Optimization in Australia’s National Electricity Market with Matt Grover (Fluence)

Hi, everyone. Welcome back to the Transmission Podcast. I'm Wendell, and we're back in Australia this week where I'm joined by Matt Grover, director of energy markets at Fluence. Australia's national electricity market is home to some really big batteries, and there's plenty more on the way soon.

However, ultimately, these things don't really do anything without having the right people in place and the right technology to actually tell them to do the right thing at the right time in the energy market. This is exactly what Fluent are doing in Australia, where they provide bidding services for batteries in the NEM. And so that's exactly what we're talking about in this week's episode, battery optimization or auto bidding trading in the NEM. I'm a battery optimization nerd, and I really enjoyed this conversation, and I hope you will too. As always, if you're enjoying the show, please like, subscribe. It really helps us to grow our audience. So let's jump in.

Hi, Matt. Welcome to the Transmission Podcast. Welcome to Sydney.

It's a pleasure to be here. Thanks for having me, and welcome, Moto, to the NEM.

Thanks. And, yeah, like, really exciting to have you on because we're gonna talk all about optimization, auto bidding in the NEM, a topic which is something I find super interesting. But to start off with, I guess, can you just give a brief intro to yourself, to Fluence? What is their role in the NEM?

Sure. Yeah. So I'm Matt Grover. I'm the director of energy markets at Fluence here in Australia. A bit about me. I've worked my most of my career in the NEM, all of my career in wholesale electricity markets, mostly helping technology companies try and do new and innovative things in the wholesale market structures. So I worked the first half of my career on the demand side of the market, helping a global demand response aggregator do new and creative things on the demand side and making loads more flexible.

And I've worked the second half of my career with AMS now Fluence, which was an acquisition Fluence made four or so years ago, Working to make the generation side of the market more flexible. We spent our first few years in the them working with wind and solar farms, helping them curtail to avoid negative prices. So I like to say that I've worked the first half of my career, on the demand side, helping loads curtail to avoid high prices, in the second half of my career working on the generation side of the market helping wind and solar farms and generators curtail to avoid negative prices. So it's all kind of come full circle in the last few years since we started working with grid scale battery projects, which are the ultimate flexible asset in the market and chase opportunities on on both the supply and the demand side.

Very briefly, I guess. Yeah. You mentioned the acquisition of AMS by Fluence. Kind of can you just very briefly just tell me, like, what is the story behind that, and, you know, where does Fluence Digital now fit in within the wider Fluence Yeah. Organization?

AMS was a plucky little startup out of California that came into the NEM in twenty nineteen to bring its new algorithmic bidding technology to the NEM, which we'll talk more about.

It was acquired by Fluence, a year and a half later, maybe in twenty twenty, and we've been part of Fluence ever since. And if listeners know Fluence, they probably know Fluence from its position as a leading global supplier of battery energy storage systems. You know, Fluence will supply the technology and do the EPC work for a battery project. And then done something like thirty eight gigawatt hours of of projects built or in various stages of commissioning around the world.

Fluent is well known for that. So I don't work for that part of Fluent. I work for the digital division of Fluent where we have two software products that we look after, software as a service offerings, one for asset managers and asset performance management software product called NISPRA. And the main product that I work on is called Mosaic, which is our algorithmic optimization and bidding platform for clean energy assets in Australia.

So we're working with a large proportion of the the NEM's semi scheduled wind and solar generation fleet and also a growing proportion of its grid scale battery fleet.

Awesome. And, I mean, very quickly, like, what is the difference in that context between, yeah, like, semi scheduled scheduled with the NEM?

That's NEM terminology. Mostly that just means big and grid scale in front of the meter, if you will. So in the NEM, any wind or solar farm that's thirty megawatts or more is semi scheduled, and any battery that is five megawatts or more is scheduled. And, actually, the new term for that is scheduled bidirectional unit. Yeah. So you might hear people say BDU. That just means a big grids big grid scale battery.

Yeah. You kind of, I guess, talked a bit about the, yeah, the Fluent Digital, like, part especially, like, the Mosaics team. Like, globally, how big is that team, and then, like, how much of that is focused on Australia?

The digital team at Fluence that works on Mosaic and NM is approaching twenty people now. So we do all aspects of kind of end to end customer journey stuff here in Australia. The people who design the product, the people who build the product, the people who sell the product, the people who deliver the product to customers are all based here in Australia.

Nice. And then yeah. So I guess of that twenty people, what is it? Yeah. Like, a lot of, like, data engineering, data science.

Yeah. We've got data scientists. We've got front end engineers. We've got back end engineers. We've got product managers on on the kind of development side, and then a whole team of, you know, energy traders and market experts and account managers help work with our customers to make sure they have a good experience using the software.

Yeah. I guess that is like like, what does, yeah, the the Mosaic portfolio look like in the in the NEM? You know? Is it all yeah. You mentioned kind of, like, the combination of renewables and batteries. Like, what is that ratio?

Yeah. When we when AMS first started out in Australia in twenty nineteen, we started out working exclusively with wind and solar farms largely while we waited for grid scale batteries to get built. It was a very kind of nascent market at the time. In twenty nineteen, there was probably only three or four grid scale batteries at all.

And it turns out bidding a wind or solar farm into the wholesale market here in the NEM is actually quite a complex game to play. You can't just leave a static bid in your wind or solar farm will lose money and, or leave money on the table. And so we helped pioneer the idea of using algorithms and automation to help trade wind and solar farms to help them maximize their revenue. They have quite tricky trade offs to manage between avoiding negative prices and avoiding being constrained off when there's congestion in the system. And so today, fast forward to five or six years later, the vast majority of winter solar farms in the now are using some form of of automation or automated rebidding technology like ours. And we're working with a significant portion of the wind and solar farms in the market.

And we're also working with a growing proportion of big grid scale batteries in the in the market. There's today eighteen operational grid scale batteries that are using our software to help form, operationalize, and execute their own trading strategies. And we're working with a wide spectrum of different types of customers and different types of assets and different size of assets, which we could talk a bit more about.

Yeah. So is that kind of batteries across because, yeah, it kind of you mentioned, like, the scheduled, then there's, like, non scheduled batteries as well. Is that across both of those?

Yes. We're working with, you know, of these, you know, first eighteen projects that are have started operating and are using our software. Some of them are as small as one megawatt. Those are non scheduled batteries and some of them are as large.

We're we're actually about start working with the largest battery in the power system, which is the Waratah super battery in New South Wales. That's eight fifty megawatts of load that'll be bit into the market as a scheduled load. So we're really covering the spectrum there from, from big to small. The small non scheduled batteries are interesting.

They have unique, some unique attributes. They're faster and easier to build, and there's less kind of red tape and grid connection process to get those built. But they're tricky to trade for their own reasons. They don't participate in central dispatch by bidding into AMO for permission to charge and discharge.

So our software plays a unique role and it it actually gives the battery at set point and tells it what it should be doing at any point in time.

And the big grid scale batteries have to submit bids into AMO. Everything bigger than five megawatts needs to submit bids into AMO and ask for permission if they wanna charge or discharge or sell and how much.

So yeah. So if we look at the kind of the the the range of things that needs to get done in both, I guess, controlling, bidding, dispatch, you know, what then is the role that, like, Mosaic plays in that? What is the role of humans within that?

Yeah. Almost all of the batteries that our software solution is involved with use some combination of software and human oversight to affect their trading. So we're working with a wide range of market participants who are working with some super experienced energy comp global energy companies that have their own twenty fourseven trading teams.

We're working with some new entrance IPPs who have never traded a dispatchable unit in the NEM before. We're working with some small developers who are building the first battery projects to small ones, just getting their, their feet wet with their first energy investments. And the challenge for us as a software company and a service provider is to provide a software product and a service offering that can meet the needs of this wide range of of asset owners and customer types. And we've tried to build a really flexible software offering.

Firstly, it's not just software. It's not just a software approach that we bring to the NEM. We provide all of our customers with a software product, but also a layer of services to support their use of that product. So we try and meet our customers where they're at.

We have some customers that are not super interested in hands on. They're pretty happy to let the algorithms drive most of the time, ninety something percent of the time. They kind of set and forget and they check-in with us every once in a while when they have a question or we check-in with them to see how it's going. But on the other end of the spectrum, we have some customers that are super interested in what's going on.

They're in there every day with their hands on, you you know, changing things around, laying on their own trading strategies, giving the algorithm algorithm a nudge, giving it a steer, giving it a hard override, and that's all totally fine. Like, we built this platform to let our customers trade their batteries however they want to trade them. We have several customers with portfolios. You know, the battery is not their only asset.

They've got load position retail customers. They're managing their battery not just for maximum wholesale revenue, but to manage their load position. We've got customers with network support agreements and contracts with network operators that require how the batteries operate at certain ways at certain times. And we've tried to build a a really flexible offering of software plus services to meet our customers where they're at and help them grow on their on their journey.

Nice. And then, yeah, that neatly brings me on to my next question, which is kind of, like, what is optimization within the context of the NEM? Like, kinda, you know, we look at battery optimization in the in the UK market, in the US market. So, yeah, what does that mean here? Like, is it the same as auto bidding? Yeah. Kind of what where does this terminology land?

It might be interesting to compare against GB where you've got this, you know, term people like to use an optimizer. And an optimizer in GB in general is almost like a managed service where the owner of the battery kind of gives the keys over to a specialist firm that, you know, has developed its own software and probably has its own team of human traders. And in combination with the human traders and the software, they trade the battery and, and kind of deliver the asset owner an outcome of revenue. And all the data is transparent and that allows, you know, companies like motor to make some pretty neat, you know, dashboards and leaderboards of, you know, which optimizers is performing the best.

And it's actually, the landscape is very different here in the NEM, but we don't consider ourselves an optimizer. In fact, I don't think there is a GB equivalent optimizer operating in the NEM. We see ourselves as a company that helps market participants, the traders of the batteries, form and execute their own trading strategies.

So we don't trade the batteries for our customers. We don't control their batteries. We don't control their trading strategies. Empowers them to build their own trading strategy and execute it mostly autonomously.

Okay. Interesting. And then I mean, so what is it about the design of the them that supports that kind of, like, difference in business model, I guess?

It's interesting. I think one of the main differences is that a lot of the first battery projects in GB seems to be owned by, like, funds, infrastructure investment funds that don't have backgrounds or capabilities in in energy trading and don't have twenty fourseven trading desks and experience doing that. So the market seemed to move towards kind of these specialist vendors coming in and and filling those gaps.

And the NEM, you might say the investor class has been different for the first crop of batteries that came in. And the NEM, the regulations are such that, you know, there's all only one participant who can be responsible for the the outcomes of the battery and settle with AMO. And it's on the hook for compliance compliance of bidding. And in general, in the NEM, we found that the typically, the owners of those batteries, but certainly the trader or the offtaker likes to stay in control of the bidding and the associated, you know, compliance obligations. Yeah.

Because I guess if we kind of break it down a bit. So, like, ultimately, you mentioned that the NEM is, like, essentially dispatch system.

And so I guess this changes some of the kind of, yeah, responsibilities that are needed in the act of, like, controlling a battery. So is it right that, like, a lot of it then falls on that bidding side? And so that's where this role of, like, algorithmic bidding tools comes in.

It might be interesting to do a bit of history on algorithmic algorithmic bidding in the m and how we got to this point. So I'd say that, you know, every generator in the history of probably any electricity market has had some form of software tool helping the trader of that asset, you know, form their bids, structure them correctly, and submit them into the market operator. And the NEM has been no different traders for decades have been using some form of tool to form and submit bids.

When first grid scale battery came along in twenty seventeen, the industry kind of realized like, oh wow, these, these assets are really flexible and they're really energy limited. And there's a lot of signals coming off them and a lot that's changing really fast. And actually the our software tools that we use to trade need to evolve to to manage the speed at which these batteries need to replan their state of energy or reforecast their operations and signal them through to the market. And so really all these tools are doing is trying to forecast market prices, plan the battery's operation through those prices over the next, you know, day or two ahead, how the battery's gonna manage its state of energy and determining the dispatch plan the battery wants.

When's it gonna charge? When's it gonna should it discharge? When should it sell FCAS and how much? And then it turns those dispatch plan the battery wants into a set of economic bids, you know, price quantity pairs that could submit it to the market operator.

The market operator accepts the bid. You're all good. You're executing your plan. If you bid too high or too low and market operator doesn't accept your bid, you've got a replan.

Unfortunately, the NEM being a very flexible market with almost no gate closure to speak of, lets you change your plan on a dime and reoptimize every five minutes if you've got your plan wrong.

Yeah. So the NEM is kind of operates entirely at this five minutely level, but you can essentially redeclare all of your bids pretty much up until the point at which that dispatch decision is made. Is that right?

That's right. Yeah. And there's, you know, several rules about the way the reasons you have to cite for rebidding, but there's so much changing at a at a battery every five minutes in terms of its physical availability that there's that in itself is almost always, you know, enough justification to slightly adjust your bids. And, yeah, the NEM's very fast moving.

It's, you know, it's pretty well known that, you know, everyone always talks about volatility in the NEM. It's it's very well known. I think that the NEM has volatile energy only pricing and that batteries historically have made, you know, a significant proportion of their annual revenue in a small number of days. That's all well understood.

But what I think is what really kind of puts the rubber on the road to me is just the fact that there's essentially no gate closure and trader pushes the button right now to adjust their bid for the next interval. That's gonna go straight through to AMO. They're gonna clear the market and twenty to thirty seconds later, that battery's gonna start ramping. So, like, those bids have this immediate physical impact on the market.

This isn't a day ahead a day ahead setup where, you know, all you're trying to do is day ahead forecasted top five you know, top four, bottom five prices. This is like a what do I wanna be doing, like, thirty seconds from now? Let me change my plan, update my plan.

Yeah. And I think I mean, because, obviously, one one of, like, the biggest constraints to a battery is state of charge.

And so how difficult is it to, like, manage that, you know, optimization in real time when you don't know what is necessarily gonna happen in the future, but you've got this limited pool of energy to use?

Oh, it's super hard, but it just means you've gotta be reforecasting, reoptimizing all the time. There's so many things that can cause state of charge to deviate from what you thought it was gonna be. You know, the ancillary services here are fairly dynamic. Anytime you're participating in those, they're gonna cause state of energy imbalances you are anticipating.

Also, the market can change very fast. The prices of can evolve very fast than them. So you need to be continually reforecasting prices as flows on interconnectors change, as AMO's supply and demand forecast change, as the wind and solar forecast changes, the market prices can evolve very rapidly. So it's important to be reforecasting continuously. Almost all the batteries in them have some form of algorithmic or at least automated system helping them update their bids every five minutes, and almost all the batteries in the them will rebid two hundred eighty eight times a day every five minutes. So it's it's very, very fast and very, very frequent.

Yeah. And I was gonna say, if anyone's ever looked at, like, the the NEM bid data for batteries Yes.

That's a big data.

A lot of data there.

And so, like, I guess, yeah, we made that comparison to GB earlier. I mean, there's a lot of kind of new companies coming into this the battery space in Australia and then then, you know, we've got a lot of new systems coming online in the next few years. For those companies maybe new to the, like, what is the biggest surprise? Like, kind of what would you highlight as the thing to look out for?

It's actually a really interesting landscape of who is not just owning, but who's controlling the batteries into them. You know, Moto's the experts on the numbers and tracking this stuff. But today, what is it? Something like two thousand megawatts ish of of operational batteries we've got.

If you look at all the commit operational plus committed projects that are in various stages of construction, it's something like ten gigawatts. Right? Ten thousand megawatts. And if you look at who's gonna be owning and controlling, who's gonna be the market participants for all this capacity, close to half of it is in the hands of brand new companies that are essentially new to the market or at least companies that have never traded or bid a scheduled unit before.

You know, they've never traded or been responsible for a dispatchable unit. And so there's a lot of new companies coming in who don't have experience or capabilities or deep benches of resources to do things like build their own bidding systems or run their own twenty fourseven trading teams on day one. And so there's a lot of new folks coming in who are they've invested, they've jumped in, and they but they need some experience and some some help getting their first battery projects going. And that's one of the the gaps that we're trying to help fill in the market.

Yeah. And so for those, I mean, yeah, like for those for those companies coming in, you know, what what do you typically see as something that they're not used used to maybe?

The comments we get from new folks, especially folks who are experienced operators of assets overseas, but have just built their first project into NEM. Firstly, everyone's just surprised the, the basic stuff, how fast the market moves, how quickly you can rebid. There's no gate closure. The fact that there's essentially no gate closure blows some people's minds.

One of the things that has caught new entrants out, and caught overseas, you know, investors by surprise is the role that grid constraints and congestion play at battery projects. So wind and solar farms, this is a well understood problem for wind and solar investors and wind and solar operators, congestion on the grid, your solar farm gets curtailed. That's bad for your revenues, bad for your business case. Lots of people focus on trying to solve that problem.

On the regulatory side, we're trying to help solve that problem through the bidding process. But everyone understands the concept. When there's congestion, you get local price adjustment, you get constrained off. There's nothing you can do about it.

I mean, on that quickly. So, yeah, like, this local price adjustment is kind of this thing which it's there, but it's kind of it's not obvious. And so for me coming into the market, it was something that I kind of wasn't wholly aware of. Do you mind just briefly just touching on what that is?

Yes. It's not a well appreciated aspect of the NEM. Everyone knows the NEM is a regional or zonal market with five regional prices in each of the five NEM regions, but few people appreciate that the NEM is actually dispatched notably. So in the eyes of AMO and their dispatch system, they're actually dispatching every unit dispatchable unit at its individual connection point with its own local price.

And so all units in the market are settled at the regional price. Everyone earns the same price, but everyone is dispatched based on their own unique local price. So when you're submitting bids, you're actually submitting bids with reference to your own local dispatch price. So for a wind or solar farm and a congested area of the grid, and there are several congested areas of the of the NEM of the power system.

When there's congestion, all the wind and solar farms need to bid as low as they can to compete with each other for the right to access to the limited amount of transmission capacity and the amount of generation that can get through to the load centers. And so that's what one of the things that our bidding system is doing to our Mosaic system is doing to add value. If you're for Windows Solar Farm, usually the worst thing that can happen to you is you get your generation constrained off and you get curtailed. For a battery, it's actually very different.

For a battery, the generator, when AMO builds a constraint formula, the generator is in that constraint formula, the load is also in that constraint formula. So if you're a battery, you might get your generation constrained on when the price is bad, you might get your generator constrained off when the price is good, or vice versa. You might get your load constrained on when the price of power is very expensive. You might get your load constrained off when the price is good and you'd like to be charging. And this can and has happened to to several battery projects in NM.

The only way to to manage this problem is with a bidding system that's got, you know, finely tuned logic and is paying attention to that local price, trying to forecast what it's gonna be, trying to put the bids in the right place so that generators and say get forced to buy power at seventeen thousand five hundred dollars a megawatt hour, which is the market price cap, which has happened to some battery projects.

Yeah. I was gonna say, like, it's a bit of a wild concept. So, I mean, maybe if we break that down a little bit because so what's happening actually happening there? So, yeah, so, obviously, there's a relative shortage of capacity in market, so prices spike.

But where the battery is located is sort of behind a area constraint. Is that right?

Yeah. The best example of this is probably southwestern New South Wales where there's several grid scale battery projects, but lots of generation capacity and only a few skinny transmission lines through to the load center in Sydney.

And anytime there's an outage of one of those transmission lines, all that generation in southwestern New South Wales has to compete with each other for the right to, serve the load in Sydney. And there's just too much generation relative to the amount of load that can flow through. And so they all have to drop their bids, their generation bids as low as possible to try and try and compete when the settlement price is good. And some of them lose out and not all of them have are able to get their generation away.

What's worse for a battery is not just getting your generation constrained off because that load is also in the AMO's constraint formula. And AMO is doing a solve to find the cheapest set of resources that can dispatch to balance supply and demand. Often to AMO, it looks cheaper for them to dispatch your batteries load, to make room for another cheaper generator to get its generation away. So you end up in this perverse situation where if you're not careful with your biddering and you leave your batteries load available to be dispatched, AMO might dispatch your batteries load on at the market price cap in order to make room for another, your neighbor to generate into that high price and, and receive that revenue.

So batteries need to be really, really careful with the way that they, they place their bids and the way they manage those local prices and bid in response to them. You asked earlier about types of situations that have surprised new battery owners or, investors from overseas operating their first battery project. So the fact that everyone's dispatched on their local price, but only settled on their regional price is actually a really important point to consider, especially when you are creating analytics around batteries and and things like benchmarks and leaderboards.

Batteries in New South Wales being a good example. A battery in Southwest New South Wales, a battery close to Sydney might as well be operating in different markets. They're they're not even living in the same them. We've got such different constraints. They've got actually got access to different prices.

So anytime you're looking at, you know, why did this battery do what it did, you've gotta check its local price because it may not it may not have even been allowed to generate into a price spike.

Yeah. And so, yeah, we've got these kind of five regions, but then within that, based on what transmission constraints there are, you can almost break that down to to more subregions.

I think the point is that anytime you're doing analysis of, you know, why did the battery do what it did, you've gotta look at its local price because without looking at the local price, you don't know the whole story.

And, yeah, again, like, I mean, if anyone's ever got some spare time, feel free to dig through the NEM web datasets to figure out what goes on in a single, like, dispatch interval to diagnose this sort of thing. And I guess, yeah, like you said, like, it's happening every five minutes. And so yeah. Unless you're kind of used to it, there's probably things out there which which you don't see coming.

So I guess if we look forward a bit, like, how do you see battery energy storage and that, you know, optimization bidding in the them?

You're gonna have more and more batteries coming into the market. You're gonna have larger and larger market share. They're gonna be controlled by increasingly by as portfolios, you know, single single market participants controlling multiple batteries and and facing a new challenge that not many folks in the market face today, which is, well, how do I manage these multiple batteries as a portfolio? These FCAS markets are shallow. I've got all these batteries, all this battery capacity.

How do I how lightly do I wanna walk in these FCAS markets, and how do I wanna be using these batteries, and how do they complement each other? So we're busy trying to build some new and interesting optimization where you can single participant could co optimize multiple batteries in concert. They could do things like consider their aggregate impact on price elasticity in the aggregate, which is a, a, a new and interesting concept. You know, as batteries get larger, they have the ability to move the market. And so trying to estimate price elasticity is a a real challenge and something that almost any large battery is trying to manage. And as traders build up bigger and bigger fleets, it's gonna be a bigger and bigger challenge to manage.

Another interesting emerging theme in them is we have a whole new class of assets that are under construction and in the process of being built, so called hybrid assets where you've got a battery, a big grid scale battery behind the same connection point as a wind farm or a solar farm. And these assets are gonna have the option of operating as if they were a single plant in the eyes of AMO.

And, there's lots of new technical challenges on on several fronts with bringing those to markets on the engineering side, registration side, the grid connection side, but also on the trading side and the way these assets are going to interface with the wholesale market and the the new complex things that bidding systems like ours are gonna need to be able to do for them to help them get the right outcomes to trade compliantly and to maximize their revenue is an an immense new challenge that we're already busy working on.

Yeah. I'm really looking forward to the first of those systems coming aligned with these, like, hybrid DU ID registrations.

I guess one other thing which which we're seeing, you know, we've had the first one coming online, is this thing called, like, virtual tolls. And that kind of leaves space for this bidding, like, provider as well as the toll provider in contrast to, like, kind of more traditional toll toll agreements. How do you see that kind of changing the role that companies like Fluent are playing?

It's definitely an emerging trend. Virtual tolls are the the flavor of the the moment it seems, and everyone everyone wants to to get one organized. Yeah. The interesting thing with the virtual toll is the you know, it's a purely financial contract. There's no no need to physically deliver that energy to to settle the contract. And it's yet another very complex job that you need your optimization and bidding system to perform for you. So if you are the the the seller of a virtual toll, you are the physical trader of the battery who has sold a virtual toll to an offtaker.

Each time you receive, you know, a nomination from your offtaker, you'll have this important decision to make, which is that do I physically deliver that nomination from my physical battery through the bidding process?

Or do I think I can earn more money by ignoring what the offtaker wants to do? And and instead doing my own thing with my physical battery and trusting that I'll make more more money overall by doing that. And what we're we're involved with several projects that are in various stages of formalizing and operationalizing their virtual tolls. And we see our role is, you know, enabling and empowering our customer, the, the market participant, the seller of the virtual toll to manage that toll, however they see fit. So you could think of a virtual toll as a spectrum where on one end of the spectrum, you're risk averse or your lenders have forced you to be risk averse and you always a hundred percent of the time wanna physically deliver what your offtaker is asking you for. So you you're, you're back to backing your financial risk with your physical outcomes.

And on the other end of the spectrum, you could decide to say, you know what? I'm just gonna take the risk and take the gamble. And I think I'm smarter than those guys. And I know what I wanna do with my physical battery, and I'm just gonna ignore the nominations a hundred percent of the time and do my own thing with the battery, or you might sit somewhere in the middle. And so our job as a software provider and in in a a solution provider to our customers is to help them figure out where they sit on that spectrum and give them the tools to trade where they wanna trade on that spectrum.

Nice. Yeah. I think we've got, like, many, many new virtual systems to come online. So definitely something to, yeah, see how it all breaks down in practice. Before I get onto my final two questions, I've got one other, which is, I guess, are there when you look at kind of other markets, like, do you think there are lessons that, you know, the Australian market, the NEM, can learn? Like, if we look at parts of the US or the UK, Europe, or vice versa, other lessons which the NEM can teach those markets.

I hope there's lots of information sharing going on between market operators and power system operators as as we the energy transition steam rolls ahead. But it does seem like there's lots of lessons to be learned from Australia. You know, there's so many power system firsts happening here. Things like minimum system demand and just the sheer volume of rooftop solar that we've got in the power system here. I imagine there's lots of eyes on Australia watching the way we we proceed with many things.

Thinking about the mechanics of battery trading, which is my world in battery market operations. There's a few lessons from overseas that I'd I'd love to let them to not learn and not adopt. You know, when I talk to my colleagues working on Mosaic in other markets in the US, in California and in Texas, the market operators there have put some tight restrictions on the way batteries trade and the the way batteries forecast themselves, and the grid operator puts a bunch of validations on the way batteries can bid. And it puts, you know, our team and the Mosaic team and our customers in this funny position where we have to forecast our make our own battery plan, forecast our own SOE, and then, forecast with the market operators plan for our batteries and how it thinks it's going to operate the battery and work with work around it.

And we we kind of need to plan around the market operators plan for our battery and seems like market operators in some markets don't trust batteries to manage their own SOE and their bids compliantly. And it seems like market operators might be worried that storage operators could put the power system at risk by mismanaging their bids or mismanaging their energy or the way they're signaling their availability. And I think the experience in the NEM improves that battery operators when they've got the right software tools can do those things compliantly because we've never really had any problems with that in the NEM and batteries have never done silly things like get their load and their generation dispatched at the same time or, you know, bid the battery available when actually the, you know, the state of energy is empty.

So I'd like to see that aspect stay stay the same in an m and some of those encroaching restrictions in the the dispatch process not make their way here.

Are there any other markets that First Digital are then providing kind of these optimization services in, and how does this compare to what is done in the NEM?

Yeah. Well, NEM is the first market that we brought the Mosaic technology to. Since brought it to California where we're helping several battery projects, you know, form and execute their trading strategies. Also brought it to ERCOT in Texas.

Brand new, the team just released a solution for MISO, which is the Midwestern part of the United States. And, the newest, newest thing we've done is we've just launched in Japan. So we've built a version of Mosaic for Japan. My team has been involved in that here in Australia, which has been really interesting.

The market in Japan, the size of the power system, the size of the market is five times the size of the market in Australia, and it's probably five times as complex to trade a battery in Japan as it is in the NEM. So we're really looking forward to getting stuck in there and and helping some projects find their feet.

I gotta ask. We need a feature episode on optimization.

That is a separate podcast.

That's a separate podcast. So yeah. So lastly, is there anything you would like to plug?

So given the opportunity to make a plug, I'll make a plug for our our sister software product we look after.

Mosaic is our algorithmic auto bidding platform. We have a sister software product called Nispera, which is an asset performance management software, for wind farms, solar farms, and grid scale battery operators.

And it's for asset managers. It like hoovers up all the data coming off the SCADA systems and runs some pretty impressive big data, big data analytics. It's doing some pretty cutting edge stuff around predictive maintenance, like using machine learning models to analyze part data, component plant data, and help asset managers anticipate when pieces of kit might fail. The reason asset performance management software solutions for batteries is an interesting plug and a relevant plug is that it's really the next frontier in marginal gains for energy trading.

So these asset performance management systems doing hardcore analytics, helping battery systems increase their tradable capacity, like their physical amount of energy they can make available. So you can really get the most out of your battery system when you're trading. So our team is busy right now linking those two software platforms together so that the, the insights that come out of the asset performance management software, NISRA, get fed in to the batteries control systems so they can do things like rebalance their cells more efficiently so that the tradable capacity made available to Mosaic, our bidding system, can increase, which translates into more revenue, which is what all asset owners are are really after at the end of the day.

Awesome. And so we always finish with the same question, which is what is your contrarian view?

I'll offer a contrarian view, but I'm gonna stay in my lane here on battery optimization and trading in the NEM. My contrarian view is that there's a perception out there that to have a successful battery project, all you need to do is build it and then bring in the highest performing auto bidding system available.

And I don't think that's true at all. And I think it's a bad procurement strategy.

Today, you know, any new grad with a modern data science degree and some experience with optimization theory and operations research techniques can spin up a optimization model that looks good in a simulation and achieves high performance in a, in a kind of desktop environment, but you need a lot more than just a high performing optimization model to be successful trading a battery in a NEM. You need like deep expertise around the new gritty nuances of how central dispatch works and NEMD works, local prices, cumulative prices, administered prices. You need to know how that all flows through to dispatch.

You need to have really deep expertise on the signals that come off a battery and how does this signal function that signal and what does this mean for my available capacity when the signal changes? And even just knowing what questions to ask of the battery system supplier during that process is, you know, we spent years and years kind of building up the IP to even know what right questions to ask.

And then you need a team of people who's really committed to reviewing outcomes and approving the software, like to be a good software provider and get good optimization outcomes.

You need to be really committed to getting it right. It really takes you know, bidding is really a lot more than just optimization.

Nice. I think that's a good way to finish up. Well, yeah, thanks a lot for coming on the podcast, Matt. And, yeah, well, thanks a lot for coming up to Sydney. And, yeah, maybe I'll see you in in in Hobart Yeah. Next month.

Thanks for having me. It's been a pleasure.

Awesome.