How do I prolong my EV battery's lifetime? - Prof. David Howey

Battery Generation by Patrick Olsen and Lena Peters. Brought to you by Celeste. This podcast is brought to you by the Helmholtz Institute Un and Celeste, the Center for Electrochemical Energy Storage, Un and Casual Germany. Welcome back to Battery Generation, your podcast on electromobility and European battery research. We will once more talk about battery state of health and degradation of batteries. And our guest is once more Professor David Howey from Oxford UK. Hello there again. Hi everyone, great to see you again. Thanks for inviting me back. Let me introduce you to everyone who missed the first episode, Professor Howey. You are a professor of engineering science at the University of Oxford. You hold a tutorial fellowship at St. Hildes College. And your current research is focused on modeling and managing energy storage systems for EVs, as well as grid and off-grid power systems. Professor Howey, we got some 10 to 15 questions, especially for you. Here we go for the first questions by an anonymous listener via email. Please ask Professor Howey why battery lifetime models are needed in the first place. Yes, we all need to use batteries better. I think that's what you mentioned in the first podcast. But here she says, 99% of us consumers don't really have a choice in our everyday life. Do we? Please provide some industry examples where battery simulations come to life. First of all, I'm really excited that people have questions on this stuff because it's like my super nerdy topic. So it's great to have like other people, you know, welcome to the cave of like battery nerd interest. So yeah, so this is a great question about why would we bother with modeling battery life in the first case? And actually, it's a question I get asked quite a lot. I mean, I think there's rightfully some skepticism about or some perhaps some questions about the value that modeling brings right to batteries. And so I think there's maybe a couple of things we could say about that. The first thing is that we use modeling to improve design, right? So that's one one scenario. The second one is that we can improve usage. And I think the question even mentioned this as well. And so I'll talk about those two things maybe for a minute. So the first thing we could say is from a design point of view, we would like to understand how to build batteries in such a way that they they last for the right amount of time. Okay, we don't want them to be over engineered. We don't don't want them to be under engineered like any other product. Okay. And so what we can do with models is try to predict for a given usage cycle what a given type of battery will do in terms of lifetime. And so this might help you if you're for example, building a new battery system to choose the right kind of cell for using batteries better. I think the answer to that is it depends who you are, right? So if I'm just a consumer, I have my iPhone. Maybe there's not too much I can change, right? I want my battery to be fully charged when I go out the door in the morning and then I plug it in when I get home. But even on something like a mobile phone, right? There's usually built in nowadays some kind of battery lifetime stuff and you've probably seen this, right? So on my phone, I could choose like not to charge beyond 80% unless I really need it urgently. And so I agree that it's kind of a limited choice, but already we have some choices about how to change our usage to prolong the battery lifetime. And behind this is in a sense a model, right? I think where it becomes much more important is if I'm dealing with very large assets. So I'm like a fleet of running a fleet of different taxis or buses, cost millions and millions. Maybe I'm running a huge grid battery. In that case, actually there's an investor behind me, investor wants to know what's the return on investment. There's also going to be decisions about maintenance, logistics, ordering spare parts. When do I replace stuff and that kind of thing? It's also very crucial for how do I price a warranty? So when I buy a electric car or a grid battery, it's going to come with a warranty. How much does that cost? How much should it cost? And so these are important questions where lifetime data and lifetime models I think are key. So I think the usual way that people do this stuff at the moment is to get lots of battery cells, put them in a lab for a year or two and test them and see what the lifetime is doing and then construct warranty or something from that data. The problem with that is it's limited to the data set that you have. And I think we are still in a difficult situation trying to extrapolate lifetime beyond two, three, five years. And therefore, this is where we actually do need some physical understanding of degradation to build models that we... They're not going to be completely correct, but at least they might tell us the difference between using the battery in one way and using the battery in the other way over 15 years, for example. We have to move on, but I'm really happy. I learned a lot. It's very important for prediction, very important for the design of batteries, the battery systems. Then again, displaying state of charge, state of health. Question number two of 10 or 15 questions comes from Alan via email. Professor, how he stated that a commercial battery was a black box. And there is not a lot you can measure besides few parameters. You mentioned the electric current, voltage and temperature, if you remember. On the other hand, his research seems quite complex. How does that fit together? One thing we need to understand is what can we get from the limited data that we have, right? So if we only have current voltage and temperature, what does that tell us? Which parameters in the battery affect those things that we measure? Then the second thing is trying to get more data from other types of measurements. And I know that there's lots of efforts around Europe and other places to build sensors into batteries, add reference electrodes, temperature sensors and so on. And I think that's great. The more the more data we can get, the better. All right. Then let's move on to the next question by Manfred. I guess a German listener sounds like we see astonishing battery cooling and heating innovations from China. The C-Line battery by C-A-T-L has an advanced construction of cooling elements for fast charging and a heating system for lower outside temperatures. What role does temperature play on lithium ion batteries? OK. I should say that I don't know much about that specific battery, but I'll comment on the final question, what role does temperature have? It affects the battery behavior in a few different ways. So one way that impacts behavior is by changing the reaction kinetics. So if you go to higher temperatures, everything kind of speeds up. You could say the battery performance improves. And what I mean by that is that the internal resistance drops or the internal over potentials decrease at higher temperature. And that's good because the battery is generating less heat and everything is kind of running better. I think the downside of that is that all the side reactions that lead to capacity fade also increase at high temperature. So there's no free lunch. But yeah, so temperature impacts kind of internal resistance and therefore kind of performance. And in fact, I think I mentioned the previous podcast. The opposite is also true. So if I want a fast charger battery, then low temperatures can be a problem because it leads to higher internal resistance. And effectively, you can push the voltage of the negative electrode below the voltage where lithium can plate directly as metallic lithium. You'd never want to have metallic lithium in a standard lithium-ion cell. Solid state, that's a whole other discussion. But in a standard lithium-ion cell, the lithium is always in a compound, right? And so yeah, if you push the voltage of the graphite below zero with respect to lithium, then you can plate metallic lithium and that's a very bad degradation mechanism. And this gets worse at lower temperatures and higher currents. So it's important for fast charging. It's important for aging, long term aging. But also temperature uniformity within the battery is important. We don't want some parts to be aging faster than other parts. And across cells in a pack, if I got a big pack, I want each cell to be more or less at the same temperature. We need high temperature for charging and we need, respectively, low temperatures to not make batteries age. So there's two emails that we got from Sophia and from Dan. Sophia is talking about an LFP home storage system. She texted Professor Howey, talked about penalties from grid battery producers. These temperature windows off around 22 degrees Celsius must be very important to battery degradation. I own an LFP home storage system by BYD which is known for bad load temperature characteristics. Should I worry about the temperature in my cellar? That's probably the place where she keeps the home storage system. Yeah, great question. I also have a home battery myself. And it's also an LFP system. It's not from BYD. Yeah, it's really interesting question. And I think there's two aspects to this question. So one aspect is the long-term degradation behavior and the other aspect is the short-term SOC estimation. So lithium ion phosphate is a really interesting battery technology, really exciting battery technology actually. It doesn't have many of the materials in the NMC cells have. So it's great from that point of view. But one of the things that happens with lithium ion phosphate is there's something called a hysteresis in the open circuit voltage. And what that means is that if I charge the battery really, really slowly and then discharge it slowly, there's a small difference between the charge and discharge voltage curves. This hysteresis can cause problems with the state-charge estimation algorithm because the algorithm is looking at that voltage from a lookup table or something as a kind of like bookend or a backstop to kind of compare against and make sure that all of the Coulomb counting, which is how a new state-charge estimation is accurate. And so if there's uncertainty about exactly what is the open circuit voltage? It's a bit like I'm shifting the level of the ground and I'm trying to measure the height and I don't know what the ground level is. And so that's a big challenge. And that uncertainty gets is bigger at lower temperature. So the hysteresis behavior at lower temperatures is more confusing. And we can talk about how to solve that problem. But I'll just say this. I also have a home storage battery and I'm really interested to see what happens in winter and we only bought it a few months ago. So drop me an email to fear and we can compare notes and see what happens. The other question is about long-term behavior and I think that the issue there is, am I degrading the battery by leaving it at high temperatures at high state of charge? That's one mechanism. The other mechanism is I'm trying to charge it at high currents at low temperatures. And I think both are somewhat concerning in home energy storage systems. The first one is concerning because many of the control systems in these batteries they don't allow you to change the upper state of charge limit. And I've discovered this with my own battery and I'm trying to get the manufacturer to like engage with me on this. But in summer at the moment, I've sold the panels on my house. We're generating way more energy than we need and therefore the battery is sitting at a 100% state of charge most of the day, right? And it's fairly warm. It's not super hot actually at the moment, but it's fairly warm. 100% state of charge, high temperature. That's kind of not great from a degradation point of view. Simple fix would be allow me to adjust the upper SSE down from 100% to 80% or 70% in the summer. So I'm not sitting at this high voltage. The other mechanism is the winter mechanism, if you like, which goes to your question about leaving it in the cellar. And that's, if it's really cold, then if I'm charging at high rates, maybe I'm charging the whole battery in an hour or two, then I'm going to make me worry about lithium plating, which I talked about earlier. I think I'm a little bit concerned about that as well because my battery is actually outside, it's in a shed outside. So it has no insulation heating or cooling. I did that because of lack of space and also slight concern about safety, although I think it's pretty safe. Mainly it's lack of space. I think that if the temperature is dropping below five degrees C, I would worry about charging it really fast. However, most of these systems are not charging at super high rates, right? So my battery is 10 kilowatt hours. Maximum charging rate is like 3 kilowatts. That's not terrible, right? I guess we could dial down the charging rate in winter. So those are my two thoughts. Different problems in summer and winter. The summer fixes to change the upper SSE limit. The winter fixes to dial down the charging rate so that it's very modest. The cool thing about that is both of those things can be implemented with software changes if the company is willing. Super interesting to hear. I wonder how you deal with the gas detector which often go together with a home storage system. However, we don't have time for this. Let's skip to Dan. He asks our inner cities in the Netherlands and Amsterdam, he said, are getting warmer and warmer, especially in summer. Professor, how we mentioned we want lower temperatures for less EV battery degradation. Prior to a purchase, everyone asks where to charge but shouldn't we rather be interested in a shadow parking lot? Yeah, so the question here is going back to what I just said. In the summer, when it's high temperature, are we degrading the battery with this kind of SEI side reaction that I talked about previously in the podcast? And I think it's a good question. I'm not totally sure that having, I mean, yeah, okay. Having the battery under a sunsheel or in a garage would help but I'm not totally sure it's the end of the world because in most of these cars, the batteries at the bottom of the car, right? So in a sense, the rest of the car is kind of shading it anyway. And the whole thing has a reasonably large thermal mass. So I think it's going to take half a day or a day for temperatures to kind of really equilibrate. So I'm personally not too worried about that issue. It's a good question, though. The next question comes from Matthias. Sounds like a German name. He is asking, I think I deep discharged my bike battery. How may I find out? And if I did, can I reboot the chemistry somehow? Honestly, I have no clue what he means. But maybe you can kind of decode that. I'll show you, actually, I've got a deep discharge battery here. Hang on, one second. Which I probably get rid of. So this is a very, very old phone, right? It's the wooden box. And what I left this phone in my desk drawer at home for years and look what happened. You can kind of see that the case has kind of been pushed off by the cell. So what's happened is there's some kind of pouch cell inside. If you're not watching this on YouTube, then basically what it is is the front screen of the phone has been pushed off because the battery is swelled up inside. Which is a bit worrying, and I should probably get rid of it. But what's happened there is some decomposition of the battery, which has generated some gas and caused the whole thing to swell. That gas is probably comprised of different things. There's probably quite a lot of carbon dioxide in there. But yeah, that's what can happen when you deep discharge or leave the battery a very low voltage. And the voltage starts to go below the voltage where we essentially start to dissolve the current collectors. So we have aluminum and copper in there. And if I remember correctly, I think it's the copper that dissolves at a lower voltage. And so that starts to cause bits of copper to float around in the electrolyte. And then it can start to cause degradation of the electrolyte. And we can have what we saw here. I don't think there is a way to reboot the battery. Sorry. I think if this has happened, certainly this one here, I just need to get rid of it. In lead acid batteries, the old-fashioned ones, there are some remedial things that you can do to improve an old battery. But for lithium ion, not really. You could take the whole thing apart, clean the electrodes, put new electrolyte in, and then it would probably be OK. But that's not something anyone's going to do. So I guess that is also a no to the following question. Digital wall is, that's the user's name in YouTube, asks, is there any chance to recover a battery capacity with some fancy discharge cycles? Not really, not really. Although I will say sometimes we see in lab tests of degradation that the capacity appears to drop and then go back up again. But I think this is more to do with very, very long term effects relating to the equilibration of concentration gradients in the cell. And if you're interested in that, there's some very nice work that's been done on that, but I can send you the papers. But I think if it's genuinely the real capacity of the battery is fully equilibrated, then it's going to change irreversibly in my opinion. You might want to tune in the previous podcast with Professor Howie, he talked on this for quite a while. So the link is found in the show notes. We have to go on to the next question. By Oliver Adams via email, he asked my cell phone always dies quickly when it hits 15% state of charge. It then takes just a few minutes to entirely switch off. Why is this? Did they fail in setting up the SOC algorithm where has this a specific chemical reason? Yeah, great question. Very close to my heart. In fact, I was talking to an industry partner about this just yesterday. So let me first explain how the SOC algorithm is working. And then I'll try and explain why it's breaking. So as I mentioned earlier, the open circuit voltage of the battery is really, really important for establishing the battery's capacity and also correcting the SOC estimation. And what this is, if you imagine a graph of voltage versus charge, as I discharge and charge the battery, I go up and down a curve. And ignoring my point earlier about hysteresis, which is like a thing that complicates this, but let's pretend it doesn't. Basically, this gives me a relationship between voltage at open circuit when the battery is relaxed and how much charge is in the battery. It just goes up and down, right? And so when I want to estimate SOC, what I'm doing is I'm taking a current sensor, and I'm adding up all the current that goes to the current sensor. And I'm just basically counting that current. And then by using this voltage versus charge curve, I can kind of correct for errors in the current sensor, OK? That's fine at the beginning of life when I've got a perfectly nicely measured voltage versus charge curve. But as the battery gradually ages, then this curve is changing because it's kind of shrinking on the charge axis, right? Because the capacity of the battery is essentially like how much charge can I get in or out between two voltage limits. And that's fine. I can correct for that by just changing the, if you like, the factor in front of my Coulomb counting. But what happens, I think, in very late life, is that not only am I shrinking this curve, the shape of the curve is actually changing. And so that's when I start to have problems. Because if the shape of the curve is changing, but I think it's something else, then it's going to think the SSE is 15% when it's actually 2%. And that's why what you're seeing is the phone is telling you is 15% but it's actually then dying like a few seconds later. And the way to change that is to go in and actually adjust not just the shrinking of this curve, but to actually adjust the shape of it as the battery gets old. So it is the algorithm, but it's probably not their fault since it's very hard to measure, right? Yeah, it's not really the algorithm. It's the data behind the algorithm, the lookup table for the OCB curve. You're just shrinking the whole thing. You're just changing like one parameter when actually you need to change the entire lookup table if I can put it like that. Understood. Okay, let's move on. Next question comes from Lorenzo. He texts too bad. You did not talk about charging algorithms for electric vehicles in the last podcast. Who decides how fast I charge the charging station or my battery management system inside my electric vehicle? Can this be manipulated? I am asking for a friend. Yeah, I think we'd all like to charge our cars more quickly. Okay, so it's a great question. And I think it's a combination of both, but let me try and explain it's mainly their vehicle, right? But the charging station sets the upper limit because in the charging station there's going to be some power electronics which converts from the grid power to the voltage that I need for the car. And that is going to have a maximum power, right? So that's kind of obvious, but I just thought I'd say that. Then beyond that, it's up to the car to decide how quickly should I go. And this happens in different ways. So if I'm charging at relatively low power from seven kilowatts, for example, then the charger would typically be on board the car and I just plug into the mains and then the thing is fully controlled by the car and the car converts it from AC to DC and charges the battery. If I'm charging from a high power charger maybe 50 kilowatts above, then you'll notice with something like CCS, if you know what that means, that there's an unplug and extra part on the EV which has two wires which go in and this is directly connecting plus and minus to the battery and DC bus. And then the charger to do that high power stuff is not on the car anymore. It's now in the charging station. And so what happens is there's some other wires in that connector which basically is the car able, the car can then tell the charger and give me this much power, right? Give me full power, give me 50 kilowatts. But then as the batteries get nearer and nearer to full charge, you pull back on the power. You know, it's a bit like if I'm parking a train in the station, I don't want to go in at full speed and then just instantly break. So I slow down as I get nearer to the buffers. And so the car is telling the charger, right? I'm at 85% go down to 10 kilowatts now and so on. And I guess there's another question behind your question which is how do you decide what profile to use, right? How long should I spend at full charge and then how do I die at the end? And that's a really interesting question and there's, you know, those companies research on fast charging is basically how do you answer that question? And let me just say this, the standard way in the lab is what's called CCCV, so constant current and then constant voltage. And that means I just push in like a certain amount of amps for a certain amount of time until I hit a voltage limit. And then I keep it at a voltage limit while turning the current down. And in the constant voltage part, what's happening there is that the concentration gradients and the battery are kind of relaxing. So I'm just squeezing in the last bit as I kind of relax everything. And you can show that that maybe is not the best for lifetime and there are some kind of other ways of doing this. They're not really going to be that different. In the end, you're always going to go like more power at the beginning, less power at the end. But if you can do this in such a way that instead of just blindly following CCCV, I'm actually obeying some physical constraint. For example, it's a temperature limit or the graphite voltage with respect to lithium where I know there's lithium plating potential. So if I can like follow that plating potential with some buffer, then I can probably essentially go beyond what I think and get more into the battery more quickly in such a way that I don't degrade it. But to do this really well, I think you have to kind of really monitor what's going on quite carefully and be ready to adjust mid-flight. But there's tons of companies working on this. It's quite an interesting area. Thank you very much for explaining that CCCV model. I think everyone knows that from his or her phone when it says it's now 80% charged, rest is kind of getting slowly, right? Now, next question comes from Gerhard. He asks, if battery health and a long lifetime was my first and single aim, it is, or is it smart to ever fast charge my electric vehicle? I have read so many different opinions on this. How large is the impact of fast charging on battery state of health? Yeah, is it smart to ever fast charge my vehicle? That's a great question. I think that we can't get away from the fact that increasing the rate of the current is more detrimental, right? So I guess there's no such thing as a free lunch, right? If I want to get lots of energy into thing quickly, it's probably going to degrade a little bit more. But if you think about how much fast charging am I actually doing? Unless I'm a taxi driver and I'm fast charging three times a day, it's probably going to be not very often. I mean, for me, I'm slow charging at home most of the time and then I'm fast charging like once every few months when I go on a long trip. And I think in that sense, it's kind of in the noise, right? We can kind of say, okay, yeah, fast charging is going to be worse than if I just charge at home, but we're talking about a very minute difference. So that's my first thing I would say. I think that I would also say that it depends on the design of the system and you would hope that a good high quality manufacturer has thought about issues like, how do I heat and cool the battery to make sure that when I'm fast charging, the temperature is uniform and stable. But there's always going to be a trade off. So you can't have everything. And I think most manufacturers would be willing to go for high energy density and low cost and maybe sacrifice a bit of life on the expectation that most users are not fast charging all the time. If I could put it like that. If we're talking about something like electric trucks, it's a totally different story, right? Or industrial users where the thing is going to be charged and discharged a longer regularly. I just have a different design perspective, I think. Probably something for the next podcast. There's two last questions. One from Tim Duncan, he asks, Professor Howie, how likely is the development of so-called forever batteries? I have read something about this, now me Patrick, from Jeff Don, who was somehow expecting forever batteries that last, I don't know, one or two million miles in electric vehicles. But Tim now asks you how likely is the development of these batteries ever? So what we're talking about here is basically taking a slightly different perspective to a lot of battery research, where the focus is a lot on my energy density and getting the next breakthrough. I think what Jeff Don has done is think, okay, this is a precious resource I wanted to last for as long as possible. So how do I do that and what are the physical reasons why batteries age and how do I extend the life? And I think that's a really important way of thinking about the problem, okay? And he identifies a few different ways of making battery life much longer. I won't go through them all, but some very interesting ideas around using single crystal materials, which therefore when they expand and contract, you don't get cracks propagating in the same way, which is something that we see in other industries, right? So if you look at jet engines, the turbine blades on jet engines are very carefully designed, almost as single crystals, I think, I remember. So that you don't get kind of failure from cracks propagating. So the kind of mechanical design of the electrodes is one aspect of this, super interesting. But of course, again, I think I said it already, there's no free lunch, these things cost more, right? There's always some trade-off, right? And generally in batteries, you're talking about a trade-off between lifetime energy density and cost. And there are very good physical reasons for that. To mention, another researcher, Dan Steingart, Columbia University, has written a very nice post about this, exact issue. I think we are kidding ourselves if we believe that you could have all three together. You've got to choose. Do I want like cost and energy density or cost and lifetime or whatever? Yeah, so I think it's a great idea in principle, I think in practice, it's going to come down to different applications and what they need. Last question for this podcast comes from Andres Blanco via YouTube. It would be nice to hear more about standalone battery grid-connected applications. The one in a million-pound question is lifetime versus projected revenue. Do you agree on that question? Or do you get many requests from the industry regarding that description lifetime versus projected revenue? Is that an issue? Yeah, so I think it is an issue and it's a good question. And if you're building grid batteries in the private sector, well, actually, and in the public sector, you are going to care about what is my return on investment. I mean, how much do I have to pay upfront for this thing? How long is it going to last? Is it going to deliver the service that I need? So it is an important question. And there are trade-offs again there. Typically, we see trade-offs between different types of services that you can provide for the grid. So one service is just energy trading. So a bit like in my home battery, I want to charge when the price is low or I've got surplus solar power and discharge when I need it or the price is high. The same thing applies on a large scale with a megawatt or tens of megawatt grid battery. So energy trading is one application, but there's a whole bunch of other applications as well, associated with keeping the power grid stable. And it varies from country to country. Here in the UK, we have something called frequency response, which, because I guess it's an island, the grid frequency maybe varies a bit more than it does in Europe. And therefore, you can inject power to keep the frequency stable. And you get paid to do this. And you get paid to provide it in a way that's not directly connected with how much energy is going in and out. And so I think to really assess the question, like, what is my return on investment for a grid battery? You have to consider all of the different services that you could offer and what each of the incomes is for each of those services. And then consider that against the cost of the battery and the lifetime of the project. It's a great question and something that I've thought about a lot over the last five years or so. Because it goes back to the very first question about the value of modeling, right? Because to answer that question, I actually need to bring together some kind of battery lifetime model, some kind of information about prices and markets, and probably some forecasts and stuff, to say, well, I'm not going to have perfect information about tomorrow, so how accurate is my forecast? And there's a whole bunch of decisions about risk and trading. And so I think that area about mixing batteries, forecasting machine learning, trading and energy and the commercial side is super exciting right now. And there's lots of companies doing work in that space. Thank you so much for this insight, Professor Howie. I would have loved to talk about the new electricity tariff zones that are probably going to be introduced to Britain as well as Germany. We're talking about this every day, basically nowadays, then batteries really come into the focus, but we're lacking of time, so thank you very much for your time and your expertise, Professor Howie. Dear listeners, if you're interested in dropping your question, then please do so and comment below this podcast or send us an email, that's helloadbatterygeneration.com. Thank you very much for listening and see you next time. Bye-bye. Thanks for having me again. This podcast is also supported by the Katzhu Institute of Technology, ULM University, the German Aerospace Center, and the Center for Solar Energy and Hydrogen Research, Baden-Würdenback.