Green Giants – Episode 94

Wes Ashworth (05:05)

Absolutely. It’s a super cool journey. Obviously, timely and very relevant. This is a very hot topic. I think everybody wants to know more and more about. We’ll continue to kind of get into those specifics as we go. But your background, too. You came out of strategy consulting. What did that background help you see about batteries that pure technologists maybe often miss?

Lennart (05:24)

The interesting part of course, strategy consulting is I think very typical if you start your career with a business background, so that’s also what I did. I started my journey there into looking for big corporations into saying we have existing technologies, where can we apply this? We looked at different paths from the automotive sector, also in the energy sector. I was also a couple of other sectors, but ultimately having a technology and seeing how that fits in the market.

When I started talking with Micah and Stefan, they had a technology and together we started looking into the market and how that applies. I think that the unique skillset that you learn in strategy consulting versus other consulting elements or other professions is really usually you know where you’re starting. You know point A andyou don’t know point B, you don’t necessarily know where you want to move. You just know that this is supposed to be bigger or better than point A. Then it is the task for the strategy consultant to figure out how you get there. That I think is a very valuable skill set as well. If you start a company and you don’t really know what is the future going to look like. You’re looking at all these different trends and you imagine a future, you build a vision, you build an idea, and then ultimately the steps that you need to take to go there.

Having said that, think in startups particularly, it’s the speed of iteration and the work together with the customer that moves you to that point quicker and makes you more successful than having a huge strategic concept. But applying some of the methodologies helps you do that more effectively and efficiently. Having said that, I mean, as I said at the beginning, we thought at the beginning that the automotiveindustry is going to be the major driver, only to be completely overtaken by the demand for energy storage and the complexity that comes with it throughout the last couple of years, which means that as a company, we now predominantly focus on the energy storage sector and have very strong customers there in a solution that adds tremendous value in a rapidly expanding segment.

That’s exactly what you want. That’s the kind of flexibility I think you need as entrepreneurs to act according to how the market is shifting. The pitch in 2018 was kind of wild because you spoke to an audience where you said batteries are going to be the future of energy and transport or like mobility. You lost half the audience because people didn’t believe in it. Then the next thing is you have to convey the idea that then batteries are actually very complicated and the problem. You kind of putprovide the counterpoint to your first statement that batteries are the future, but battery are complicated. Then you provide a solution. By the time you arrive at your own solution, you’ve already lost 90 % of the audience. Making that pitch in 2018 was very difficult. Fast forward two years and people do understand that batteries absolutely are a core part of both renewables as well as transport or automotive, and that they are a problem.

Then at least you can skip these first two elements and just start talking about the solutions space more and how you actually contribute to helping people do their jobs easier or helping cars drive further and better and safer.

Wes Ashworth (08:34)

It’s cool to see that evolution. Obviously, that’s very quick too, from 2018 to now just how quickly it’s evolved, how quickly it’s changed and you guys have obviously gone along right along with that and had great success there. Over that time period as well, I know your understanding of let’s call it performance and storage has changed as well. Can you tell us about that sort of that that change in the evolution of performance?

Lennart (08:56)

As the industry matures, you do obviously see different problems having a higher emphasis. I think the initial hypothesis was that health, battery health, the degradation of cells, meaning the loss of capacity over time and over cycles, is going to be a major problem. I think that’s to a certain degree still true. It’s the unrecoverable capacity loss that ultimately degrades battery performance over time.

That still constitutes a major challenge. But then came battery fires. We had a number of fires in South Korea. We had a number of fires on home storage solutions. We had a number of fires on larger scale facilities and the industry learned. That is the way batteries are built is different now. The way battery fires are contained is different now. The way you prevent battery fires is different now.

Part of that is, of course, the analytics, which is the part that we provide. But then also you notice the bigger the storages get and the more complex the architectures get, the more difficult it is to bring all the battery cells into an alignment and to get the algorithms on the BMS, on the controls, so suddenly you have all these errors that are kind of stacking up on top of each other’s within a energy storage hierarchies. From cell to module to rack to the container through the PCS to the different transformers until they ultimately reach the point of interconnection, meaning the grid.

Now, with all these issues, you suddenly see that on the, as you call it, the balance of system side, you get all kinds of elements that you need to address. What used to be degradation is now a perspective that we call usable energy. For us as a company, it’s important to enable the users of our software to dissect the recoverable energy, so the part that is not yet usable, into the different elements that cause it to not be usable and give mitigation measures. Are there imbalances? How can you get rid of these imbalances? Is that just a rebalancing cycle? Do you need to replace cells? Do you need to replace modules? Are there issues on the algorithm side for the SOC?

Do they need to recalibrate? Do they need to be fixed as a whole? Sometimes it’s just a software fix. Sometimes it’s a maintenance fix. Then enable their teams to actually bring this performance back up to the levels you need. That is translating into hard revenue numbers. If you don’t have that performance, you can’t make money with a storage. That’s independent of any kind of market regulatory aspect.

Then that adds the other complexity. If you start in 2018, most batteries are kind of pilots. They may be connected to some other generation asset, maybe to a big demand side, so like a big factory that its peak shaving for. Really, it’s not part of a market mechanism. But the more it becomes part of a market mechanism in America, typically, CAISO in California, and ERCOT in Texas, the two massive gigawatt-hour markets, the more you need to adhere with these market regulations. That makes it so much more complicated and so much harder to do that. That helps, well, it’s obviously a task for us to then support our customers.

That brings us to that last bit that you don’t really see on pilot projects, but you definitely see on these mature scaling assets. That is, there’s a lot of work that goes into making sure that the market participation works and that the maintenance works and the operations work. To automate all these processes as much as possible. That starts by automating the data analytics part, but it goes on to kind of automating reporting, automating the calculation of KPIs that you need to ensure that your warranties work, automating the kind of reporting towards the QCs or the market participation entities, right?

All these aspects. Ultimately when you come back, it’s these pillars of what is the performance? Kind of degradation, recoverable energy. What’s the safety aspect? With that also comes all the bankability topics. Then lastly, how do you actually streamline the operations of the people that work with it? Now we see so many companies that had one storage, two storages, maybe even big ones, but it’s very limited technology. Suddenly they now expand their portfolios. They mostly buy new storages because it’s so difficult to develop them. Interconnection queues are long. You buy them, meaning different cell, different integrator, different EMS system, completely different data structure, completely different contractual obligations. How do you bring that all together without hiring a bunch of new engineers that do that for you?

Wes Ashworth (13:16)

I love the way you kind of get into it and the detail of it. It’s certainly a much more kind of honest, robust version of performance. Then cool to see that how that’s evolved. But that that focus on availability, usable energy, being ready to participate in the market, all those kind of pieces that you touched on is fantastic. Well, stick on this and just kind of maybe getting into more misconceptions. But when people talk about battery energy storage today, what do you think they still fundamentally misunderstand?

Lennart (14:05)

I think the key misunderstanding for most people seeing a battery or seeing a battery storage is it is literally a white box. Like it is a big container, usually painted in white for thermal reasons and thinking that it just works. You switch a button and it’s a battery you charge and discharge and then you don’t have to worry. But how many components are working in harmony in there or disharmony and how much work it actually is to work on them.

It is a very common misconception because that’s also what a lot of vendors want to have you think. I’m not going to name any vendors, but there is almost always, if not, it’s a different issue, a massive overbuild on these batteries. This is to account for all the operational challenges that you get, for all the imbalances, KPI readings, kind of offline components that can cause a battery to lose some capacity so that it can still participate at nameplate in the market because otherwise you get penalized.

Now, that might work in year one and two, but it gets worse and worse over time with degradation, increasing also the challenges of all the other issues that plague a battery. You might be very happy for the beginning of the battery life, but then problems accumulate. That I think is typical for people that are new, they’re like, but it works. Yes, it does for now, but you already see the issues and someone in the background is trying to fix that. There is vendors that don’t give you any access to data, so you never know what actually happens and you just can hope and ideally, you don’t only hope, you also have a contract with them that will ensure that performance throughout the entire lifetime, because otherwise you will have a gap at some point and then you’re responsible for it. Then it becomes a problem.

But I think it’s that misconception of a battery as this extremely easy to manage, easy to drive asset that is still sometimes out there. Just to give you one example, that 30 megawatt hour battery that I mentioned at the beginning, which was one of the first storages to connect, has 2 billion data points, I think, somewhere in that magnitude per day, just cell data. We’re not even talking about all the other alerts that are produced on a BESS. We’re looking at billions of data points already for small storages.

That’s just magnitudes bigger than anything anyone has ever seen on the solar side. A lot of the players in the market obviously come from solar or wind to complement their renewable portfolio now with that latest technology in the renewable space. They’re applying their mental framework from solar and from wind to batteries. Then they just completely miss the point on how dynamic a battery needs to be managed and what can be induced in terms of imbalances and inverter issues. All of these are kind of also already there on the solar side, but just completely different magnitudes and completely different mitigation mechanisms.

Wes Ashworth (16:49)

A huge misconception there. Appreciate you getting into that. We’re talking billions of data points. Not as simple as just that white box. It just works. We’ll get more into it as we go too. Appreciate you kind of setting the stage there, some foundational stuff. I want to shift a little bit into what you actually see when projects are operating and kind of going through some of those.  From the outside, storage looks like this big success story. From the inside though, where do things sort of quietly start going wrong? You started to touch on some of that there, but what are you seeing from your perspective?

Lennart (17:17)

One concrete example maybe it makes it easiest. It’s a California based storage, couple of hundred megawatt hours. They operated it. They had an overbuild of 30%. For simplicity sake, let’s say they had 500 megawatt hours build out there. They only needed 400 megawatt hours. I’m aware that that’s only 25 % overbuilt, but let’s take that.

They still missed their nameplate. It was four-hour system. They needed 100 megawatts per hour and for four hours. I’m 100 megawatts for four hours. They consistently missed that. They got penalties, they got derating, and they didn’t know what happened. That’s how we started working with them. Initially also their supplier was fairly antagonistic, didn’t really want to play ball, didn’t really want to work with us because obviously they had to pay liquidated damages for that missing capacity.

When we worked with them, we very quickly identified not only that it was imbalances, which I think was kind of obvious, but also where they were, how they built up, and what caused them. That meant that they actually started looking into the details. You have these thousands, it’s hundreds of containers in that case, full of battery racks and then full of cells. I mean, it’s hundreds of thousands of cells.

We pointed them to the, I think, 25 that were causing problems immediately. Then they could go in there, replace these modules, looked at them as well, figured it was a manufacturing defect. It was impossible for them to ever get the storage to run with these cells in place because they limit the operation of everything else. One single cell can then very rapidly degrade, depends on the architecture, but in that case, the architecture was set up in a way that very quickly degraded the capacity of that entire container massively. That meant that they lost very quickly, 20, 25 % of their capacity, which meant that they went under nameplate. Then if you stack up other issues that came, operation issues, you just miss nameplate. For the integrator, the nice thing was, was the manufacturing defect.

It actually goes back to the cell manufacturer so they could claim a warranty claim with their own supplier. You see these kind of back-to-back warranties that you have very often in this industry. They’re never perfect. You obviously always have your own share that you need to pay. There’s often deductibles that you need to take into consideration. But ultimately what happened throughout that process, we could go down to the root cause. The battery would then work not perfectly. No battery ever works perfectly.

But worked within what you would expect, absolutely met the nameplate and we brought back, I think the equivalent of $1 million in revenue to that asset per year in a very short period of time. That’s just troubleshooting. I have another example of a massive site also in California. We’re talking gigawatts here. We onboarded our solution and they wanted to test it.

You see the first kind of dashboards come up, the algorithms are tuned, and then you see the analytics running and showing you the weak spots of the system. Of course, every system has weak spots. But this system, given its huge amount of containers, look pretty green. It’s like a heat map in our solution that shows you what are the problematic components. I look at it, I’m like, huh, that’s all green. It’s nice. It’s good for the customer.

Might be difficult for us to prove the value. It’s going to be an interesting customer conversation. They started using it. I never really expected them to love it, given that what insights will they get. They were extremely excited, because the reason everything was green was that they had people on site immediately running the moment any alert turned orange. That was a huge effort and they called it.

They just ran around slapping Band-Aids on the storage whenever something came up. With our solution, they suddenly saw a lot of issues before they were actually arising. They could prevent or preempt this, move from reactive to proactive maintenance. That made a massive change to their operations. While the performance technically didn’t really improve, the way they looked at it was, well, now we have the resources available to actuallyproactively manage this and root cause analysis that used to take us two months to figure out what went wrong suddenly were available and visible with the click of a button. They absolutely loved it and they applied it in retrospect and that was a part that I couldn’t obviously see but they used some examples from the past where it had taken them months to find out what actually went wrong and the storage was down and it cost them millions and they wanted to figure out, okay, what caused this. They did find it, but it was an engineer looking at Excel sheets, building his own dashboards and trying to find it. Ultimately, it was just flagged immediately as, well, this inverter went down because you had a DC bus imbalance there, and this is what you need to do to bring it back up. A very, very kind of direct connection of what happened and just accelerating this so much.

It combines nicely and ties together this idea of, you want performance, but what might be problematic is the operational aspect of it. How do you actually do it and how much of it can you automate? How much of it can you just make easier? Ultimately, I think any software solution is either doing anything completely automatic or it just makes the life of the user easier. I think we’ve achieved that a number of times.

Wes Ashworth (22:32)

Amazing stories really paint the picture of one, the problem that is out there and then obviously TWAICE’s solution and why that is so important and so valuable. We’re talking millions of dollars there and all of that as well, too. It’s funny that second story, I just had this like visual imagery of these people just running around frantically, like solving all the things, the alarm goes off and they’re running around and you’re like, OK, it was working, but it wasn’t sustainable. Do you want to do that? No, of course not.

Obviously these, these great solutions exist, but that’s cool. Even, even one where you look at it and you’re go like, everything’s green. Like it looks like it’s okay. We’ll see what happens. But, but really the underlying is they needed it even more so.

Lennart (23:21)

The sad truth is, you know from computer kind of troubleshooting, have you tried turning it off and on again? Even in the BESS industry, I think a very, very common maintenance mechanism is resetting the component and just hoping that the error doesn’t come up again. I mean, that might work, might also be the right thing to do, but it might not be. Then you just kind of carry on this problem until it comes up again in the future. In the worst case, it’s maybe also something that might be safety relevant. You might even cause an even bigger problem down the road. That obviously, this idea of really understanding what caused the issue is fundamental there.

Wes Ashworth (24:00)

Absolutely. You’ve mentioned too, even some people kind of look at it go like, we don’t really have that many issues, but you’ve mentioned too even like a small number of faulty modules can really drag down an entire system. Can you tell us a little bit about that in layman’s terms? Like, why does that happen and what should people be thinking about?

Lennart (24:16)

The topic of imbalances and kind of different components limiting each other is if you think of different… I mean, I think liquids are imperfect because the liquids would just flow where there’s space. But essentially, you could think of open glasses, and you cannot have a spillover because the moment it spills over, it causes a short circuit of fire. It’s essentially what happens in a battery.

You have eight glasses put next to each other and you can drain them and you can fill them up with one pipe. All of them are filled equally. Now the problem is with all kind of frictions and imperfections in that, I mean, to stay in the image in the piping, they are not evenly filled up. Now you need to stop though when the first glass is kind of reaching its limit because if you charge more, it will spill over. Same for battery.

You can’t overcharge it. It’s actually one of the more dangerous things to do because that might actually cause a thermal runaway fire. Right? You stop. Now you stop and all these other glasses still have space. This space represents capacity, in this case, mean, water that cannot be used in battery terms, well, it’s capacity that gets unused and that can stack up quite dramatically, especially if the more you put together into one of these groups.

The same applies if you drain it. Well, the first time the glass is empty, you need to stop. kind you can’t go less than zero, right? That is deep discharging can cause similar issues as overcharging. Ultimately, then all the water that is still in there is also lost capacity. If you add all that up, you very quickly see, OK, I’m missing out on 20 % of what I can use here. Obviously, that numbers can be infinitely higher or lower and especially is amplified by different system architectures that prioritize putting more in serious there.

That becomes a huge problem on the imbalanced side. That’s where you kind of, in battery terms, you have balancing algorithms or balancing protocols meaning essentially like a drip current that goes from one battery to another or like water spilling over from one glass to another, but that takes time and that needs to happen in specific state of charge windows. You need to trigger that proactively. The problem is if the imbalances are too bad, sometimes you are too late to be able to trigger that because some of the batteries will never reach the state of charge that they need to reach in order to actually qualify for that balancing.

To identify what’s the right approach now and to validate that the balancing actually worked is something that obviously you need some degree of intelligence or analytics for.

Wes Ashworth (26:51)

Absolutely. Good analogy. For those listening, they’re dealing with storage projects. They’re seeing and hearing storage projects often or experiencing it themselves. They end up in disputes between owners, integrators, and suppliers. Talk to me about TWAICE and just how does data change those conversations and really create a solution here?

Lennart (27:11)

Data is such an interesting aspect because BESS is, I think, is one of the few assets where you really get almost unlimited amounts of data. So many, it’s a fully digital system, so many data points flow somewhere. That’s, think that’s the first conversation you need to have as someone who operates batteries is, A, there’s data. What do we want to do with that data? What’s our data strategy?

Who’s using the data? Who’s collecting the data? Who’s putting the data in a data lake? Who can then use the data? That’s very often overlooked. We just conducted a big survey in the industry. It’s called the best pro survey, where we ask over 100, I think 117 active operators, what are your challenges? 85 % said they have ownership and access to data. 10% are working on it.

Very high amount of people say we actually own data or have access. The interesting part is that one of the main challenges is making it usable or getting access to it. Well, that seems like a complete contrary position, but it comes from the fact that the amount of data is actually so big that it is not that easy to access.

Often it resides on a local data historian, maybe it’s used in a local EMS, but not necessarily saved, but to bring data into a data lake in the cloud and then make it usable and actually interpret it at scale is very, very challenging. That’s where TWAICE comes in. Our customers often utilize different approaches to data collection. We have preferences. We have our own data collection that can work redundantly from kind of existing EMS structures and are very powerful because it does create this unique data lake that enables you to tie in different tools and analytics that don’t only have to be TWAICE, but can be different elements as well.

It can be your own team kind of running queries on this or running analysis on that, building their own Grafana dashboards. That’s the number one. How do you get the data from the storage into the cloud? Finding a reliable and robust way there. Then how do you make sure that there is a scalable way of getting insights from this data? That is A, requires, of course, like a powerful front end that makes it easily accessible and usable and kind of replicates the workflows of the people and automates them. But in between, you also need a layer of analytics and KPIs that ensures that errors don’t just get reported upwards. Very typical errors are on the state of charge side. You see sometimes 20 % percentage points off errors. They also accumulate in different components.

All that interpretation, making sure that the sensor logic is, or the data logic is correct so that you see that you can actually compare if there’s an error, that you see the associated components immediately next to it and see what are they doing. Is that like an isolated problem? Is that related to that component only? Or is it caused by some connected components? Typical thing is that imbalances can cause an inverter or PCS to trip.

Do you want to fix the PCS or is the problem actually more on the DC side, so more on the battery side? Where do you send your maintenance to ultimately? To have that middle layer of advanced KPIs to correct the basic KPIs, to identify what is just a sensor noise, what is an actual error, what is an actual risk, and feed that then into that kind of user interface layer and automation layer that might connect with other tools or with a human in the loop that then looks at it and takes a calling action.

Wes Ashworth (30:45)

I love the way you spell that out. I’m kind of like thinking, how would you go about it without a solution like this? When you’re talking about billions of data points, being able to interpret it, use it correctly, being able to know what to do with it, get down to those root causes. I mean, that’s, that’s a highly complex thing. How critically important this is. I’m sure there’s, there’s lots that are listening. Hopefully that’ll, they’ll reach out to you about that as well, too. You mentioned something too, that I think is interesting.

We see people that have maybe one or two kind of smaller scale, right? These problems maybe don’t show up as pronounced or there’s a bit of heroics that kind of keep it flowing. You gave a couple of those examples. Then as they start to scale up, that then these problems start to really explode. guess sort of beyond the obvious, like why is that? Then what can be done about that early on?

Lennart (31:32)

You have a number of really, really smart people in the industry on the performance engineering side that know an amazing amount about batteries. They love to do that themselves. It makes total sense. If you have a small storage, you want to learn about it, build it yourself, figure it out, and invest the time. Now, the problem is, of course, the time you’re investing into building this out, something that technically also exists in the market you could just buy.

This means that you can’t focus on other aspects. You can’t focus on actually optimizing the performance. You can’t focus on expanding your own platform and kind of reviewing new projects, building new projects. But for a pilot, I think it’s kind of very compelling because you do actually want to build that in-house expertise about batteries. You start digging into that yourself. I think there’s a lot of players out there that have created some kind of data pipeline, that have created some kind of Grafana dashboard, built their own alerts.

It works, it might not be the most pretty, it might not be the most stable. Usually, it’s only one person that really understands it and that’s the person that built it, especially that’s probably the only person that can maintain it. I already kind of hinted at the, I think, issues when you scale or when as an organization you want to de-risk against being reliable only on one person. But it becomes particularly problematic once you move to bigger terms because then you also need to account for the large amounts of data.

How do you ensure that this is done efficiently and you’re not racking up cloud costs very quickly because Amazon, AWS, Azure, or whoever, they love it because it’s huge amounts of data that need to be processed, but that can become very expensive very quickly. Then you need to suddenly look into, how do we optimize that? Then you get a number of different storages with different data points with different logics. You need to make sure that your original data logic also applies to these. Then suddenly the only thing you’re doing is trying to maintain a data platform.

Then you’re starting to more or less compete with companies like TWAICE that has an entire team of 50 engineers looking at maintaining that data platform to ensure it works across all batteries, all battery platforms and reliable. Then you very quickly, I think, risking of just rebuilding something that is out there in the market at a lot of a higher cost instead of leveraging your resources to really focus on their core jobs, which is to make sure that the batteries work, that the platforms can scale, and that you meet your performance targets.

Wes Ashworth (33:51)

On the other side of that, if an operator is out there and they want to want to scale correctly, not have these issues, they want to think in 20 year asset terms, your perspective advice, like what has to change about how they run systems today and what should they be thinking about instead?

Lennart (34:05)

It’s really that thought about, how much of the value chain do you want to own? I think that is extremely important strategic decision. And then how do you go about the rest? Then how much of the O&M do you outsource? How much do you rely on LTSAs? If you negotiate LTSAs, how do you make sure that these are enforceable? How do you make sure that you can actually get the suppliers to do the job that they signed up to do and just to give one example, one of our customers, we’re monitoring the LTSAs for them. We looked at the contracts, it’s all visible in our platform. It’s almost 100 % representation of their contracts. But our engineers looked at it and were like, we don’t think that they will ever be able to get any money from their supplier because the way the LTSAs are written, it’s almost impossible for their supplier to break it. Now, that’s bad. That’s not ideal.

I mean, that’s something you can not really fix later on because that’s a contract you signed up front. But it’s also something if you do rely on that, you need to make sure that you get that right? That complete idea of who’s going to be responsible for what, how are we going to make this enforceable and what needs to be in the contracts to do that? Also, if they don’t do it, how can we step in? If they go out of business, I mean, the example of Powin in last year. when they went out of business and caused this entire cascade of operators suddenly not having someone to stand in for their guarantees. It’s like, what happens then? How do we ensure that we still get access to the technology, that we’d still get access to the software and that we are still able to operate?

To have that concept really worked out and validated. Then I think there’s no way around really ensuring that you have that data access and then you have the capability to use that data. That can have different forms and we also account for that. I mean, most of what we talked about was kind of more on the operations and maintenance side. Performance engineering, very deep time series data, very advanced KPIs.

There’s the other extreme as well where you get just a report a month or quarter that says, yeah, this is what your storage has been doing. This is the amount of cycles you’ve been doing, this is the current degradation and availability numbers of the storage grade, all in the green or not grade, this is all red and this is what you should send your supplier to fix. Ideally you do that a bit quicker, so maybe you get like a daily alert if something happens, but it is problematic and it almost doesn’t matter even if you say we only focus on the commercialization and the entire technical side is irrelevant to us.

Just had a fantastic conversation with someone in that market who said, well, I’m sending, so they had a tolling agreement. It’s basically a battery that is being used by another entity and their only responsibility is to make sure it works. Then there’s different KPIs. They cannot use it more than one cycle a day, but it needs to be available, I don’t know, 98 % of the time, 100 % of the time for specific periods. Doesn’t matter, right? Ultimately then they send an invoice every month.

That is a fixed fee and then some deviations depending on whether they did more cycles or whether the availability wasn’t quite there. Penalties or bonuses. They sent the same invoice every month because they don’t really know how much the battery has been cycled. They can’t even charge more. They also really don’t know how much it has been available, but that is something that probably the customer is figuring out with complaints and saying like, at the point of interconnection, that didn’t work. Now, of course, where that happened on the battery side, you don’t know. You might not even know who to speak to about your warranty claim. Talking about this quite extensively, but to get that transparency to know how your business is impacted is fundamental.

Wes Ashworth (37:38)

I think as we move towards scale, obviously becoming much more proactive, much more strategic and doing this the right way so you don’t end up with some of these bad examples that we’ve all heard and witnessed as well too. I’ll ask you a few more questions as we get into it. When you’re seeing it, from the data perspective, when a system is maybe heading towards a serious problem, whether it’s performance or safety, what do you typically see first in the data? Like what are some of those early warning signs?

Lennart (38:09)

That is opening an entire can of worms here, I think, on how do you see that? It’s really, I think, it’s a whole range of things that can cause issues. It’s to understand that entire width and to tie it back to what fundamentally causes that is the important part.

If we just take a look at what causes fires, for example, we did an extensive study with EPRI, which is the research arm of the American utilities, what causes battery fires and which components and what phase? You can check out the study online. But what was interesting is that only 11% was actually tied to the battery cell and in the battery cell that’s then usually it’s manufacturing defects. What you could see is resistance increases, anomalies, self-discharging happening.

These are all indicators that there’s something wrong with the cell. But the majority, 45%, was coming from controls. So really, the BMS or the EMS, overcharging, deep discharging, so kind of causing glitches in the system. I mean, it sounds trivial, but if you’re charging so many cells at the same time, slight coding errors can mean that the one cell at the end is just being overcharged. That was 45%. Then 40 %, more like 43 % was related to the balance of systems. It’s just electrical fires. An HVAC system failing. So, identifying that and identifying there’s a heat spot in the system. Like there’s a temperature anomaly that’s coming up. Identifying that and tying it back, the HVAC system is failing. In due cause, this can cause a fire.

Now looking at imbalances, now this is tied to that potentially, right? You can have like thermal imbalances, meaning that battery cells are charging and discharging differently, so causing imbalances in the system that are stronger than needed. Now you can obviously go back and increase the amount of balancing you want to do, but maybe you also want to revisit the cooling system. Now if you do that and you ramp up the power of the cooling system, your round-trip efficiency might suffer from that. You kind of see that causality that exists on the battery storage and why it’s so important to understand how they all tie together, but also to get a holistic picture.

The risk, I think, if you are just focusing on that one red light that comes up, then you might miss the bigger picture of what else has been happening and how can I go one step further and solve a systematic issue that I might see on the storage instead of just solving that one kind of mini symptom, I think, if you were a doctor treating the symptoms or the actual disease and to really look into that and I think that’s a big issue there.

Wes Ashworth (40:51)

I love how you put that into comparison with the doctor. I think that really paints the picture really well. You could treat the one little thing, but it’s like, what else is out of balance that’s causing that? Why is that? Just the importance that I’m oversimplifying it, but just massive amounts of data and being able read and interpret that correctly, catch those things early. Just how critical that that is and huge cost savings and downtime. We can go on and on safety concerns and all that as well through that, too. You did mention so like that you believe the safest systems often end up being the most profitable ones. I think that’s a powerful point. Why do you believe that? Tell us a little bit about that.

Lennart (41:27)

I think the safer systems usually should be the most performing ones as well. You could make the case, of course, if you just cycle in a very conservative range, the battery might be safer. But the interesting part is that if you’re proactively managing your storage and you proactively replace components that are dragging the performance down, these are usually also components that in due time might cause a safety risk.

The more you see components on the system failing, the more likely it is that something goes wrong on the controls or something goes wrong on the cell side as well. Or you get like just a occurrence that you don’t want to have in the battery and then ultimately causing these causes for failure of the battery. Do you just react to something happening and that might be potentially catastrophic or do you generally use predictive maintenance, preventive maintenance to ensure that none of the components ever cause an issue? If you do that, that usually also translates to very, very high performance availability and therefore high revenues. Now, if you look at the costs, of course, you kind of need to find that ideal balance of saying, like, how much do we want to invest in cycles? How much do we invest in truck roles to ensure that it’s safe and profitable without costing us more than it brings in. That is this business case perspective is fundamental throughout. It starts with the data because collecting all the data and put all the data in the cloud and run all the data all the time, unbelievably expensive, doesn’t make sense. Figure out which data points do you need and how do you run them efficiently and so that they actually provide the insights to you. Same goes, understand which imbalances cost you money or which components that are down cost you money.

But also understand that if there is a component broken and you do a truck roll, well, fix it. Because then you’re going out there anyway, and that’s going to be a cheap fix for you than just having it run until suddenly it becomes an emergency. Then you need to have another truck roll that you didn’t expect. Really do it smartly and do it in understanding what the impacts are. Oftentimes it’s not necessarily impacting the cost side so much as it is just an activity to do. From running the imbalance cycles in off times that doesn’t necessarily cost you too much money. To time this right, to understand when it’s necessary, and to just run a business case in that.

Wes Ashworth (43:48)

Absolutely. I think it’s so important to look at it that way and kind of seeing that just a holistic picture how that all ties together and does affect performance and profitability. At the end of the day, there’s a great business case for it. As we get right up on time, I’ll start to close a little bit with some couple of questions. Just thinking about battery storage, battery storage data, everything that you’re doing as well. When you’re looking ahead to the future, five, 10 years out, what are some things maybe you’re excited about? What are you looking forward to? What are some things you’re paying attention to?

Lennart (44:15)

As looking at the energy industry in general, it’s extremely exciting to see, I’m not sure from an ecological standpoint, but from an industry standpoint, it’s extremely interesting to see that the overall demand for energy is growing. That’s the first time in decades. We suddenly see a shift away from static generation to, we need more.

This more is likely going to come from renewables because it has the lowest levelized costs of energy or electricity. It is very quick to deploy, but it has a massive intermittency issue. You will need some degree of energy storage to balance this out. Batteries are currently in the front spot there to provide that intermittency coverage. Now, the other element, of course, is on data centers and their unique operating way and the oscillation issues, the speed of deployment, kind of the huge loads that they put on the grid, but also the risk of curtailment, which means that they almost automatically will be coupled with massive batteries just to ensure their uptime and just to ensure that they’re not becoming too much of a liability for the grid.

I think this entire trend of AI. Let’s see whether all the data centers that they’re planning at the moment will actually be built. But I think we will see a massive rollout of AI through all elements of the knowledge economy or the economy in general. That will require processing power and that will require build out maybe not only in the next two years, but I think over the next decades and that will drive ultimately load on the grid. I think in conjunction, that’s a very exciting time for renewables, hopefully renewables, because that’s the only way of doing it somewhat climate neutral. That I think is something that I’m very excited to contribute to and work on in the next couple of years.

Wes Ashworth (46:03)

Absolutely. Same, a lot to be excited about going forward and heavy growth that’s coming as well. Final question, anything else you want to add? Anything you want to leave the audience with that you didn’t get to share? I’ll just let you take it anywhere you want to go.

Lennart (46:14)

I think maybe one aspect and that I really love that about the industry is how small it is in terms of the people that are actively working on the industry and the amount of collaboration that we’re currently seeing. But on the flip side of that, I think everyone should be aware or is aware that there’s a huge lack of talent in the industry and that is going from simple tech to that are actually going out on the sides, up to the performance engineers and even the asset managers. The amount of customer conversations I’m having where the call ends with a, hey by the way, if you know someone who could support our team in that, which my answer to that would usually be then I would hire them myself.

But it’s that lack of talent and the limited capabilities of just scaling the teams in accordance with your portfolios that I think is something that will probably change over time where we see more talent flowing into that industry, but it is already becoming a massive kind of bottleneck for a lot of companies in scaling and a lot of companies in operating their batteries. I think that’s the way you guys come in.

Wes Ashworth (47:15)

Absolutely. Great way to end it, I agree. Obviously, I see that firsthand and what I do. I think there’s a true kind of war on talent. To be able to win in that race is going to be important. You can’t do it without the people. As you said, the growth ahead, that’s only going to get worse. Start being proactive, get on that now. But a great way to wrap it up, Lennart thank you so much for bringing grounded, honest view of what it really takes to operate battery storage at scale.

Your perspective really cuts through a lot of the noise focused on which actually matters over the life of these assets. I appreciate just the holistic view of that and some of the solutions there. To everyone listening, thank you for joining us on Green Giants. Please check the show notes. I’ll include several links there for you to check out and learn more about TWAICE and their work. If you found this conversation valuable, please subscribe to the show, leave a rating, and share the episode with your network. With that, we will see you next time.