in the portfolio. We've developed our optimization platform to work with a variety of technologies as well. But I think what we're seeing is it really is a scale game. There's a lot of investment that's required to build and maintain an optimization platform like this. So we would expect that in the long run, it's. The space is going to be, you know, seeing maybe three to six key optimizers, because it does require a certain level of scale to sustainably run a quality optimization business.
- [06:45]
A
Because it's both like the software, but it's also having the humans as well that can dispatch these assets alongside whatever sort of AI or systematic approach you've got.
- [06:54]
B
Exactly, exactly. It's everything. It's experienced traders who deeply understand the markets, it's having good quants and developers who can take that market understanding and convert it into strategies that can be run more autonomously. All the IT teams that develop the infrastructure that this lives on. Yeah, there's a lot to it.
- [07:20]
A
The list goes on. And I think it's quite interesting because I often get asked, people say to me, oh, do you have a. When we've got sort of 200 or 300 megawatts in our. So let's say they're an independent power producer, they've got 200 or 300 megawatts under management, they might say, ah, Ed, should I. Should I bring this in house? Should I do my own trading and optimization and I kind of go, well, it maybe not like it could cost just to run the desk. Might be half million. A million? A million and a half. Do you have enough volume to make this work?
- [07:49]
B
Yeah, yeah.
- [07:50]
A
I think it's a critical thing that people often think, oh, actually I could just do that. And in reality, it's a lot harder than it seems. I'm going to come back to a question that I asked, but I don't think you necessarily gave me the answer to, which is, is there a natural ceiling for one optimizer? So you're sort of Circle 5 gigs today. If I said to you, Brian, you could be like 20 gigs from tomorrow, do you think your risk committee would say, oh, no, Brian, that's too much, we can't do it?
- [08:10]
B
Yeah, I think there's two aspects of that. There's the operational side being able to run the optimization for a portfolio that large, and then there's the risk side. So starting with the optimization side, because of how we've set up our systems to deliver a quality optimization asset by asset and have the end to end automation and transparency to also feed that into the intraday market, trace back trades to the asset they were generated for. I think we, we could handle 20 gigawatts and we already have about 200 assets that are live in the portfolio, ranging from 250 kilowatts right up to 300 megawatts. So I think from an operational side at least we would be equipped for such growth. Of course that's going to be dependent on individual optimizers. From a risk perspective, I think I would probably get a hard time from the risk committee if I was coming and requesting approval to write 20 gigawatts of battery floors.
- [09:14]
A
So not overnight, but you think maybe a more gradual thing.
- [09:18]
B
Exactly. I think the battery market should evolve and grow along with the growth of renewables. So yeah, over time, we would expect as the energy transition progresses and there's further renewable deployment, batteries will grow and we will grow along with that.
- [09:39]
A
I think to your point, you said if there are five or six optimizers in the long run, if we get 50 gigs of batteries, which is a scenario that may come through by 2050, then it's nice, simple maths for me, but it's about 10 gigs. But then it's also plus all the other forms of flex that you have, you've been managing as well. Okay, and let's. I want to get into one sort of very specific part of trading that I Don't get, I don't think gets talked about often enough. And that is retrading. So I think lots of people listening will sort of immediately be okay, what, what on earth is retrading? That sounds very complicated. So maybe could you just talk us through an example of like what someone retrading looks like and then we'll go on to some sort of extra questions around like how that's evolved in time.
- [10:22]
B
Yeah, yeah. So the assets in our portfolio, they participate in several markets and in each subsequent market we're looking to see if we can re optimize the position and extract more value. So for, for a given battery, we would start the day by participating in the hourly auction, selling a base profile for the battery. Then we have the ancillary services auction where we clear and now have certainty on what the ancillary services commitment will be. And from there we can re optimize our position from the hourly auction, including the half hour products that are available in the afternoon auctions. So that's the first step of retrading is going to be re optimizing across the auctions. So taking what you sold on an hourly basis and re optimizing it, including half hour granularity, once you have certainty on what your ancillary services commitment is, there's three auctions that we participate in and then, then we move into the intraday market. And this is where there's a lot of the focus in our optimization platform and a lot of value we think is in re optimizing from the day ahead across the continu continuous intraday markets. So one form of retrading could be we sell a specific profile in the auctions day ahead. But then intraday, because of forecast error on renewables, because of forecast error on demand, or because of trips on an interconnector on a large thermal unit, intraday prices change. So let's say the highest price period of the day moves forward. Well, in that case we would buy back the profile that we sold at the day ahead, stage the hour that we sold day ahead, let's say buy that back and sell intraday at a higher price. So that's creating incremental value beyond just the day ahead. Dispatch.
- [12:18]
A
Okay, so let's say, let's make this real, right? So day ahead you sold at 100 pounds at the, the peak.
- [12:24]
B
Yep.
- [12:25]
A
And then something trips, let's say a large gas plant trips or an interconnected trips, all of a sudden the system goes short for that period earlier in the day, let's say it was a, let's say we'd sold at 5 o', clock, but at 4 o' clock something trips and all of a sudden the price goes to £150.
- [12:38]
B
Yeah.
- [12:39]
A
You would then sell into the £150 and then assuming the market was imagining the trip would get resolved, that then you could buy back what you had originally sold at the deadhead market. So you would have sold £100, bought that back for a hundred pounds, had the energy available and you'd have sold that in at £150. And so if you're kind of following the maths, you'd be up 50 pounds in this theoretical example. And so that's like the super simple version of this.
- [13:05]
B
It's exactly, that's the simple version for batteries. It's just price is shifting from the highest price being in your example, 5 to 4 o'.
- [13:15]
A
Clock.
- [13:17]
B
Then there's other versions of this as well. So the intraday market, it is continuous and different products are more liquid at different time horizons. So there's four hour and two hour blocks which are generally liquid further away from delivery. So perhaps the first retrading option for a specific period is doing a block trade. And that's not a perfect profile for a battery because not many batteries are four hours and maybe part of that's already been sold, but there's still an opportunity to use maybe a four hour block to lock in some value.
- [13:53]
A
Yeah.
- [13:54]
B
And then as liquidity improve, liquidity improves closer to delivery. Then use the two hour product to unwind part of that position and re optimize in other hours. And then finally we're getting into the half hour products where then we can do the fine tuning and the final reoptimization.
- [14:13]
A
Yes.
- [14:13]
B
And, and because the market is continuous and it's constantly responding to what NISO is doing and what forecasts are doing, prices can trade up, they can trade down. So for a given half hour there could be a really widespread of prices that were traded.
- [14:29]
A
Yeah.
- [14:30]
B
Which gives us a chance to. Yeah. Maybe in this example sell at 150 at 4 o'. Clock. But then maybe in a different term of turn of events, if the trip was resolved early, then that 4 o' clock could come back down to 90 and then we could buy that back for 90 and still have the energy available to sell in a subsequent half hour. So we see churn day ahead to intraday within period intraday and then finally into the bouncing mechanism as well.
- [15:01]
A
Yeah.
- [15:02]
B
Where we could submit a pn, but then have NISO unwind that PN or take us to a new position, which would also be some, some form of churn trading.
- [15:12]
A
A PN being a physical notification. So like a commitment you make to deliver that, that notification unless you're told otherwise by the system operator.
- [15:20]
B
That's right.
- [15:21]
A
And I think one of the things that some sometimes people listening. So a few things a, if you're an aspiring trader or asset optimizer, I hope you're listening. This is a great insight into how a lot of this works. I think a second thing is what does this mean for consumers? Because I often think from an asset value perspective and is it good for storage to be doing these things and particularly intraday trading. And I think the thing that's worth sort of remembering is that when an asset is kind of trading around this to get hit in the intraday, you have to be the most competitive asset out there to be either selling or buying. And essentially what you're stopping the market doing is spiraling off to a very high price. So you know, remember you occasionally see in the news like, you know, x unit gets called at £3,000 per megawatt hour because this thing happened in the market. Well, the great news is that now with more or less 7 gigawatts of batteries on the system, we don't see that so often because actually we've got these flexible units that sit in intraday and as soon as we see a bit of a spike, the batteries shift their strategy and they say, actually you know what, I can do that. I can react to this. And so from a consumer perspective, I think that's quite a good story.
- [16:34]
B
Yeah, absolutely. I think everyone wins when the markets become more efficient. And that's what bringing this additional capacity and flexibility into the market does. It makes the markets more efficient and yeah, benefits everybody.
- [16:47]
A
Okay, and you've been thinking about this since 2018. Has your approach to sort of intraday retrading changed at all?
- [16:53]
B
Yeah, it has certainly evolved over the time. So when we launched this business, StarCraft already had established algorithmic intraday trading tools that have been developed, you know, kind of around 2013 or so. So we had an algorithmic trading platform already available to build upon. We went live with our first asset in autumn 2018. One week later, Apex had the new version of the exchange rolled out in gb and we were also using using that for algorithmic trading for that asset. We've started to bring in more products. So initially we would only be trading half hour products. And then over the years we expanded to do more with the block products. As the markets evolve, we've seen that intraday markets in GB have also become more liquid. So when we started this business there was only decent liquidity, maybe in the next three half hours. Now that's changing and the spreads are getting to be fairly reasonable going out maybe six to eight hours. So that gives us more, more opportunities to trade and retrade because this, we're not losing too much value from crossing the spread in later periods. We also have looked at how different strategies perform around where we position our volumes in the merit order and in the order book intraday. So generally we, we aim for passive execution, which would mean that someone is crossing the spread and coming to us and allowing us to keep that or the flex assets to keep that additional value in the spread.
- [18:44]
A
And when you say crossing the spread, that's the bid offer spread. Right. So the gap between what's, what someone is willing to pay for it and what someone is selling it for, let's say it's 20p per megawatt hour or something.
- [18:55]
B
Exactly. So we, yeah we, we aim that in that example if we're looking to sell, it's the buyer who comes and accepts our price as opposed to the other way around. Because that, exactly that, that 20p and sometimes that's 1, 2 pounds, that's value that is then captured by the asset in the setup. So yeah, we've been, been working on strategies on how we position the volume, how we manage an increasing volume in the intraday market as well. Because you know, our portfolio has grown from 20 megawatts to currently 2 gigawatts active. So at times that can be a lot of volume to try to get sold in the intraday market. So we have to understand really deeply the dynamics of the intraday market and how to get this volume effectively sold at a fair price in the market.
- [19:49]
A
I think, I think we took, we did some work on intraday retrading a few months ago now and I think we found that sort of going back to 20, 23, 2024, it was more like 15, 15, 20% you could get from intraday retrading. I think that number has dropped a little bit recently just because of how volatile the market has been and the opportunity that's presented itself. Is that the ballpark that you see intraday retrading in or is it higher or lower?
- [20:14]
B
So I think we have to be very clear what the benchmark is here. So if we say that we have a two hour battery and the reference is only doing a day ahead Dispatch an optimal day ahead dispatch and comparing that to the value that we would be captured by re optimizing across the subsequent markets, I would say that the uplift is more like, you know, in most cases 50% or more with a range of probably 25 to 100% across a given month. Yeah, and that would be for a battery which is only doing wholesale trading. Now. That's not where the market is today because even though there's been a quite a build out of batteries, there's still a lot of room in ancillary services for the existing fleet. So if you consider that batteries are getting a fair amount of value from ancillary services from the balancing mechanism, it's not necessarily 50% incremental value from day ahead to intraday reoptimization currently because there's other value coming from ancillary services. But the market is evolving. The ancillary services markets are not necessarily growing at the same rate as the installed base for batteries and therefore more batteries. Batteries will be earning more of their revenue from wholesale markets and as batteries get longer, there'll be more of a focus on wholesale market optimization. And this is really where we think things are heading and why we've made these investments very early on and always set out with the view that wholesale trading is where these assets will land and the entity which can do the wholesale trading most effectively will, you know, create the most value.
- [22:12]
A
And I suppose the thing that you've also seen since 2018 would have been the balancing mechanism like really increase in terms of the activation for these batteries. I imagine you'd have been sort of going, seeing the frustration in the early days and now seeing batteries being dispatched more and more. How do you see that from the Stackcraft side?
- [22:27]
B
Yeah, it's been a very positive journey. I think, you know, NISO has a tremendous challenge, or had a tremendous challenge in integrating all these smaller units into existing systems and into the market. And I think that by and large they've done a commendable job of in, in how they've implemented the open balancing platform to make dispatch more efficient and more merit order based. Yeah, we, we started bringing assets into the balancing mechanism probably about five years ago and you know, the, the, the first wave was reciprocating engines and there was teething issues with that, issues with skips. A lot of what we saw in batteries we also experienced with the gas peaking portfolio five years ago or so. But then that got to a good place and what I think was really good to see was that when there was Sufficient volume of a technology, at least from the outside, from where I was sitting, it looked like it maybe gave NISO tools to manage the grid differently. So if they knew that there was several hundred megawatts of reciprocating engines available, maybe they could do things differently and not necessarily turn on large thermal units because there was a lot of fast acting flexibility that was available but would only be called on if it was actually required, as opposed to paying all that cost to, to synchronize a unit for margin which might not end up being needed. So there's been a good journey that we've gone on with distributed flexibility in the balancing mechanism. That's definitely continued with batteries. We're seeing quite a strong uptick in how batteries are being dispatched in the BM and being done in merit order. But I think there's still probably room to go as well as, as there's more batteries, battery capacity available in the bm, then the whole system can probably be managed differently.
- [24:32]
A
I really like that, I really like the sort of strategic elements, so the balancing mechanism for those sort of less familiar with it, it's almost like that last hour when you pass control of the system over to niso's control room and then they take decisions for system security and to make sure the grid is balanced. And as you're saying, if you go back, say 10 years, you maybe didn't have the same number of gas reciprocating engines or you didn't have the same amount of batteries available to them. And if NISO now know that they've got, let's say, 10 or 15 gigawatts of flexibility that they can cool on, then they can take different decisions. Because maybe one thing that people often get wrong about the balancing mechanism, as I've just described it as, is that it's the last hour, but also they take strategic actions ahead of that time because they have to warm up certain units. And so you're kind of saying, well look, as they start to get more understanding of how this fleet works, they could maybe do fewer things like warming up larger thermal units because they can rely on, they know they can rely on this flexibility to turn up when it's needed.
- [25:31]
B
Yeah, absolutely. And there's been some ancillary services development alongside that, which I think also gives some interesting tools and hopefully we will see some growth in things like balancing reserve, quick reserve, slow reserve, but especially balancing reserve as a mechanism for, for NISO to ensure that there's capacity available in certain time windows.
- [25:55]
A
And that balancing reserve is essentially when NISO pays someone to be available in the balancing mechanism ahead of time. So they're sort of securing an insurance option effectively.
- [26:04]
B
Exactly. And that comes at an opportunity cost for the asset owner, which we have to evaluate when we're bidding into the balancing mechanism or bidding into balancing reserve or quick reserve or slow reserve, basically any ancillary service. So we have to look at is there going to be a cost to deliver that service? In some cases there will, in some cases there won't. And what would I give up if I make a commitment to the system operator that I'm going to maintain headroom in this period? And that usually comes with an opportunity cost which needs to be evaluated.
- [26:38]
A
We have spent a while talking about retrading and it's incredibly complicated and I think unless you spend a lot of time in some of these trading desks, I don't think you really ever understand it. So how do you prove to your asset owners, who frankly have other things to be worrying about, that they are sort of seeing the benefits of that retrading? How do you prove to them that it's real and that they're seeing sort of a fair rub of that retrading?
- [26:59]
B
Yeah, absolutely. That's a really good question. So the way that we prove this to the asset owners is by providing a full trade log of every trade that we've done for the asset. Because of the automation that we have and the end to end systems, we can trace back specific volumes for the assets that we traded those against. So at the end of the day we, we produce a trade log, we show all the trades that we did, how volumes were bought and sold across different markets. Those are timestamped, so asset owners can look at the price that we allocated for the trade or that we actually captured for the trade. And if they were so inclined, compare that to tick data from the Apex exchange and see where the market was, where the best bid, where the best ask was at the time that those trades were done. They can see the trades that we were entering into from the Apex data and match up trades that were done well, that we tell them were done from their asset. Two trades that were visible in the market.
- [28:10]
A
And this is the, this is the only way you can do it, right? You can't. Otherwise it feels like how do they know that they're getting a fair view of that churn that happens?
- [28:19]
B
Yeah, absolutely. We think this is really important because different assets will capture different prices across the intraday market. And that's just because different assets have different level of flexibility. And different constraints. So for example, we would expect that longer duration assets on average are going to be capturing better prices in the wholesale market. The reason for that is they have, they just have more options. So there may be. Yeah. More flexible in how they can buy and sell volumes in given periods. It might be the case that, you know, certain trade decisions, if you just took an index price, because Apex does public published index prices add at, at the index price, which is the volume weighted average price, certain trade decisions would have not been optimal for some of the assets in the portfolio. However, it's a continuous market which trades can trade up substantially from the index and then that trade does become optimal and we need to make sure that the, that value is properly ascribed to the asset that it was created for.
- [29:25]
A
Yeah. So to help people understand, you know, that that value can sort of up until delivery can bounce up and down, up and down, up and down. It's not just one final volume weighted price. That's not how the market works. You could, you could see a whole range of prices before then. So yeah, transparency is key is the message.
- [29:40]
B
Absolutely.
- [29:41]
A
Well, we've done the retrading piece and I think people will really enjoy or found it really useful to have got sort of behind the scenes of how people think about optimizing these assets. It's complicated but like it's critical if you want to understand these markets in GB or in Europe, you have to understand how this kind of retrading works because it's a, it's a key part of the value. I want to move on to perhaps more of the type of the longer term origination type work so more like working with clients to support their projects with flaws or tolls or perhaps swaps. There are several ways to do that these days. How do you think about these instruments and does Statcraft have a preference in terms of how you would do this kind of longer term contracting?
- [30:25]
B
Yeah, this has been a really exciting space in terms of the innovation that has come into the market around how to, how to support projects in getting financing. So yeah, early days it was mostly floors. Going back to 2019, we did some of our first first floor PPAs and this was good because it unlocked project finance and debt for these projects so they could get built and come forward to market. Usually we see floor arrangements as being mostly for the benefit of debt and project finance, but there's also been cases where floors are done to support equity investment as well. So if there's, let's say an infrastructure fund with a more conservative risk appetite they may seek to have a revenue floor to also support the equity investment into the project. And we have different forms of floors that are maybe more efficient to use for debt finance as opposed to equity finance.
- [31:29]
A
And so is floors your sort of preferred route?
- [31:33]
B
By and large, yes. But we also do like to use tools in some cases. Usually we would see a floor as being something that we could write for, you know, probably eight to 15 years to, to really support the project life lifetime for tools. We would be looking towards shorter term tenors for that. So maybe three to seven years and also potentially do a tool in combination with a floor. And what, what a tool does that a floor doesn't is it would let you lock in revenue which is closer to the base case. I think there's a, a view that of a floor provides downside protection but it doesn't provide much revenue certainty other than the downside. So it enables finance, it prevents you from the worst case. But tolls because you can locking something closer to the base case. Yeah, that's useful in terms of smoothing revenues for the asset owner and also supporting some additional debt early on in the project which is then paid off quite quickly. So yeah, we, we think tools are interesting and we also really do like the day ahead battery swap.
- [33:02]
A
Okay.
- [33:03]
B
This is something that we've done with asset owners. It's also something we've done with other as a structured product towards other market participants. We think it's a really interesting product. It's nice because it could also be done potentially financially outside of the ppa and we have it as an option in all our PPAs to use as a fixing mechanism. So that's a really nice product. The challenge that we found with that product is it can be hard for asset owners to value because a lot of the revenue projections are either not broken down across the various markets or if they are broken down then it's presenting the asset owner what the view is of for a fully optimized battery across all the markets and they can't back out what would be a comparable day ahead swap price? So that's the challenge that we've seen with asset owners and the day ahead swaps. But that's starting to evolve as well.
- [34:07]
A
Okay, so it's sort of three, three options. Yeah, maybe I might put some numbers on there for the, for the first sort of floors and tolls just so people get a feeling for it. So if your expected value of say to our batteries, maybe 90k per megawatt per year for the next 10 years, I would Say your floor is kind of in that range of. It depends what your profit share percentage is as well. There's lots of. You're going to have lots of caveats on this, but let's say it's in like the 35 to 50 range. As you say it's, it's further away from the expected value but it's helping to protect, protect the debt. And then your toll might be something like 20% off that number. So you're like maybe 25%. So you're in the 60 to 70k range. So closer to kind of the expected value. But still like there's a, there's a haircut for the optimizer. I think sometimes people just, just don't really know where those levels are. So it's nice to put some themes. Is that sort of the ballpark you see?
- [34:54]
B
Very roughly, yes. And, and then the day ahead swap that's probably going to be. Because that's maybe let's say in some ways more, more hedgeable, more understandable. That's yeah. Going to be done over a shorter time horizon but probably also with less of a discount to the expected as well.
- [35:17]
A
Okay. Okay, that's really interesting. And we obviously are also looking, we've got sort of FCA authorized indices at modo and we're quite interested in the concept that. So right now that swap is often done with a sort of day ahead spread in mind. We also think you can do that same swap around a modo index. And yeah, watch this space because by the time this comes out we'll certainly have more detail out there for anyone looking to, looking to trade this and could be a useful tool.
- [35:44]
B
Yeah, absolutely. That's. That space is something that we are, you know, we're actively engaging with, with you with prospective counterparties on. We think it could become really interesting. I think it's especially for, for portfolios. Yes. Yeah. For individual projects there may be some challenges but it's definitely a useful tool to be able to capture all revenue streams of a battery as opposed to just the day ahead portion which is the limitation on the day ahead swap.
- [36:16]
A
So maybe going back to your earlier example of let's say there's five or six optimizers out there who are the sort of, the best optimizing. They've got 10 gigs each. Maybe their risk committee says I'd rather we had slightly less exposure to batteries for three years. You could sort of find a new counterparty whoever has like a big wind portfolio, has a big retail portfolio and sort of would want Virtual batteries in their portfolio. They could sort of take some of the floating off you and give you some of the fixed back. And then from a risk perspective your, your risk committee is getting, getting a horrendous deal out of this conversation. I'm sure they're much more positive than this.
- [36:48]
B
Yeah, they're. Yes.
- [36:50]
A
Yeah, they're a good bunch. Okay. So. But, but if it affects me, that would be a way for you to sort of reduce some of your risk and then for someone else to hold
- [36:56]
B
some of it potentially for. Yeah, either for parties like us with large exposure, for asset owners with large exposure. That could be interesting and we'll see who, who ends up wanting to be on the other side of that.
- [37:10]
A
This sector, how it like handles risk is something I think we're only like the tip of the iceberg on. So yeah, exciting space. Okay. We. I would like very briefly just to touch on CO location. It feels like something that is big in Europe and has done a little bit in gb but it feels like sometimes it feels sort of forgotten in gb. How do you think about colocation? Are you seeing. And more specifically on the asset optimization for colocated sites because I know you have some. Do you think that it's. Do you sort of optimize each sort of asset individually on these colocated sites or are they sort of being co optimized?
- [37:49]
B
Yep. So yeah, colocation is a really interesting space. It's an interesting optimization problem. It's also an interesting structuring and finance problem as well.
- [38:01]
A
You have to be careful here because interesting to Brits means potentially bad. No, no.
- [38:05]
B
Yeah, good. Interesting.
- [38:07]
A
Good.
- [38:07]
B
Interesting. Exciting. Exciting. Let's use the word exciting instead. So yeah, we have had co located assets in the portfolio for quite a while. I think we signed the first purpose built co located PV and battery project several years back. We have a DC coupled project which is also really fun to work with and optimize. That's a four and a half hour battery which has also been really great to work with across the wholesale markets over the years and see how longer duration storage can be effectively managed in the trading markets. Colocation didn't grow as quickly as we were expecting it to over the last five years, but that is changing very quickly. I think there was some challenges with early CO location because of complexity around metering and existing solar sites which were supported by the RO scheme wanting to ensure that by co locating a battery alongside the PV production it didn't jeopardize the RO accreditation and also battery capex was higher so there was the opportunity cost of co locating with solar didn't quite work with the higher battery capex and it was better just to put the batteries standalone to make sure that they're unconstrained from an optimization perspective. But this is changing. We've got, I think more certainty in the metering arrangements with sites that are supported by the CFD scheme and that does have a clear path for how a grid connection can be shared between solar supported by the CFD scheme and a battery. So that combined with improvements in battery capex, scarcity of grid connections means I think we're about to see a lot more co located projects come forward. And that's, you know, we have signed more lately than we have for quite a few years before that. So. Okay, yeah. And in terms of the optimization problem, yes, the shared grid connection does impact how we optimize the battery because we have to understand what the solar is going to be doing and therefore adjust the the battery schedule based on that. But it's very site specific. So sites with much larger solar relative to the grid connection, potentially oversized. That's where the battery is going to have to change its dispatch schedule more than it would if the solar were smaller relative to the grid connection. But yeah, the way that we optimize these batteries is we have a forecast of the solar production, we have uncertainty around what the actual solar production will be. So we feed some headroom into that potential solar profile to the battery and then optimize the battery based on these constraints that we're seeing arise from the solar in other markets where we have different forms of colocation. So for example, in Germany some co located projects where the battery cannot import from the grid and there we have to use the solar production exclusively to charge the battery. And the optimization problem becomes different because you lose that ability to import from the grid. So you have to keep that energy in the battery in different ways than you would if you could fully discharge it.
- [41:50]
A
Yeah, it's a fascinating space and I really like how you framed it in terms of, go back 10 years. Batteries are quite expensive relative to say the grid connection and so you wouldn't want to jeopardize your battery by putting it with something else and getting it sort of blocked off at certain times of the day. But now as batteries get far cheaper, grid connections become the scarce constraint. Well, yeah, like how quickly can you get batteries onto each of those sides? I think if you're offered the ability to do it through connections reform, then yes, you want to go ahead and look at It.
- [42:20]
B
Yeah, absolutely. And yeah, from a financing and structuring perspective, we also think it's interesting because if you have a CFD supported solar project then that maybe that's enough contracted income to also allow you to finance the battery without a floor. But equally we are able to and do put floors on co located batteries as well, provided that we understand the the solar profile and can model what the constraints are likely to be. So I think that aspect is also interesting. It's a nice blended risk profile between subsidy supported solar and between the battery and for us taking on both because we have a large PPA route to market business, there are some synergies between the battery optimization and the offtake and balancing risk that we would get. So we can also potentially be more competitive when we're doing both legs of the site as opposed to only one.
- [43:26]
A
Okay. Yeah. So sometimes when someone comes to you with a solar and battery site because you can compete on both, it's a nice place to be.
- [43:32]
B
Yeah, absolutely.
- [43:33]
A
So I want to ask you two more questions. One is around size of battery. So you mentioned earlier that your battery fleet goes up to 300 megawatts. Does that battery get managed differently to a 50 megawatt or a 10 megawatt system?
- [43:48]
B
Yeah, the way that we manage the batteries is generally similar but we do have to consider market impact, market depth and liquidity and that becomes more of an issue for larger batteries. So for example, for a given price level in a certain ancillary service, maybe it's optimal for two hour battery to commit 30% of its capacity. The challenge becomes that if you have a 100 megawatt unit and you're trying to put 30 megawatts into that fairly shallow ancillary service market, that price is going to move quite a bit when you commit those 30 megawatts to it. So something that we have put a lot of effort into is understanding and managing market depth, liquidity, price impact as we start to put these progressively larger units in, into different markets. So while, yeah, the core optimization is the same, but there are more considerations around market impact and depth.
- [44:56]
A
Yeah, how you approach. So do you put in all the volume all in one go? Do you phase it across a few stages?
- [45:01]
B
Exactly how we phase it, how we feather it, how we spread it across, across markets, how we target deeper markets. Generally the larger batteries in the portfolio are also slightly longer duration, so they're doing more wholesale trading as opposed to smaller units which tend to be perhaps a little bit older and shorter duration and there may be a bit more Skewed towards frequency response markets.
- [45:22]
A
Okay, now a fun question. So for someone coming into this space who wants to be an asset optimizer, wants to be a trader outside of listening to this conversation, which if they're here, they have done, what would you recommend to them as something that they should be doing to try and get
- [45:37]
B
a job getting into the space? So let's say for a university graduate or someone go through their studies trying to secure an internship with a party like us, someone who's in a trading house is really a great way to start getting into the space. I think an ideal profile is someone who is very curious because it's a really complicated market. When you're, especially with the market design in GB and still things like embedded benefits, they still exist. So understanding, being able to ask all the right questions in order to fully understand the market design, that, that's important, being analytical is important because at the end of it a lot of what we do comes down to, you know, math, statistics, probabilities and also being tenacious and resilient because you know, a lot of times things go your ways but they don't always. So you have to pick yourself up and yeah, go again. So yeah, getting a foot in the door with a trading house, with an asset optimizer perhaps with a large asset owner, asking lots of questions, trying to understand as much as possible about the industry, that's, that's a good way to start.
- [46:53]
A
And is there still like a progression element to this? So people would want to be say a power trader but nobody comes in as their first job as a power trader. So sort of they come in and they do like an asset, a physical operation style job. So making sure that sort of auctions are met on certain times or maybe they're working in kind of mid office so they're seeing the risk or assessment part perhaps is, is that, is that still true that there's a sort of way from those teams into the more sort of trading element?
- [47:20]
B
Yeah, I think one of the progressive paths which we see a lot with colleagues internally but by no means the only one is to perhaps start as a shift trader. So working in energy management, managing the large renewable portfolio that we have in GB or in Germany and, or the Flex portfolio. So starting out very close to the market with the 247 team under the wing of experienced traders to really understand the market. From there moving maybe into more of a analytics platform role. So working on the automated trading systems, once you understand the market then you can work on systems to help automate Some of that decision making. So that that frees up headspace for the traders to work on the more challenging problems. And then from there potentially coming forward to the origination team and working with clients, explaining what we do, how we do it. And then you have that end to end coverage from having to learn a bit, you know, learn about the project finance side of things, but also being able to relate back how the physical operations work and basically how the value is created for that battery around the structuring as well. But that's one way, that's a common way. But we do have colleagues who have come in maybe from a risk management function or from a analytics or consulting background as well. So there's several ways, there's lots of
- [48:57]
A
routes and it's great to hear because it's probably one of the most asked questions, I think for people outside of the space. How do I get into it? How do I do it? Okay, let's wrap up a final question. What is a contrarian view you hold about energy markets?
- [49:10]
B
I think a contrarian view that I hold about, let's say the battery market in particular is, I think there's probably too much talk about megawatts, megawatts of bez and not enough talk about megawatt hours. I think also maybe in some of the forward looking projections around how much BEZ is required for the system, I think there's maybe an overemphasis on the megawatts. And you know, I wonder if a more efficient solution could be to make projects longer as opposed to building lots of nameplate capacity of two hour projects. So we are seeing the evolution in the market where our projects have gone from one hours to two hours, they're progressing towards three and four hours. I think we'll see more of that, More of that. I think there's a big role for augmentation to play. So taking existing projects, making them longer duration and yeah, I think it's that megawatt hour piece that we need to start focusing on more than just the megawatts.
- [50:18]
A
If you read Clean Power 2030, it says 22 to 27 gigawatts. Right. Of batteries. It doesn't say gigawatt hours in there. Right. So as you say, the problem is being described in the sense of power, which is very useful in terms of dealing with peaks on the system. And that does make sense. If you're dealing with ramp rates or peaks, then power is a great thing to use. But also there's another problem which batteries help to solve, which is when we get to a period in winter where the wind's not blowing, sun's not shining, we need gigawatt hours or we need terawatt hours. We need to be better equipped to describe what the problem is that we'll face. And so I think your call for more transparency on that feels much needed.
- [50:57]
B
I think that would be something I'd like to see. And I do wonder if certain assumptions were made in that Clean Power 2030 analysis around, for example, are all the batteries assumed to be two hours? If that's the case, then maybe if some of those batteries were simulated as being four hours instead, it would be a different answer as the optimal and
- [51:20]
A
some of our one hour systems today are morphing, augmenting into three four hour systems. Right. So we already see this happening.
- [51:26]
B
Yeah, absolutely. And it's something that we do see a lot more from prospective counterparties and existing counterparties, this appreciation that projects will need to augment and there needs to be technical design and land availability to allow them to augment as well. So we think that is really important, this ability to add storage as the market needs it.
- [51:54]
A
Agree. Well, Brian, thank you very much for coming on. You've been a fantastic guest and I think everyone has learned a huge amount about how optimization actually happens.
- [52:02]
B
Yeah, thank you very much for having me.