New study finds V2G discharging harmful to EV batteries

Prius Plug-In (Charged EVs)

Vehicle-to-grid (V2G) technology is seen as a potentially valuable resource for balancing electrical grids and facilitating the use of renewable energy. However, a new study by a team at the University of Hawaii suggests that the additional cycling to discharge vehicle batteries to the grid is detrimental to battery performance.

In “Durability and reliability of electric vehicle batteries under electric utility grid operations: Bidirectional charging impact analysis,” published in the Journal of Power Sources, Matthieu Dubarry and colleagues write that using battery packs to discharge power to the grid could shorten their useful lifetimes to less than five years. However, the researchers found that delaying the grid-to-vehicle charge in order to reduce the impact on the power grid had a negligible impact on the cells.

“The potential of using EV battery capacity for grid storage has been the subject of much discussion in recent years, but few studies have tested grid impact on EV battery degradation,” write Dubarry and his team. “Understanding the real impact of bidirectional charging on batteries is essential to weigh their viability as a grid support service.”

“V2G will induce more usage of the cells by discharging to the power grid and having to recharge the battery again for transportation. Controlled [grid-to-vehicle charging], on the other hand, will allow the cells to rest at a different state of charge compared with charging immediately upon arrival, and therefore might induce a different calendar aging. Since battery degradation is path-dependent, any changes in usage can lead to different degradation mechanisms and drastically influence the reliability of the batteries.”

The Hawaii team performed laboratory testing on Panasonic 18650 NCA batteries. They investigated the effects of V2G/G2V combined with different charging schedules (1 or 2 charges a day, immediate or delayed charging) and different charging currents (Level 2 or fast charging). The effect of calendar aging at different temperatures was also investigated in a second set of experiments.

The team concluded that a V2G step twice a day increased battery capacity loss by 75% and resistance by 10%. This step once a day accelerated capacity loss by 33% and resistance by 5%. Forecasts based on the measurement results indicated that V2G implementation would decrease the lifetime of the battery packs to under 5 years.

The team also found that calendar aging influenced the cells little enough that it was beneficial to charge the cells twice a day instead of once. Charging twice per day resulted in 5% less capacity loss and similar resistance increase, compared to once per day.


Source: Journal of Power Sources via Green Car Congress

  • Lance Pickup

    It sounds like this might have been a detailed study, so maybe some of the details are available in the paper itself (not planning on paying $40 to get it myself). This is potentially a serious issue, but without some supporting details, it’s a bit hard to quantify exactly how valid the impact is to a real world scenario (although I would hope they maintained a realistic scenario).

    For example: what power level did they use for V2G discharge? Did they scale that down to reflect the fact that the batteries being discharged would be part of a full pack? and therefore each cell’s contribution to the full discharge power would be less? What kind of depth of discharge did they assume? Reading between the lines, it sounds like they discharged the heck out of these batteries, and if so, is that realistic?

    If this is a potential problem, can it be mitigated by limiting depth of discharge to a certain amount (e.g. 5% of the full battery capacity–if enough cars are part of the pool, this may be very realistic)? And/or limit the discharge power to a very small amount (again, counting on a large pool of vehicles to provide the necessary total power)? And/or limit drawing on a particular vehicle to once per week?

    I have been considering the fact that hooking up your EV to a V2G scheme would make battery warranties that specify a number of MILES essentially meaningless. Instead, I think batteries need to be warrantied to a total amount of energy delivered to account for some of this “stationary” use.

    And if all else fails, maybe the use model we need to go for is a home-based stationary storage with batteries optimized for grid stabilization use, and then just charge the car from that source each day.

    Anyway, thought provoking article. Thanks.

    • nordlyst

      I agree. But I’m more suspicious than you seem to be, because of the very dramatic reduction in battery life which must mean the V2G cycling is highly significant compared to the ordinary usage cycling. This strikes me as quite unrealistic, but what is realistic depends quite heavily on what EV density you assume and whether V2G is the only grid storage mechanism or just a part of the solution. We don’t know any of this, and it’s not easy to say what “the right assumptions” may be.

      But I tend to prefer a fundamental view. So I take a very long term view and try to ask what V2G can potentially achieve if all cars are EVs. My household consumption of electricity is more than twice the energy I use to charge my car, but I’d only need to provide some of it for some of the time. To supply 50% of it for 30% of the time I’d need to put one third more cycles on my battery pack than I do now. With an already somewhat degraded (I’ve lost one bar) 24 kWh LEAF that might be a problem for me, but I have great difficulty imagining it being so on a 60 kWh pack with better chemistry and active temperature management. It is the future we’re talking about, after all, so even 60 kWh may well be much too conservative.

      Don’t get me wrong, it’s not like I think I can work this out correctly and my napkin calculations overrule a study. But not knowing much about the assumptions they made or their methodology and having my own attempts to figure it out lead to the opposite conclusion is a decent basis for saying I don’t understand this and therefore cannot believe it on simply the basis of one study of unknown quality/robustness.

      • Lance Pickup

        I think we’re both basically on the same page.

        My first LEAF lost 4 bars (1250 miles over the 60K mile limit 🙁 ) so I too am sensitive to capacity loss. And during the last 7-8K miles before 60K I VERY closely monitored my capacity loss on a daily basis to determine what effect various things were having on capacity loss. Things like ambient temperature; temperature when charging; leaving the car fully charged; leaving the car relatively empty; fast charging vs. L2 charging, etc.

        What I found confirmed my suspicions that the original LEAF battery chemistry is basically a loser. By far more capacity was lost simply on hot days and leaving the car parked outside at work. Charging to 100%, even when letting it sit there, actually reduced capacity loss (presumably because the pack used the opportunity to do a deeper balancing). I don’t think that lack of an “active cooling system” is necessarily a culprit, as that normally would only kick in while charging or in use, and I found that significant loss happened simply while the car was sitting in the parking lot at work. Having said that, if an EV were to be used for V2G, even at relatively low power levels, I would certainly advocate for active cooling, particularly since there will be no airflow during discharge like there would normally be if the car was being driven.

        Second, I would certainly expect some reasonable limits on the power rate and depth of discharge. The kind of numbers you are talking to supply a significant portion of your own home’s electrical needs I would hope are completely out of bounds. I am thinking of a scheme where, as you’ve said in other posts, we are talking more about knocking off peaks, and doing so in a rotational fashion so that (a) each EV on the grid only gets tapped every X hours (or even days) and only to a very small percentage of capacity (5-10%-ish?)

        Like you and I both said, there is not enough details in the part of the article quoted to give enough information on what their test methodology was. It sounds to me like they used a pretty deep cycle, which is probably not a realistic scenario once you get a fair amount of stationary storage and EVs on the grid and are aiming mainly to “knock off peaks” or maybe provide some supplemental power the wind drops or clouds move over a solar array somewhere.

  • dogphlap dogphlap

    Dr Jeff Dahn has made it clear that current Li-ion cells have a limited number or charge discharge cycles before they rest in peace. He also has said it matters little whether these cycles are a large number of small events or a lesser number of larger events. Just so long as you stay away from the top and bottom extremes of the cells capacity cell life is a function of Wh supplied and subsequently replenished (of course there are other factors, temperature and age being a couple). So hardly surprising that acting as a peaker unit for the grid will take some toll on the EV battery life expectancy. J.B. Straubel has also come out against the idea of using Tesla EV batteries for vehicle to grid. We know the cells Tesla employs for electric vehicles are optimised for maximise energy density and high discharge rates rather than than maximum cycle life. Panasonic do make cells for the Tesla Power Wall and Power Pack which have the opposite properties and would be more suitable but even then I don’t think Tesla is in favour of propping up the grid with residential Power Walls.
    For me the battery pack in my EV represents a lot of money and if I’m required to decrease its usable life I would hope for a sensible (i.e. large) level of compensation from the electrical utility.

    • nordlyst

      Well, Tesla may have the knowhow, but they also have obvious motive to avoid V2G. At least if they work on the assumption that EVs will replace ICE regardless of whether V2G happens. If “everyone” must buy batteries to install in their home in addition to the ones in their cars, the market is much bigger.

      I’m not really convinced either way yet. My average electricity consumption is almost exactly 1 kW (9000 kWh/year => 1027.4 W). A single EV with a 60 kWh pack could provide 50% of my electricity 30% of the time – it’s smoothing out the peaks we’re trying to do here! – adding 3.6 kWh or 0.06 charging cycles per day, or 1.8 cycles per month.

      Of course there’s losses and my numbers can be quibbled about. But equally the average battery pack may well be 75 or even 100 kWh before V2G could be deployed in any real scale. And 50% for 30% of the time may be much more than is required to be useful. (It could also be less, although I find that hard to imagine.) So to me the battery life concern isn’t very convincing.

      Which parameter values are right depends on what assumptions you make about EV density, how many cars are plugged in at the time the grid needs them, what capacity the packs have, how much energy they should supply and how that is distributed over time, and probably several other factors I as a layman don’t realise are relevant. This means it’s difficult to work out a robust conclusion, but it also means you can easily choose parameters that lead to a result that shows V2G either to be idiocy or a complete no-brainer. When I attempt to reason about it based on my personal numbers the estimates seem to demonstrate that it would be very easy to have V2G – the extra charging is so insignificant as to be easily compensated with a slightly larger pack, which is much cheaper than having an extra powerwall-like pack in my home AND gives me additional range for the first half or more of the cars life. But obviously my numbers may be totally unrepresentative, and they implicitly assume there’s one EV or household. So my point isn’t to say V2G works, just that this study, or Tesla speaking against it, doesn’t really settle the issue. You may know more than me, but I at least can’t conclude anything based on this.

      • dogphlap dogphlap

        I see from the length and number of comments you have contributed to this thread that you have a strong interest in V2G and have given it some real thought. Thank you for that. I don’t think I can add anymore but the grid is highly regulated here (and for good reason) so even if I wanted to I could not do any practical trials so I’ll just forget about it for now other than to observe that I don’t recall ever seeing such a volume of comments for an article in Charged EVs before so this subject is of more interest to folks than I had realised.

  • jstack6

    They should compare their results with the Univ of Delaware that has been doing V2G for many years.


  • Vincent Wolf

    Owning two Leafs with a total of 54 kW of storage I wish I could hook them up to my home solar to moderate electrical usage at night. But so far Nissan has refused to import their Leaf-to-Home systems to America. Why is it okay in Japan but not in America? Japan makes them for their own citizens just not for Americans.

  • nordlyst

    Be skeptical. The abstract doesn’t provide enough information to evaluate the study even superficially, and says nothing about who funded it; and the full text is paywalled.

    Clearly putting more cycles on a battery reduces its capacity. But the parameters used dramatically modify the impact. So the first step in such a study would be to explain and justify the choice of parameters. For example, when they say “two steps a day”, what exactly does that mean, and is that realistically what is needed in order for V2G to be really useful? I can’t tell. But if it means two full cycles extra per day, and those are accomplished by fully discharging at a significant rate, and I sort of suspect that it does because of the huge decline in battery life, then it starts to look a bit suspicious to my mind.

    EVs going forward seem likely to have 60 kWh or more on board. I live in an apartment and use 9000 kWh per year of electricity (includes heating my place, excludes charging my car). That’s an average of very nearly 1 kW consumption. For V2G to be useful it doesn’t need to supply ALL of the electricity in the grid at any point in time – even a ten percent contribution will greatly help to load balance the grid.

    I’m no expert on this stuff, but I do think ChargedEVs should be careful publishing studies, positive or negative, without having independent experts examine them and weigh in on them. It could well be that this study is perfectly good and unbiased, but it seems very possible that a bad study in which parameters are deliberately chosen to get a result that undermines a very important “selling point” (in this case to regulators, not consumers) could get exactly the same coverage.

    I do realise it takes more resources to do this kind of quality control, but it’s not like I’m saying you should be able to vouch for the study. Just try to get someone qualified to read and comment on the study so readers get some idea about possible objections to it. Studies sometimes have extremely robust evidence for their findings, but can also be narrowly true and simultaneously completely misleading. Focusing only on their findings and ignoring whether they are robust makes it very hard for readers to get any real insight.

    • Gillacey

      This study has been done by a very well known academic in probably the most prestigious peer-reviewed journal for this topic. Charged EV have every right to highlight it and I’m glad they have.
      you don’t seem to understand how academic research works and you haven’t read the article (which is freely available to anyone with a address), so please be assured that it is valid and I’m sure ChargedEV and Prof Dubarry welcome informed discussion of their findings

      • nordlyst

        Then name the journal and post the link. It’s not difficult to just say stuff. I can do it too, and like you provide zero justification, only the bars claims themselves: “This study has been done by a known fraudster and was published in a fraudulent pay-to-publish repository.”

        You provided nothing to give me any reason not to remain skeptical. In fact, that you think you did demonstrates how naively you think! “Trust me, you should just trust them” is what your comment boils down to.

        If you’re right, be more convincing! I’m very much interested in reading the study, but I can’t find it and I don’t have “a address”.

        You’re right that my understanding of how research works is limited, but you don’t seem to know the basics of how skepticism works!

        • Gillacey

          nordlyst, I am not asking you to trust me, I’m telling you that you can trust the journal that published it. The journal is the Journal of Power Sources. I’m afraid I can’t send you a copy of the article because it is only available for a fee or to a university, which pays for a subscription so that their academics can have access to the latest research in their area,. The idea is that someone does some research, they write a paper and submit it to a journal, all journals are rated in two ways; they belong to a publishing house with a good reputation, or by the number of citations they have over a period of time. known as the Impact Factor. You can look up the journal and see its impact factor in any search engine. The journal may reject the paper outright, or send it for review to other academics who have published in this field. the paper is then reviewed, sent back for edits to the author, reviewed again and hopefully eventually published. all academics need to publish to progress in their careers and the publishing process is carefully regulated, as I hope you can see. So please, don’t say “that doesn’t seem right” without having some evidence that has been subjected to at least the same peer reviewed scrutiny as the original.

  • Doug

    Dr. Dubarry and his colleagues are well respected, have conducted years of research on battery degradation, and have impressive lab facilities that are specifically dedicated to this type of research. I’ve read the paper and think it is a very good, they did a good job of modeling drive/duty cycles and other important elements. Those that have access to Elsevier can find the full investigation at: