Battery health: What is the ideal CAC ?

[djp]

In general I agree CAC is just an estimate and it can overshoot the true capacity of the pack, but it is the best number we have. The only time that CAC gets accurately measured is when the car is driven from a full charge down to close to zero. If you're only using the middle of the range the CAC slowly drifts upward and overstates the true capacity.

The range numbers supplied by the car are apparently calculated from the same information used to calculate the CAC, as the two move in the same direction. We do know the relationship between the two is not just a simple multiplier. My guess there is that how you charge, and from what initial charge, has more of an effect on reported Ideal Range than on CAC.

As far as I've seen the displayed range is a simple multiplier of CAC and SOC LIM. The pack reaches a higher SOC LIM when charging on 120V. I usually see a Standard charge SOC LIM of 83% on 240V and 84% on 120V, which accounts for the difference in displayed range.

Not doing a full Standard Charge is not worth the effort unless you're not going to be using your Roadster for months (plural) and/or is being stored in very high ambient temperatures.

Disagree on this one. The NREL study has a graph of degradation rates at various temperatures and SOCs. I agree with Martin that the biggest step is from 100% to 80%, but there are still meaningful benefits in maintaining a lower average SOC. For example, at 25C the graph shows it would take 11 years to lose 20% capacity at 80% SOC and 17.5 years at 60% SOC. I'll take the extra 6.5 years.

View attachment 23281

 

[gregd]

Also depth of discharge is a HUGE factor on how many cycles you can get out of your pack, so 2% is really low. Look at the table in this link and compare the number of cycles when a pack is discharged to 25% vs. 10% over time:

How to Prolong Lithium-based Batteries - Battery University

From the above paper is a table that says:

50% Depth of Discharge gets 1,200 - 1,500 cycles
25% Depth of Discharge gets 2,000 - 2,500 cycles

If depth of discharge is a direct indicator of miles driven, 50% depth is twice the miles driven as 25%, right? So, one should get somewhat MORE miles by doing a 50% depth and then charging, vs doing two 25% depths with a charge for each. Two times 1,200 is more than 2,000.

So, without hitting the very low SOC levels, it's better to do fewer, longer cycles than it is to plug the car back in after every short errand. Let it get down to 50% or so, then charge it. (They didn't publish a 75% Depth number so I can't tell where things begin to get worse...)

Is this correct?

 

[smorgasbord]

For example, at 25C the graph shows it would take 11 years to lose 20% capacity at 80% SOC and 17.5 years at 60% SOC.

Maybe you didn't attach the right graph? The ones you attached only show 5, 8, and 10 years in what is a non-linear time-scale that's not marked beyond 10. Or, maybe there's an equation the uses the "Capacity fade rate" that's in the article, not the graphs?

There are also the usual caveats to this:
• This wasn't done on the exact chemistry that's in Roadster.
• This was done with the battery not being used, simply stored. I agree that for long term storage, lowering the state of charge is a reasonable thing, just not for an overnight charge when the car is to be driven within two months.
• This was obviously done on some sort of accelerated/predictive process. That is, they didn't actually take 10 years to generate the data for the 10 year life, so there's some kind of acceleration and/or extrapolation going on.
• We've seen better results that Eberhard predicted, and that's almost entirely from cars charging to full Standard most of the time. I think this also calls into question the accuracy of the predictive models being used, including in this report.

Another data point is a fellow Roadster owner who has a very short commute and so didn't charge his car every night - he has ended up with a worse CAC than other cars of his vintage (and much more mileage). Was he better off keeping the car at lower states of charge than he would have with the car on the charger overnight? Not according to his CAC. Is his battery actually far better than his low CAC value is reporting? I don't know. From what we can tell from CAC/estimated range, his car is worse off for being kept at lower states of charge than it would have been had he full standard charged every night.

So I continue to take a dim view of both relying on CAC as anything other than a +/- 10% battery charge capacity estimate and that not charging your car everynight to keep it at a lower state of charge is worthwhile. Tesla recommends and the data we have shows that plugging in and charging is the best thing for your car's battery. I believe it's better to charge to full Standard every night in every case except where you're not going to drive the car for 2 months, and even then, you have to plug it in at Storage Charge to get a benefit. Keeping the car unplugged at low SOCs is worse than having it plugged in and charged to full Standard.

Your opinions may differ, of course, but I know of no compelling data that contradicts what I've said.

- - - Updated - - -

So, without hitting the very low SOC levels, it's better to do fewer, longer cycles than it is to plug the car back in after every short errand. Let it get down to 50% or so, then charge it. (They didn't publish a 75% Depth number so I can't tell where things begin to get worse...)

Is this correct?

Like I just posted, I know of one owner (perhaps he'll speak up) who did just that and his CAC/estimated range have suffered far worse than other cars of his vintage and with more miles than his.

Now, perhaps his battery is really really good, but his BMS doesn't know it and thinks its much, much worse than it really is. Which would then go to show that CAC is very inaccurate and raise the question as to how we know how good a battery's health really is.

 

[djp]

Maybe you didn't attach the right graph? The ones you attached only show 5, 8, and 10 years in what is a non-linear time-scale that's not marked beyond 10. Or, maybe there's an equation the uses the "Capacity fade rate" that's in the article, not the graphs?

You can calculate any time span by reading the capacity fade rates off the graph and doing some math. The 5, 8 and 10 year lines are there for reference but any other combination of % fade over time can be reverse engineered from the data.

Agreed with the caveats, but the data is still useful. A typical car is sitting in storage 95% of the time, unless you're driving more than an hour a day. I'd rather have my car parked at lower SOC than higher. I generally keep mine between 40-60% and after 5 years my CAC is still 155, even after full cycle road trips. Of course all the usual caveats on being a single data point and the inaccuracy of CAC.

 

[shrink]

I generally keep mine between 40-60% and after 5 years my CAC is still 155, even after full cycle road trips.

How often do you charge to full standard to balance the pack?

 

[djp]

How often do you charge to full standard to balance the pack?

About once every two weeks - I usually top up the pack on Friday nights for the weekend. The balancing cycle wraps up quickly and all bricks are within 0.02V of each other, so balance looks good.

 

[bart513]

Just as an FYI about 1 year ago I had a conversation with JB Straubel about my CAC, ideal and range mode miles and my weekly commute. In summer and winter when having to use the AC and heat I would barely make it ( or not) of course being aware of speed . His response was not to worry about Range Mode charging as much as I needed to in order to reach my destination. This reply isn't to challenge the data collected by my fellow and well respected Roadster colleagues who are far more sophisticated than me, but rather to share information from a high level, intelligent Tesla battery expert.

 

[hcsharp]

From the above paper is a table that says:

50% Depth of Discharge gets 1,200 - 1,500 cycles
25% Depth of Discharge gets 2,000 - 2,500 cycles

If depth of discharge is a direct indicator of miles driven, 50% depth is twice the miles driven as 25%, right? So, one should get somewhat MORE miles by doing a 50% depth and then charging, vs doing two 25% depths with a charge for each. Two times 1,200 is more than 2,000.

So, without hitting the very low SOC levels, it's better to do fewer, longer cycles than it is to plug the car back in after every short errand. Let it get down to 50% or so, then charge it. (They didn't publish a 75% Depth number so I can't tell where things begin to get worse...)

Is this correct?

It's not correct for an EV in my opinion. In addition to smorgasbord's caveats:

• This wasn't done on the exact chemistry that's in Roadster.
• This was done with the battery not being used, simply stored. I agree that for long term storage, lowering the state of charge is a reasonable thing, just not for an overnight charge when the car is to be driven within two months.
• This was obviously done on some sort of accelerated/predictive process. That is, they didn't actually take 10 years to generate the data for the 10 year life, so there's some kind of acceleration and/or extrapolation going on.
• We've seen better results that Eberhard predicted, and that's almost entirely from cars charging to full Standard most of the time. I think this also calls into question the accuracy of the predictive models being used, including in this report.

Another data point is a fellow Roadster owner who has a very short commute and so didn't charge his car every night - he has ended up with a worse CAC than other cars of his vintage (and much more mileage). Was he better off keeping the car at lower states of charge than he would have with the car on the charger overnight? Not according to his CAC.

These studies do not emulate the use of a battery in an EV. Specifically, they don't take into account the fact that temperatures are higher when driving at lower SOCs. In all those studies they were careful to keep temps constant so that it wouldn't corrupt their data. They were testing the effect of depth of discharge, not the effect of temperature. If you are driving 65 mph at 80% SOC your pack will not generate as much heat as it will at the same speed at 60% SOC.

 

[gregd]

It's not correct for an EV in my opinion. In addition to smorgasbord's caveats:

These studies do not emulate the use of a battery in an EV. Specifically, they don't take into account the fact that temperatures are higher when driving at lower SOCs. In all those studies they were careful to keep temps constant so that it wouldn't corrupt their data. They were testing the effect of depth of discharge, not the effect of temperature. If you are driving 65 mph at 80% SOC your pack will not generate as much heat as it will at the same speed at 60% SOC.

Ok, interesting. So as one would expect, there are multiple factors affecting the outcome, and the evidence is that temperature management is more significant than depth of discharge in determining long-term pack life, so long as one doesn't go too deep.

So, now my own dilema seems to be answered... My commute is about 60 e-miles round trip, nearly all of it freeway. I live about 1,000 ft higher in elevation than my workplace, so it takes about 20 e-miles to get to work, and 40 to get home. But I get 4 hrs of free charging at work, and 3 hrs will cover a full round trip. Do I:

A) Continue to be cheap, and only charge at work? That means that the car sits at home at night and on weekends with ~135 miles of ideal range on it (about 60% SOC)? The lower SoC drive is down-hill in the cool of the morning, so not much heat is generated anyway. Why increase the home electric bill?

B) Not bother charging at work, since it's a hassle because the charging ports are a long way from my cubicle, and it's only 20 e-miles worth of charge coming down the hill.

C) Charge at work and at home, even though the at-work charge only needs to replenish ~20 ideal miles worth of juice? That way, the ride up the hill is with a pack that starts at full Standard charge, and a few hours later, the pack can rest easy with a full stomach for the night / weekend.

Sounds like I should be doing C), and pay for the 40 miles worth of charging every night, to keep the car topped off at Standard levels whenever possible, and suck it up and get my exercise walking to the more remote parking lot so the car can start the up-hill ride home with a full battery. Yes?

 

[ecarfan]

In your case I would not bother charging at work. You've only used 20 miles of range to get there. That's trivial. Let the car sit at a lower SOC during the heat of the day (and where you work it is pretty warm for half the year). Do a Standard charge every night at home when it's cooler.
Don't over think it. [emoji3]

 

[spaceballs]

In your case I would not bother charging at work. You've only used 20 miles of range to get there. That's trivial. Let the car sit at a lower SOC during the heat of the day (and where you work it is pretty warm for half the year). Do a Standard charge every night at home when it's cooler.
Don't over think it. [emoji3]

Yeah I agree with ecarfan, save time and wear and tear on your charging connectors and just charge from home.

 

[wiztecy]

I agree, if you're commuting only 20 miles, don't worry on plugging it in. But for me, I'd rather look at the ESS temps and make a judgement call. If its hovering above 33C I'd cool it down. But if you're only going to spend 2 hours at work, then no. If 8+ hours, then yes.

So with the CAC, the reverse could be true. Our CAC could be better than what its reporting right? If its just an estimate. Lastly, all this looking at the CAC when people are buying a used car is phony?

As I mentioned before, yes, you need to not take just data points from the middle/85% of the battery discharge/charge group but most importantly it needs to be pulled up and most of the way down to re-calibrate.

Just my observation from 3 battery packs, I did all that and my CAC is rather close to what it reported in the middle. I've see at most 2 maybe 3 CAC point rise/fall from it after doing the full Range mode, drive to 15% on one key and charge method. Maybe if I was lazy or neglected to balance my pack, but I keep my pack balanced every day with 0's on the grid before my 1st key-turn in the morning. My spread between Vmin and Vave are all within 0.02v. Again this works in my Universe, different Universes may yield a different result. Everybody's driving / charging / balancing behaviors are different. Again, I'll do the calibration after I've hit a ceiling and come back to the CAC discussion.

As for the Range mode charging, I have to disagree with Jerome's answer. That's why Tesla warns you about dong Range mode charges when you select it. I think he's thinking more on the Model-S side of things. Now if you do Range mode charges every so often, then its placed in the mix and averages out across the overall health of the pack. But we do recall that person who tried the PR stunt with Tesla claiming his pack degraded faster than expected and wanted a new one, but then we found out he was doing Range mode charges most of the time. Now in that case, I believe he was on the edge of the std. mode range in the beginning for his commute, then he ran out of capacity, but instead of using a charger to bump up the miles (which would be more healthy for the pack) he chose to do Range mode charges. And we all know where that went who were following that thread.

And in the end, yes EVs battery packs are a little different than the raw Lithium-Ion cell. However, the base / root of the battery pack is the Lithium-Ion cell, and that's what we have mostly out there in terms of datapoints for life cycle, care, health, and capacity over time. We also have datapoints from EVs such as the Leaf and what not to do in terms of degradation. i.e.: 100% charges, 0-10% discharges, no thermal cooling, etc...

 

[supersnoop]

Lastly, all this looking at the CAC when people are buying a used car is phony?

When I see people ask questions about Roadsters for sale, they always want to know "the standard charge miles." Since that can vary based on how the car was charged, obviously the CAC is the more (and most) accurate measure of battery capacity and health. I guess if we're splitting hairs, it's the car's interpretation of the battery's health. And it's the best number to use.

 

[smorgasbord]

When I see people ask questions about Roadsters for sale, they always want to know "the standard charge miles." Since that can vary based on how the car was charged, obviously the CAC is the more (and most) accurate measure of battery capacity and health. I guess if we're splitting hairs, it's the car's interpretation of the battery's health. And it's the best number to use.

Would you pay more for a Roadster with a CAC of 153 but 4000 more miles than a Roadster with a CAC of 146 in similar condition?

(careful, it's a trick question).

- - - Updated - - -

B) Not bother charging at work, since it's a hassle because the charging ports are a long way from my cubicle, and it's only 20 e-miles worth of charge coming down the hill.

This. No need to put it on the charger at work for such few miles and no need to cool the battery.

Actually, if you want to optimize, get an OVMS box and set both work and home locations for COOLDOWN cycles at, say, 26 degrees, but set for actual Charging only at work. That way, you'll charge at the "free" place as much as possible, and use not-free charging solely to cool the battery. And, if you need to charge differently, you can do so from your smartphone.

 

[supersnoop]

Would you pay more for a Roadster with a CAC of 153 but 4000 more miles than a Roadster with a CAC of 146 in similar condition?

(careful, it's a trick question).

Are you sure it's a trick question and not a loaded question?
Depends if the CAC's are accurate (with recent 100%-10% single-key-turn drive)
Depends if the 4,000 mile difference is between 8,000 and 12,000, or if it's 44,000 to 48,000.
Obviously, it will also depend on the accessories included. And maybe the color.

But, for the sake of argument, and so I can hear the "trick," I'll say the higher-CAC car would be worth about $1,500 more to me.

 

[wiztecy]

Show me a Roadster that had a 153 CAC, got re-calibrated, and now shows 146 or below. I know that some owners have taken long Road trips in their Roadster, had sub 148 CAC, seen it drop after a series of Range mode charges on their trip, but then later seen their CAC recover back up, typically not reaching the pre-CAC of the trip but coming in pretty darn close.

If that was true, we'd see owners who based their buy on CAC, and the first range mode charge, the CAC dives and goes to crap. I've never seen it nor heard any owners complaining about their CAC diving after that event. And that's what the majority of educated Roadster owners have been basing their buy upon.

And in terms of a buyer picking up an "out of balanced" pack, from my experience they'll always make out. All the packs whacked like that typically get better (ie: Tesla's Refurb Pack) and the out of balance CAC reading is the worse case scenario for the pack.

And yes I'd pick up a Roadster with higher miles with significant higher CAC. Look at the refub pack case, and in any event yes, someone who takes care of their pack vs some guy who bakes the Roadster in the sun and hot bathes it at night.

 

[hcsharp]

So, now my own dilema seems to be answered... My commute is about 60 e-miles round trip, nearly all of it freeway. I live about 1,000 ft higher in elevation than my workplace, so it takes about 20 e-miles to get to work, and 40 to get home. But I get 4 hrs of free charging at work, and 3 hrs will cover a full round trip. Do I:

A) Continue to be cheap, and only charge at work? That means that the car sits at home at night and on weekends with ~135 miles of ideal range on it (about 60% SOC)? The lower SoC drive is down-hill in the cool of the morning, so not much heat is generated anyway. Why increase the home electric bill?

B) Not bother charging at work, since it's a hassle because the charging ports are a long way from my cubicle, and it's only 20 e-miles worth of charge coming down the hill.

C) Charge at work and at home, even though the at-work charge only needs to replenish ~20 ideal miles worth of juice? That way, the ride up the hill is with a pack that starts at full Standard charge, and a few hours later, the pack can rest easy with a full stomach for the night / weekend.

Sounds like I should be doing C), and pay for the 40 miles worth of charging every night, to keep the car topped off at Standard levels whenever possible, and suck it up and get my exercise walking to the more remote parking lot so the car can start the up-hill ride home with a full battery. Yes?

The best answers to that are quoted below. Basically you don't want to come home with an elevation gain which heats the pack and then let it hot-soak until you go back to work the next day. So charge when you get home at least enough to cool it down.

This. No need to put it on the charger at work for such few miles and no need to cool the battery.

Actually, if you want to optimize, get an OVMS box and set both work and home locations for COOLDOWN cycles at, say, 26 degrees, but set for actual Charging only at work. That way, you'll charge at the "free" place as much as possible, and use not-free charging solely to cool the battery. And, if you need to charge differently, you can do so from your smartphone.

I agree, if you're commuting only 20 miles, don't worry on plugging it in. But for me, I'd rather look at the ESS temps and make a judgement call. If its hovering above 33C I'd cool it down. But if you're only going to spend 2 hours at work, then no. If 8+ hours, then yes.
...

 

[dhrivnak]

A wealth of information is at the Roadster owners battery survey. And if you have not taken it please add to the results. It looks like the CAC starts out at the high 150's and shows a slow decline mostly with usage. Plug In America

 

[frequencydip]

I thought ideal miles are partly based on driver behavior. I have noticed when I drive the car more inefficiently it shows a lower ideal rang for a few days until I drive it more Eco and reset the ideal range based on recent historical performance.

 

[wiztecy]

I thought ideal miles are partly based on driver behavior. I have noticed when I drive the car more inefficiently it shows a lower ideal rang for a few days until I drive it more Eco and reset the ideal range based on recent historical performance.

I see the behavior you metioned with "Estimated" miles, but Ideal miles is directly correlated to your CAC and there's a formula to figure that out. Possibly your pack is out of balance and your CAC is fluctuating?

However if you have 180 ideal miles and you drive with a lead foot, you won't get 180 miles that day! I know you know that.... but that driving behavior won't carry over to the next day's charge cycle and ideal mile estimate.

For my daily driving commute, I'm avg. at 235kWh if I take it easy, 245-255kWh if I drive a little aggressive, and up to 265-275+kWh if I'm really having fun (playing on the back roads). I'm in the 245-255 range typically....