02 - Building technology to control energy storage systems with Charlotte Johnson [Chief of Staff @ Krakenflex]

[MUSIC PLAYING]

QUENTIN SCRIMSHIRE: Hi, guys. Quentin here from Modo. We're here with Charlotte Johnson from Krakenflex, who we've been wanting to get on for a while. And today, we're going to talk about technology in the energy storage world and how it's really changing how assets are operated.

So Charlotte, thank you very much for coming on and speaking to us. Before we get started, I guess it would be good to have, for the guys at home watching or wherever they're watching this, who is Charlotte Johnson? And how did we end up here?

CHARLOTTE JOHNSON: Well, thanks firstly, for inviting me. But I'm Charlotte Johnson, a senior analyst at Krakenflex. I've been with the company for nearly four years now, or just under 4 years, and there's two main areas that I focus on. One is the commercial side, so basically market modeling and market insights, and the other is strategy.

So looking at the business strategy for Krakenflex in alignment with the wider Octopus Energy Group.

QUENTIN SCRIMSHIRE: Great. And so on your background before this, how did you end up in the world of energy? Did you always plan for a life in energy?

I think that's slightly tongue in cheek. I don't think anyone.

CHARLOTTE JOHNSON: So at university, I studied geography, physical geography, and then I did a Master's in Climate Modeling, really looking at the melt rate of the Greenland ice sheet. And I think from that, obviously carbon emissions is a massive thing and a huge driver, and I think I looked at the energy industry. They're responsible for a huge proportion of carbon emissions, so it felt like kind of the next step to do something practical in this space.

QUENTIN SCRIMSHIRE: Awesome, OK. And so Krakenflex.

Let's get this out of the way. So Krakenflex and Octopus Group, whose Krakenflex? And how do you fit in with Octopus? And what do you do?

CHARLOTTE JOHNSON: Yeah, so Krakenflex was originally Upside Energy.

QUENTIN SCRIMSHIRE: Upside, sorry, yes.

CHARLOTTE JOHNSON: So originally Upside for probably the past five, six years--

I think since 2014, 2015.

We were acquired late last year by Octopus Energy, or the Octopus Energy Group. And within the Octopus Energy Group, you've got multiple companies. So there's Krakentech and Krakenflex that are kind of related together, and they sit on one side. And then you've got Octopus Energy Services, Octopus Electric Vehicles, Octopus Renewables, and then there's a number of other areas of the business too.

QUENTIN SCRIMSHIRE: OK. And so you're--

Krakenflex, so you're based in Manchester, right?

How many people are there in the company?

What's the team? Is it software developers or is it commercial people? How is it split? And then who's the customer?

CHARLOTTE JOHNSON: Yeah. So we are based in Manchester, the northern quarter, although we do have an office in London as well.

The split is mainly [INAUDIBLE]

the largest proportion of the company is software developers. So we've got a huge number of tech teams, and then we've got a small commercial side.

QUENTIN SCRIMSHIRE: OK. So Krakenflex is the company, right?

QUENTIN SCRIMSHIRE: And then what's the product called? And who is the customer?

CHARLOTTE JOHNSON: Yeah. So the product has--

it's a modular platform. So we've got four separate modules, and they're basically control and dispatch, and that's where we connect to assets and dispatch them. You've got optimization, where we basically pull in market price signals and optimize the assets, and decide when you should dispatch them. You then have alerts and notifications, and then you've got reporting and analytics, which is all around sort of operational performance and financial performance of the assets.

QUENTIN SCRIMSHIRE: OK. So if I've got this right, you guys do--

like the technology are the controls--

and do you do market access as well? So do you do the optimization bit, like habitat word or an EDF? Or do you do something different?

CHARLOTTE JOHNSON: Yeah. So it would vary between customers. In terms of ancillary services, if we start there, we basically do access to ancillary services for customers that potentially don't have a megawatt to enter the market.

QUENTIN SCRIMSHIRE: In case it's too small.

CHARLOTTE JOHNSON: Yeah, because they're too small. So we would aggregate them, and we would basically bid them into the market for them. So we do that for a number of customers at the moment, usually retail customers. They've got smaller portfolios.

QUENTIN SCRIMSHIRE: So just like [INAUDIBLE]

with his electric vehicle, or is it, what--

CHARLOTTE JOHNSON: Slightly larger than that, to be honest. I don't think there's many electric vehicles in the ancillary service market at the moment, but it's usually CNI or small businesses. So for example, some have got 200 kilowatt batteries, 100 kilowatt batteries, that kind of range.

QUENTIN SCRIMSHIRE: OK, cool. And so in that case--

So just going to take this back. So you provide the technology and some tools for folks who are doing optimization, right? So is that screens for trading teams, or is that algorithms? Is it price forecasting?

CHARLOTTE JOHNSON: Yeah.

QUENTIN SCRIMSHIRE: How does that all fit together?

CHARLOTTE JOHNSON: The idea is is that customers upload their own price forecast, or they perhaps their own sort of market forecast. They also will look at--

if they have got a VPP, or a virtual power plant with renewable generation in as well, they'll pull in their own generation forecast. Our optimizer would look at the market prices, look at the assets that are in the portfolio, and decide when to basically dispatch them.

So you can do it that way, or you can send us a dispatch schedule via an API. So there's two different ways. It all depends on the customer and what their capabilities are like, really.

QUENTIN SCRIMSHIRE: OK. So who are your biggest customers right now? Are you able to say who they are, and what you do for them?

CHARLOTTE JOHNSON: Yeah. So we have four key customer groups, and the first one starting with would be asset owners. So they're sort of large fund owners that have invested in grid-scale battery projects or renewable projects. So you've got kind of the likes of Eelpower--

would be one of them. You then have off-takers. So these would be companies like EDF Energy, Habitat Energy, Shell, and their people that would use us to connect to a large number of distributed assets.

You then have retailers, so that would be the likes of Bright Energy, Haven Power, and Octopus Energy, who have a growing portfolio of either domestic flexibility or CNI flexibility. And then the final group is system operators, and that would be the likes of transmission system operators or distribution system operators, and that's the kind of work that we're doing with Elia Group.

QUENTIN SCRIMSHIRE: They're saying, just sign a deal? Or something was just agreed and [INAUDIBLE]

were taken? Congrats to you and the team for doing that. Big moves abroad, right?

CHARLOTTE JOHNSON: Yes, one of our first international deals, actually. So Elia are the transmission system operator for Belgium and also 50 Hertz in Germany. So Germany's got four, or a few different transmission system operators.

QUENTIN SCRIMSHIRE: OK. So in Europe, they have so--

National Grid ESO is our transmission system operator. One of the--

yeah, basically our transmission system operator. And then in Europe, they don't have one per country, they spread across countries. So what is exactly is Krakenflex going to be doing for Elia over there?

CHARLOTTE JOHNSON: Yeah. So we're looking at building out two modules for their platform. So they're building out a new digital platform, and the idea is that it will all be sort of plug-and-play, that they can then license out to other TSOs. And the modules we're focusing on is LFC, which is load frequency control, and shadow settlement.

QUENTIN SCRIMSHIRE: OK. Octopus, of course, acquired Upside, which became Krakenflex. And now you have--

which is incredible, you've got now hopefully some more money to spend to grow the team, deliver on the vision, deliver on the technology. So my question is, what does that look like? Now that you have the backing to deliver on what the vision of the company was, what does that look like in 2030? And where does Krakenflex fit into that?

What's the technology going to do to your customers that's going to make their world better?

CHARLOTTE JOHNSON: Yeah. And I think if you look across the four customer segment groups, you can see that we're trying to branch out and diversify from just one customer type, and the idea of that is to really help us be at the heart of the technology that then enables the energy transition. So to do that internationalization is a huge one for us, so Krakenflex and Octopus have a very strong presence in the GB market, and now it's really looking at how we can enter new geographies strategically to sort of access that vision.

QUENTIN SCRIMSHIRE: OK. And then of course Europe is first, and then Octopus is in the states now, isn't it?

CHARLOTTE JOHNSON: Yes. Yeah, they're in the states. So Texas is a state they're in. They're in Japan with Tokyo Gas, Origin in Australia, and then, yeah, Europe.

QUENTIN SCRIMSHIRE: Brilliant. OK. So then last question about Krakenflex. So what's the business model? How do you guys make money?

Do you charge per license? Do you charge per seat? Is it a box that you sell? What does the business model look like?

CHARLOTTE JOHNSON: Yeah, so there's two different sides to it.

We're a software as a service company, so we have a subscription fee, and customers basically pay a pound per kilowatt per year fee for the volume that they put on the platform.

And so you'll see that that varies between customers because it depends on how many modules you take and also how big your asset is, or how big your portfolio is. On the other side--

QUENTIN SCRIMSHIRE: So question on that because, of course, having 100-kilowatt assets--

the math is going to be in my head now, but we put a zero on where it should be. A lot of small assets is not the same as one big asset, right? Because the amount of effort to get those small assets into the market and all the metering, and registration, and stuff, it doesn't really scale like one. So I guess it can't cost the same for that as a big unit, and that's one of the big problems that our industry needs to solve, right?

CHARLOTTE JOHNSON: Yeah. So at a smaller level, it tends to be a pound per asset fee instead. So yeah, that's the slight difference between the front of the meter and behind the meter proposition.

QUENTIN SCRIMSHIRE: OK.

All right. So now about the wider industry. So you've been around for a while. You've been in lots of writing, and we've been following Charlotte, and I'm sure a lot of folks watching have been following the stuff you've been writing and putting on LinkedIn, and I just want to say it's absolutely terrific. So thank you on behalf of us for putting the time in and releasing that work. So in your view of the industry as a technology provider, where does Krakenflex sit?

From people who build assets and own them, all the way up to the end user, how does Krakenflex sit within that? And what are the changes that Krakenflex is making to the industry?

CHARLOTTE JOHNSON: Yeah. So if you're looking at it from a customer point of view, it would be sitting between the asset owner and the optimizer, or the off-taker. If you're looking at it from where we physically sit in the sort of integration stack, you would have the site controller, which is someone that does all the lower-level functionality on the site. So with things like HVAC systems, fire suppression and detection, all the kind of monitoring, controlling of switchgear, and they're usually done by--

In our industry, they're called SCADA systems, but you often hear the term PPC, so a power plant controller, an EMS, an energy management system, or a BMS, a battery management system. So they're the collective terms for SCADA technology. We interface with that technology and we do so by putting a piece of hardware on site which we call a gateway, and that gateway then allows us to communicate with the asset and also back to our cloud. And that's obviously what our customers then use to dispatch the asset through the cloud.

So it's that kind of bridge between the physical asset and the cloud.

QUENTIN SCRIMSHIRE: OK. Do you provide the metering as well? So the comms for registering, and metering, and all the stuff for being a BMU, and--

do you do that?

CHARLOTTE JOHNSON: So we don't own any of the meters on site.

So we don't take any kind of liability there. That's all done by the EPC that usually organizes or configures the site.

But we would obviously advise on, when you're looking at going into different markets with your asset, you really need to think about what metering you need, so the accuracy and also the resolution of it. So we can advise on that, but we wouldn't go out and procure it, and install it on the site.

QUENTIN SCRIMSHIRE: OK. Well, let's talk about that for a second because you guys--

I don't want to put you on the spot and say, "How many assets do you control now?" But quite a few. They're in a lot of different services, and so you have to pre-qualify for a lot of different services, and they're using your technology to do that. So if it's OK, can you just talk a little bit about pre-qualifying for ancillary services?

So for example, in the last year, we've had dynamic containment. One of the interesting things about dynamic containment, which is the new frequency response service, is it's new, it's fast, and the regime for testing to pre-qualify to get into that service, it's quite a bit more complicated than it used to be for FFR, for our older services. And so the technology requirements to pass those tests are pretty stringent. So what's your journey been like at Krakenflex?

I keep on trying to call you Upside, and it's very--

But Krakenflex, what's your journey been like and what have you learned? What can you share with folks who are listening on how they can make sure they don't fall down any traps?

CHARLOTTE JOHNSON: Yeah. There's been some really great developments with dynamic containment because if you think about--

it's the first ancillary service that grid has brought out that actually allows you--

or gives you specific rules on how you manage your state of charge, your state of energy, which is really important for energy-limited assets.

So that's great. When you think of, though, the metering, it requires 1 Hertz operational metering and 20 Hertz performance metering.

QUENTIN SCRIMSHIRE: So one type of--

operational--

what does that mean, operational metering?

CHARLOTTE JOHNSON: So operational metering is the metering that goes in real-time to National Grid's control room, and it's secondly, so 1 Hertz. And then the performance metering is 20 Hertz, which would be delivered an hour after the event to National Grid.

QUENTIN SCRIMSHIRE: OK. So National Grid, once a second can see generally what you're doing. And then if there's a big frequency event, like low frequency, something falls off the system, wind farm trips, something like that, then they can look back through the data at 20 Hertz, or 20 times a second, and see did you do what you're supposed to do?

CHARLOTTE JOHNSON: Yeah. So the 20 Hertz is all there to basically validate your response and to prove that you did it. In the past with DFR, the way grid validated it was through audits, or spot checks, and they would ask for it secondly readings for a 4L block.

QUENTIN SCRIMSHIRE: In Excel sheets. I know they used to be like, "Send us an Excel sheet from last month." Like ah, really?

Is the world [INAUDIBLE]

on Excel sheets? OK, so we don't use Excel sheets anymore?

CHARLOTTE JOHNSON: No.

That's good. That's great. And so passing the test--

so hypothetically--

because we are just talking about energy storage on this podcast, although we do tend to talk about all sorts of stuff--

hypothetically, we've got the Modo awesome energy storage limited, and we have got 20 fantastic containers on a great site that are ready to go and get into [INAUDIBLE]

containment, and we have used Krakenflex for our technology provider.

And so we want to get these 20 megawatts site, or whatever it is, into the market. What's the process of us pre-qualifying? What do you have to do at Krakenflex?

CHARLOTTE JOHNSON: Yeah. So we need to look at the metering requirements to make sure that you've got the right meters on site, and the site is configured in the right way.

Then you would go speak to an independent technical expert. So there's someone that essentially--

or an IT that was set up by National Grid a few years ago, to try and speed up that process from when you've done your DeFi testing or your DC testing.

Instead of sending it to National Grid, you can send it to an ITE who would sign off it, and then you'd be able to send it to National Grid. So it's supposed to make it much quicker.

QUENTIN SCRIMSHIRE: We'll talk about the test in a second, but you do the tests and then you get an expert in who is some sort of engineering bot, someone from one of the engineering consultancies I'd imagine, come in, look through your data, and then they sign it off, and then National Grid says an ITE, independent technical expert is happy with this. So therefore, National Grid is happy with it. Is that how it works?

With the ITE, though, there's something to sort of bear in mind that the earlier you bring them in, or the earlier you engage them, often is better because all ITEs are very different in how they work, there's not a consistent approach. Some of them will look at the test results on an Excel spreadsheet and just sign it off and be happy. Others will actually want to go on a site visit and look at how the site is configured, and look at the metering, and start to probe more into about how you've generated the results.

So the earlier you can engage them, the better you can de-risk the projects.

QUENTIN SCRIMSHIRE: Stop me if I'm wrong here--

so you do the test, you get out the results, which is data in the Excel sheet, or something else, right? And then the ITE's job is to say, did that battery do what the data says it did? And are we measuring it in the right way?

CHARLOTTE JOHNSON: Exactly, yeah.

QUENTIN SCRIMSHIRE: OK. And then what about the actual tests? So what do I need to prove to National Grid to make sure I can get into DC?

CHARLOTTE JOHNSON: Really that you have a stable internet connection in the sense of when you're looking at the operational metering that's the secondly data that goes to National Grid, you need to be able to--

not all sites will be configured in a way that they can deliver stable readings to National Grid every second. So there's a bit of leniency with some dropouts, but there's a threshold on how much you can kind of have. So you need to be able to prove that you've got a stable connection.

QUENTIN SCRIMSHIRE: [INAUDIBLE]

of mine having been involved in quite a few projects over the years, for anyone who's developing projects or building projects, make sure you get your openreach connection pretty early in the build out.

You don't want to come to switch it on and then think, "Oh, OK. Where's this BT box I need to connect to?" Because it can take ages for such a small thing. And so these are mostly wired connections for Kraken, is it?

Or do you do 4G cards? How does it work?

CHARLOTTE JOHNSON: So there's a bit of both. So our preference is usually to have a direct line.

In reality, it's not possible on all sides. So a 4G connection is often used as well, and what we tend to do on those sites is to add an additional 4G SIM with a different network provider so that you've got some sort of redundancy if one of them goes down.

QUENTIN SCRIMSHIRE: What about different assets? So we need to pass some dynamic containment tests, we need to do a step test, we need to do frequency following test. We need to do all that good stuff, but all batteries are a little bit different, and so you may have an asset that can respond within, whatever it is, 500 milliseconds. It was pretty fast.

Slow for Ireland, but we'll talk about another time. So you've got a fast asset, and then you connect Krakenflex's technology to it, and then how do assets differ in their response?

Do you have some that are faster than others? Some that don't quite track frequency properly?

What should asset owners look for when they're building these kind of sites, or looking to connect to your technology? Is there a latency question to worry about, anything like that?

CHARLOTTE JOHNSON: I don't think we've ever had a site that we've tested for DC and it's not been able to deliver, aside from your really crappy small batteries.

QUENTIN SCRIMSHIRE: Oh, OK.

CHARLOTTE JOHNSON: So I don't know if it's--

what we find is that the smaller the site, the worse the inverter tends to be, and it's the inverter that's the one that you need to deliver the frequency response. So we've got about 300, 400 megawatts on the platform. 95% would be front to the meter.

All of them can do DC, and we've never had to retest. You test them once, they do it, and you go. So I don't know if that's completely different to what other people have found.

QUENTIN SCRIMSHIRE: I think that some folks find it really difficult to pre-qualify assets in certainly the first half of this year because we saw in our data, assets being out of other markets, like FFR. And you could see they were testing, you could see BMU registered asset--

you can see they're testing, you can see what they're up to. There was certainly a delay. And anecdotally, we know that some assets have struggled to be fully qualified.

So you might have a 40-megawatt asset, and it might only mainly pre-qualify for 36, but maybe you've managed to scoot around all of those difficult assets.

CHARLOTTE JOHNSON: No, I don't necessarily--

so I know, for example, there's a 50 megasite that we've got 50 mega, and I think it's a 1 and 1/2 duration, so 75-megawatt hours.

That took a long time to get into the market because it didn't have the fixed-line connection that it needed. So using the 4G sims were not good enough, it didn't work, so we had to wait, basically, for a fixed line to be put in. That was the blocker, though, not necessarily the inverter, or not being able to actually deliver the response.

QUENTIN SCRIMSHIRE: So the asset could do what it needed to do, you just couldn't report on it in the right way for the National Grid.

CHARLOTTE JOHNSON: Yeah. because if you think at the very beginning when Grid brought out dynamic containment, I think it was in the October 2020, you didn't have to do the real-time metering. There was that derogation, or that leeway period, where assets could participate in DC, providing they pass the tests, and didn't necessarily need to actually give the real-time metering. So for that asset, it didn't make a difference.

But then I think it was January the 1st that the rule kicked in that you needed to be real-time metering. And yeah, that kind of brought the asset out of the market.

QUENTIN SCRIMSHIRE: Great, OK. So coming back to assets for a second, what are the differences between the assets that you connect to? So a battery does not equal the same--

all these battery assets are different. We should do a calendar of these batteries one day, but they're all slightly different, they all respond very differently. You've got containerised systems out there. You've got in 40-foot containers, 20-foot containers, separate inverters, integrated systems, loads of different ways that these assets can be configured.

And so if I was building an asset today and I knew that I wanted to use Krakenflex's services in the future, what should I think about when I'm specifying this asset to make sure it's going to be a good asset for you to connect to? So we need a fixed-line, a broadband connection.

CHARLOTTE JOHNSON: Or a strong 4G connection, yeah.

QUENTIN SCRIMSHIRE: What else do we need?

CHARLOTTE JOHNSON: I think you've already kind of alluded to it, thinking about the markets you want the asset to participate in. So if it's dynamic containment or if it's the balancing mechanism, can you do that 1 Hertz operational metering? And so for that, you'll need the right meter. Likewise, with dynamic containment, you'll need a 20 Hertz performance meter.

So have you got a frequency meter that can monitor frequency at 20 Hertz?

Then also an active power meter that can actually report on your readings at 20 Hertz, so 20 readings a second.

QUENTIN SCRIMSHIRE: It's a lot of data to be constantly sending out.

CHARLOTTE JOHNSON: Yeah, a huge volume of data compared to previous markets.

And then the final thing is then looking at, what else is on the site? So do you need any submetering? So a lot of battery sites later down the line, want to put electric vehicle charging or other assets behind the meter. So thinking about, do you have the right submetering? And how is the site configured so that you can still deliver your DC response whilst also having other assets on the site that don't invalidate that response?

QUENTIN SCRIMSHIRE: Krakenflex connects to all sorts of different types of assets, right? So you've got big, sexy, big batteries that we love.

I hear that the smaller ones are also very interesting. So can you talk to us a little bit about behind the meter small assets? How are they different from the big ones, and what's different in the technology realm?

CHARLOTTE JOHNSON: Yeah. So in principle, when we connect them, it's the same in the sense that they all have the same level of safety, security, and reliability built into the site. The difference, though, is that on a large site--

so take a 70 megawatt battery farm, or 110 megawatts of wind, you would have more redundancy built in, or resiliency built in. So we would do additional frequency meters, we would do additional industrial PCs just so that if anything goes down, you've got a backup.

When it comes to domestic sites, or smaller sites--

so a few hundred kilowatts, you might have a cloud-to-cloud integration with the manufacturer, and that just makes it more cost-effective for the customer, and also less disruptive because you don't need to--

QUENTIN SCRIMSHIRE: So cloud-to-cloud, does that mean that Krakenflex is talking to--

I don't want to use Tesla again, let's use a different one. Say you've got a BYD battery in your house, and so Krakenflex would have an API to BYD's cloud server, which would control the battery. So you wouldn't be speaking directly to the battery, you'd be going through the OEMs cloud service?

CHARLOTTE JOHNSON: Yeah, which is great for cutting the cost down and making it more scalable. The only drawback that you might have is latency between the two car systems, which means you might not be able to do a service like dynamic containment.

QUENTIN SCRIMSHIRE: OK.

And so these smaller--

so there's more redundancy in bigger sites. And sometimes, you won't put your own hardware on the site, I assume, so you just control it all via an API. How about the performance? Are smaller assets--

do they perform worse than bigger assets because they're--

I'm going to say something tongue in cheek here, but cheaper? Of course there's a scaling cost, but are they more difficult to get into ancillary services? Are they more of a headache, or are they easier? Quite a loaded question.

CHARLOTTE JOHNSON: Yeah, it is a loaded question. There's a few things. So if you're a small site, you'd need to be aggregated to get into an ancillary service market, which then means you need to be tested individually and also at a portfolio level. And every time you then add in a new asset, or drop it into that portfolio, you have to retest the whole portfolio again.

So making sure that you've got a time window blocked out to perform those tests with--

If you think about electric vehicles, you're talking about hundreds of electric vehicles that you would need to get to a megawatt to then be able to test. So orchestrating that test is difficult. You also have the challenge that some of the inverters are not necessarily the same, they haven't been set up to do dynamic containment. Whereas if you're buying a large 50 megawatt, 70 megawatt site, the containers you're buying have been made and spec'd out to be able to enter these markets. So there's different sides to it, yeah.

QUENTIN SCRIMSHIRE: Makes sense. And now we're going to get to the bit that we really want to talk about. So everybody knows Charlotte Johnson from some of the stuff you've been writing recently. We are massive fans of the work that you've put out on LinkedIn and on your blogs.

Not only the work itself, and the analysis, which is awesome, but the titles of some of these pieces. And one of them that we particularly liked recently was, "An Imperfect Storm" within brackets about this winter.

And some of the stuff that you talked about in that piece has really come to the front of everybody's attention since it's been written. So you talked about this winter being tough for our industry in lots of different ways. And I wonder whether you could just talk through what that research involved, and what your view is of the market now? And let's just chat it through.

CHARLOTTE JOHNSON: Yeah. So I originally wrote that, I think it was end of September, beginning of October time, when there was so much going on in the media about the crazy volatile market prices that we were experiencing. And I really wanted to sort of disaggregate, or separate, the difference between the extreme pricing that we saw in the first few weeks of September versus the consistently sustained high prices of gas and electricity due to rising commodity costs that we'd seen the whole year.

And I think that was kind of mixed. I don't think the message was really delivered clearly, and particularly not in the media with all the noise that there was about what was going on.

QUENTIN SCRIMSHIRE: So what was the purpose of the piece?

CHARLOTTE JOHNSON: Yeah. So the first few weeks of September, the market prices on the Day-Ahead market and the Balancing Mechanism were significantly higher than what they'd been the rest of the year. So I think the Day-Ahead price reached a new record of over 2,500 pounds per megawatt hour, and the Balancing Mechanism bids and offers were accepted over 4000 pounds per megawatt hour.

QUENTIN SCRIMSHIRE: Which is insane, right? It feels normal to us now in December recording this, but to everybody in the industry, this is crazy, right?

CHARLOTTE JOHNSON: Yeah. and particularly when you think about an average Day-Ahead price, which could be anywhere between 40 to 50 pounds per megawatt hour, 4,000 pounds per megawatt hour in the BM, or 2,000, 3,000 pounds per megawatt hour in the Day-Ahead market is a massive difference.

QUENTIN SCRIMSHIRE: Especially Day-Ahead, right? Because Day-Ahead is, of course, Day-Ahead, no flies on me there, but there's so much longer after 24 hours ahead to--

usually 24 hours ahead, you can take so many actions to bring the cost of that electricity down. There's so many things you can do. But because markets are so tight, we were seeing just insane prices. And what does this mean for everybody?

CHARLOTTE JOHNSON: Yeah. Well, I think September was unique, though, in the sense that we weren't into the winter months yet, and we were still sort of coming out of summer. So you had so many plants offline, like the nuclear outages for summer maintenance, and the same with thermal generators as well. So you had a huge volume of capacity that was missing off the system.

And our market, or our grid, is so heavily reliant on gas.

So when no one else is around, it's the gas generators that will have to set the price and will have to kind of step in, and I think that's why we saw the kind of extreme pricing. And it didn't help that wind was obviously unseasonably low during that three-week period, and very low. I think it was below two gigawatts, and there's about over 20 gigawatts of wind installed. So 10% of what is installed was obviously generating.

QUENTIN SCRIMSHIRE: So we've got unavailability--

it is the perfect storm, right? So we've got the unavailability of plant, we've got crazy high gas prices that--

we don't yet, even in December of recording this, we can't see an end to that.

In many cases in the electricity market, power prices are set by the most expensive generator in many cases. And when it's not very windy, the power prices are set by gas generators anyway--

but especially when it's not very windy, except by gas, and in some cases, coal, which we saw some really high prices there.

So what does this mean for flexibility, and batteries or gas peakers, or--

what does this mean for our energy system?

CHARLOTTE JOHNSON: Yeah. And I guess that's where the title came from, this sort of imperfect storm because it was perfect for flexibility providers, or generators, that had dispatchable technologies that could take advantage of the low Day-Ahead prices for batteries, and the extreme high prices to discharge that. On the other side, you've got suppliers or customers that don't have a flexible load book, or can't shift their load.

QUENTIN SCRIMSHIRE: Does Krakenflex control DSR? Because I've been thinking this is perfect for DSR, and even some of these backup diesel generators that have got capacity market contracts and whatnot. Surely they should be running now and making an absolute killing. I don't know whether that's actually happening in practice, but does Krakenflex do DSR too?

CHARLOTTE JOHNSON: Yeah. so we do have a growing portfolio of DSR customers that are all on the CNI side. So that kind of encompasses industrial heat pumps, factory lines that you can turn down, refrigeration, HVAC systems, and then some back-up generation.

QUENTIN SCRIMSHIRE: I guess if you're a plant that produces stuff in the UK, and that requires a lot of energy, you are basically doing DSR whether you like it or not now because every day, you're looking at your electricity bill and you're going, "Holy smokes. What can we turn off to save electricity?" Which is basically DSR without an aggregator. Anyway, we digress.

So the imperfect storm this winter, craziness across the board, what does this mean for flexible assets? And what's the opportunity out there?

CHARLOTTE JOHNSON: It's a huge opportunity for providers that have got dispatchable technologies or flexible assets, like you said, because if you have a battery, you can obviously discharge into those peak prices, whether it's Day-Ahead or you can opt to enter into the BM, and the same with gas generation as well. A lot of gas generators were holding back capacity on the Day-Ahead market in the hope that they would be successful in the Balancing Mechanism.

QUENTIN SCRIMSHIRE: Yeah.

After so many years, so many peakers got built. They were developed and got built in 2017, '18, '19, '20.

And then everyone said, oh, we don't want peakers anymore, we only want batteries. Of course, we're battery enthusiasts, so that's great.

But now if you own peakers, despite peaker evaluations this year went really low, and nobody wanted to touch peakers because they were dirty gas--

and then of course peakers now running are making an absolute killing. So just very, very interesting.

We do need some sort of peaking generation out there, and there is a big opportunity for it. And so coming back to imperfect storm--

I'm wandering around this conversation, just knocking on doors, and just going crazy. So imperfect storm, how did you finish? Because if I remember correctly, there was scenario analysis.

Was that your own forecast? Or was that the Krakenflex forecast? How did the scenario analysis work, and what was the outcome?

CHARLOTTE JOHNSON: Yeah. So I think it was a combination of some of the historic things we saw in the market. So we looked at how much generators were making.

I think for September, it was double what they'd made in previous months. And likewise, for batteries, what a battery could have made or what their opportunity would have been during those weeks. But really, it was looking forward to looking at National Grid's winter outlook, which looked at the D-rated margins, and trying to see what's the chance that we're going to see those extreme prices that we saw in September again in the winter months.

And so I don't know if you saw National Grid's winter outlook, but they basically said that the D-rated margin was going to be between three to four gigawatts, and the D-rated margin is just basically used to illustrate any excess capacity above the underlying demand.

QUENTIN SCRIMSHIRE: So peak demand, they're saying when everybody--

when [INAUDIBLE]

is on, everyone's using their kettle, no.

Peak demand on a weekday, trial period usually, there's going to be x gigawatts of demand on the system. And we at National Gird, we think we've got three or four gigawatts on top spare capacity. Is that what they're saying?

CHARLOTTE JOHNSON: Yeah.

QUENTIN SCRIMSHIRE: That's quite frightening. And so what's the normal number for a D-rated margin?

CHARLOTTE JOHNSON: Well, it's still higher than what we saw in September. September was below two gigawatts. So it's not too far off what we usually see in winter. I think the lowest that we've had in winter in the past five years has been slightly below three gigawatts, but the fact that it's three to four is above the lowest.

So not the highest, but not the lowest.

QUENTIN SCRIMSHIRE: OK. Still though, my basic mind--

a big nuclear plant is not far off that if, say we had a trip, so quite worrying. Of course there's other things they can do. There's other things--

we had a capacity market notice this week, or last week, which was rather exciting.

Do you think we can have a capacity market event soon? We're all very excited to finally have a capacity market event. Is it going to happen?

CHARLOTTE JOHNSON: Probably likely that it's going to happen. It's just whether or not it actually turns into a capacity market event. It is another thing because the market tends to be quite good at responding, and we saw that last year. I think there was a six electricity margin notices and a few capacity market notices, but nothing really resulted in a capacity market event.

It's quite rare that we get to that point because we've got a market that's very responsive. Although sensitive, it's very responsive to events that are happening.

QUENTIN SCRIMSHIRE: Yeah.

It would be interesting to see how much actually--

in an actual capacity market event, how much capacity you could actually switch on for it.

I don't know. We should do a poll on this. It's not 100%, so what is it?

All right. So Charlotte, anything else you want to bring up while you're on the podcast?

CHARLOTTE JOHNSON: Oh, no. You caught me on the spot. What am I supposed to say to that?

QUENTIN SCRIMSHIRE: Well, I don't know. What are you working on at the moment? Is there a new piece coming out?

CHARLOTTE JOHNSON: Yes, there is a new one.

So I don't know if you saw the one I did last year, sort of a year in review of 2020. I think it was called "The Good, The Bad, and the COVID".

I'll do a repeat this year, not the same title. I need a new title.

QUENTIN SCRIMSHIRE: Like a Spotify Wrapped? But OK, cool.

CHARLOTTE JOHNSON: But yeah, looking at the highs and lows that have happened this year for 2021.

QUENTIN SCRIMSHIRE: And for anyone who is listening or watching, so you can find Charlotte--

how do people follow you and your work? What should they do, follow you on LinkedIn? How do we do that?

CHARLOTTE JOHNSON: Yeah. So I have a Medium account which I put all my blogs on, and then also LinkedIn.

QUENTIN SCRIMSHIRE: OK, awesome, and do that because you will learn a lot. So I just want to say, Charlotte, thanks so much for coming on. It's been awesome, and we really look forward to hopefully having you back in the Modo offices sometime soon. And everybody who is watching, listening, do follow Charlotte.

Can they send you a DM and ask any questions?

CHARLOTTE JOHNSON: Yeah. Of course, yeah.

QUENTIN SCRIMSHIRE: Yeah. And see you next time. Thanks very much.

CHARLOTTE JOHNSON: Thank you.

[MUSIC PLAYING]