Battery health: What is the ideal CAC ?

[smorgasbord]

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.

They're the same Roadster, as reported earlier in this thread: CAC 146 in May and CAC 153 later, in September. Figuring an average of 1K miles/month. So, the earlier, less used version of itself is the better buy.

- - - Updated - - -

Show me a Roadster that had a 153 CAC, got re-calibrated, and now shows 146 or below.

Hey, Wiztecy, I'm talking about your Roadster. At least one person here would pay you $1.5K more in Sept than they would have in May for your car, even tho you added thousands (I'm guessing) on the odometer.

Have I made my point?

 

[wiztecy]

Have I made my point?

No.

I understand that CAC can jump up and down with Roadsters. And I'd be happy to buy a Roadster that had a dip and was able to recover that CAC any day. I always can get out exactly the ideal miles (cac@153&181 ideal) to match the real miles my Roadster if I drive 65mph. Same when cac@146&174 ideal. So with that I'm confident with what my CAC reports is very darn close to the actual and there's no smoke and mirrors.

So I'm pulling these magical miles out of thin air right!? Honestly, where I'm curious, is that "Unknown" zone. The lower 20% SOC, and how the BMS/Roadster handles that with the CAC going down and up like this.

Truth is that the cells in the ESS need to be Range mode charged and then driven down to a lower SOC, then the BMS will get all the information it needs to recalculate and provide the user with the most accurate CAC. I'd like to try doing it with a std. mode charge and do a 30% SOC to see what the numbers reflect. And with my theory / belief, they won't be off too much but we'll see when that time comes.

 

[smorgasbord]

And I'd be happy to buy a Roadster that had a dip and was able to recover that CAC any day.

Of course, you could be buying a Roadster whose CAC has climbed higher than the battery chemistry can really support. So, paying more for 7 additional Amp Hours of capacity isn't worthwhile unless you know the recent driving history of the car and whether the CAC is at a dip. Sellers of Roadsters should spend a few months to convince their BMS to report higher CAC numbers - we've seen that buyers might be willing to pay thousands of dollars more for an older car with more miles!

I always can get out exactly the ideal miles (cac@153&181 ideal) to match the real miles my Roadster if I drive 65mph. Same when cac@146&174 ideal. So with that I'm confident with what my CAC reports is very darn close to the actual and there's no smoke and mirrors.

You're not understanding the mechanisms at play here. Let's go back to basics:
1) What is an "Ideal Mile" really? Well, it's not a distance. It's a certain amount of energy, around 215 watt-hours (assuming a full Range Charge stores 53kWh in a new battery that's rated at 245 Ideal Miles).
2) That you can drive a mile and use exactly 215 watt-hours says nothing about how many watt-hours your battery can hold.
3) That the car's BMS ticks off every 215 watt-hours used says very little about how many watt-hours are still left in the battery.

Therefore, the accuracy of the CAC calculation is not bolstered by being able to drive a mile and consume only 215 watt-hours.

Truth is that the cells in the ESS need to be Range mode charged and then driven down to a lower SOC, then the BMS will get all the information it needs to recalculate and provide the user with the most accurate CAC. I'd like to try doing it with a std. mode charge and do a 30% SOC to see what the numbers reflect. And with my theory / belief, they won't be off too much but we'll see when that time comes.

Does anyone know how the BMS actually calculates CAC? I've seen lots of black magic things, like one-key driving, range charging, slow charging, etc. but haven't seen any rational explanations as to why those are helpful/necessary.

 

[wiztecy]

So what test do you want me to do? I'd be happy to go head to head with your CAC and do a range test on a route that's relatively flat/straight, set the cruise control to 60/65mph and see where our mileage ends. We know each other and live close, so that's do-able.

And what makes you confident my CAC dip is cell degradation? And not Tesla who possibly found a bad datapoint or possibly even a bug that caused it to dive, and then some action that was done to bring it back? With this pack I've done at most 3 Range mode charges and always performed cooldowns and never hot-bathed the pack. I don't see any of my habits adversely affecting the pack in any unusual way to account for the dip. I definitely am not contributing habits that typically build up resistance on the cathodes and anodes from build-up other than the 30k miles.

Anyways, I'm using the CAC as "my" gauge on pack life. Either way, I'm happy with what I have. And no-matter if my CAC is high or low, as long as my Roadster can get me to where I want, then my Roadster & I have accomplished all that I need which always brings a smile to my face, and that's what counts. And if I follow what's been found to be healthy for Lithium Ion batteries for longevity and practice and see promising result, I'll keep doing it.

And honestly its the brick's voltage we need to look at, a data point needs to be taken when a 100% charge is done and also when a low discharge is done. That will show the weak brick for that it will charge faster & discharge faster. A metric will also be recorded for the voltages of that/those bricks as well as for all the bricks in the ESS from the ahr.log file. And that will show a better snapshot of the pack's health, in particular the weakest link or brick.

But also remember that Refurb packs usually start out in the 145 CAC and climb. So that's not the walking dead happening here. I will say that most of the gain does come from the pack having time to balance, but still after its balanced the CAC climbs in most cases if its a "healthy" pack.

 

[bart513]

Once again may I complement the Roadster community? Another example of Roadster owners disagreeing on a specific topic, and discussing it in an intelligent manner without the need to insult one another!! Of course I love Wiztecy's idea of doing a head to head trial. Wish I lived closer than NY LOL :wink:

 

[wiztecy]

Ok, so I talked to some people @ work who're hardware engineers and have worked with Lithium Ion batteries. I discussed what we put on the table with this thread, and thanks smorgasbord for keeping it going and helping us understand what's really going on here.

So it is correct, once a cell/battery pack has degraded chemistry-wise it *cannot* be reversed. Secondly, the CAC is *NOT* an accurate way of judging a pack's health as smorgasbord has argued appears to be correct. Reason being, Lithium Ion cells don't bounce up and down in health. They gradually go down in capacity for an ideal cell, or dive tremendously for a sick one. But they will never come back up. Recovering a cell from a very low voltage is a different story, we won't discuss that one here. My friend also said what Tesla is doing is "calculated", so their doing some funky business behind the scenes. Same with rotating out the lowest bricks and he found that to be odd and does not know why they do that. He also indicated there can also be "bugs" in their algorithm, so this diving and coming back up very much well could be that particular scenario. That is not "real" or useful data that has any meaning.

Ok, so how to we gauge "true" health of our packs. That was the question I asked next.

He said the full SOC charge, recording the voltages of the bricks and finding out which bricks reach the highest voltage first is one data point. Lets call this (a). The second, call it (b) is the close to full discharge, same thing, record the first brick(s) that discharge their voltages first. So the once that reach the highest voltage first upon a full charge are identified as "weak" bricks/cells. And for the discharge test, the first ones that drop and have the lowest voltages are also the weakest. There should be an intersection / relationship between (a) & (b).

Another test to identify the health or weakest brick/cell is to monitor the voltages while the battery is under load. Meaning floor the accelerator and in real time or somehow log the individual brick's voltage, the bricks that drop the fastest under load are the weakest bricks. I'm curious if we can pull this data from "diags".

He also state that "Ah" or amp hour capacity is a more accurate way if measuring the health of the pack/bricks.

Again, who knows what and why Tesla is doing the CAC and what ropes the "Oz" behind the curtain is pulling but it does appear the CAC might not be the best way after discussing this to fully gauge the Roadster's ESS health.

By the way, I'm still down for the CAC range test and if others want to join this could turn out to be educational as well as really fun.

 

[ecarfan]

?..the CAC is *NOT* an accurate way of judging a pack's health as smorgasbord has argued appears to be correct... He also state that "Ah" or amp hour capacity is a more accurate way if measuring the health of the pack/bricks.
Again, who knows what and why Tesla is doing the CAC...but it does appear the CAC might not be the best way after discussing this to fully gauge the Roadster's ESS health.

How is the "Ah or amp hour capacity" different from the "CAC" (Calculated Amp-hour Capacity)? I'm confused.

 

[shrink]

How is the "Ah or amp hour capacity" different from the "CAC" (Calculated Amp-hour Capacity)? I'm confused.

I was going to ask the same thing.

Per Tom's paper:

Calculated Amp-hour Capacity
Another common way to measure battery capacity is amp-hours. An amp-hour is theamount of charge required to deliver a 1-amp current for one hour.

http://www.pluginamerica.org/surveys/batteries/tesla-roadster/PIA-Roadster-Battery-Study.pdf

 

[wiztecy]

My best guess is:

Amp-hour (Ah) is how much capacity a battery has when discharged over a given period, say 20 hours, at a constant discharge rate.

CAC or Calculated Amp-hour Capacity relates more to Peukert's law, which reflects more of the EV world, where the battery is measured and discharged in a shorter time, say 1 to 3 hours vs. 20 hours, at a higher discharge rate as well as being given a wider range of current variations. The constant for Peukert's law also takes into account and can be adjusted to take into account aging of the battery. So CAC has more variables going into it and its rather dynamic and also part of the reason why I believe we see the CAC jump up and down with our Roadsters.

https://en.wikipedia.org/wiki/Peukert's_law

 

[markwj]

We know what Tesla has available in the roadster. It is clearly shown on the CAN bus, and displayed on the VDS 'ESS SOC' diagnostic screen. Tom (and others) have also found these in the vehicle log files.

Not sure if I should post the screen shot here, as I guess it is Tesla proprietary information (and we all know they didn't like the Model S diag screens being posted publicly). So, I'll err on the side of caution and describe it.

At the top of the screen is the calculated stuff (SOC, CAC, AR), then the instantaneous stuff (#Ah, #kWh, EVC, kWhR, RR, dRMS, dAI, dWhM), then a historical log for each day (date, LRV, duration, SOC, #Ah).

Anyone with access to a roadster can put in the not-so-super-secret code and see this for themselves.

It is pretty clear that Tesla is calculating the SOC, CAC, AR, because they can't directly measure those. You can't look at a battery and say "what's your Ah capacity?". You can measure instantaneous voltage, and sum up current over time to know what you've put in / taken out, but you can't measure capacity directly. So, you have to estimate it, averaged over time, based on the behaviour of the battery (voltage, etc) during daily driving charge and discharge cycles. They call that 'calculated' because it sounds better than 'estimated', or 'blind guess'. If you go for a long drive, or do a long charge, that is going to give the algorithm more information and it'll be able to get a better estimate (calculation). If all you do is short discharge/charge cycles at the top of the pack, the estimate is going to be extremely limited and could very well drift off 'true' over time.

It is also pretty clear that CAC is scaled to 160 for a new pack, and that it is related to ideal miles by a conversion factor. I've always suspected that it is just "AR" times a factor to get 160 at new, and that factor is stored in the car's config somewhere. No idea why they chose the magic number 160.

In an ideal world, if you were buying a second hand roadster, you'd drive it for a long distance, over time, and see how the range changed. See what real world range you were getting out of the car, and how the battery held up. Right? Well, that is exactly what CAC is measuring (changes in real world observed range/battery capacity, averaged over time). Variations in CAC are most likely caused by variations in the driving/charging behaviour behind the calculation, not the health of the battery itself.

So, is CAC perfect? Obviously, no. Is it the best we have? Yep.

 

[supersnoop]

Have I made my point?

Can you maybe restate your point? Because I don't think you've proven anything. So, either we disagree, or I have no idea what you're trying to prove.

 

[markwj]

How is the "Ah or amp hour capacity" different from the "CAC" (Calculated Amp-hour Capacity)? I'm confused.

Well, Ah would be a specific measured thing, as described previously. But let's assume it is estimated in some way.

In my opinion, CAC is this estimated Ah value normalised (via a calculated/entered factor, and the current SOC%) to the magic number 160 for a full new pack.

The actual Ah usable remaining in the pack goes up and down, depending on state of charge. The CAC is always for a fully charged pack, irrespective of the current SOC%. At it simplest, 100-((CAC/160)*100) is the estimated percentage degradation in the pack since new.

Also remember that SOC%, ideal range remaining, Ah remaining, are all related via simple factors.

 

[djp]

In my opinion, CAC is this estimated Ah value normalised (via a calculated/entered factor, and the current SOC%) to the magic number 160 for a full new pack.

I don't think it's normalized, the 160 number is Ah per brick.

160Ah * 3.7V nominal * 99 bricks = 58.6kWh

A Range charge is 95%, which gives us the rated capacity of 56kWh.

 

[smorgasbord]

Can you maybe restate your point?

OK, to re-restate the obvious: You were willing to pay $1500 more for the same car, despite it being older and having more miles on the battery, simply because its CAC number was higher by 7 AmpHours. Clearly, CAC is not an accurate number, and its calculation to hundredths is silly. One should not spend more money on cars solely on the basis of reported CAC values.



To give this post some new relevance, if I were buying a used Roadster, I'd want as many of the log files from the car as are available and run the VMSParser on them. That would show me:

A) How long the battery spent at very low or very high SOCs
B) How long the battery spent at very high temperatures, and what the maximum temp was.
C) What the average lifetime temperature of the pack is.
D) How balanced the pack is (brickahmin, brickahave, and which brick).

If the previous owner didn't collect logs, or doesn't want to share them, you'd still be able to pull the latest log and run VMSParser on that, which would still be better than relying on CAC alone.

For instance, today my CAC is 149.69 at 39,378 miles. I have a OK log file history of 5.5 years with 166 days missing (mostly from when I first got the car).

A) My SOC has been below 18% only for about 13 hours, never exceeded 96%, and was above 90% for only 107 hours total.
B) My battery temperature never exceeded 41° C (and was there for only 13 minutes), and was never below 4° C.
C) My battery's lifetime average temperature is 21.08° C.
D) The worst brick has varied a lot over the 5.5 years I've owned the car - there's not just one or two bricks that are bad relative to the others.

The largest difference between brickahmin and brickahave was 4.38, with the average difference being 1.21 and median difference 1.26.
Brick 54 was historically the worst brick most of the time (705 days), but most recently it's been oscillating between bricks 88, 32, 35, 54, and 81. So, no big worries there.

 

[wiztecy]

OK, to re-restate the obvious: You were willing to pay $1500 more for the same car, despite it being older and having more miles on the battery, simply because its CAC number was higher by 7 AmpHours. Clearly, CAC is not an accurate number, and its calculation to hundredths is silly. One should not spend more money on cars solely on the basis of reported CAC values.

I would pay more for the same car being older with more miles if the CAC has climbed for that I have confidence in the Tesla electrical engineering team for that they know and understand what their doing with Lithium Ion battery packs in many respects, including their CAC algorithm computations.

I had a talk with my hardware engineer discussing my dip and rise of the CAC, and we know that Tesla is feeding in variables into this algorithm to best estimate the health of the pack. What we've concluded is that since I've been doing these additional cool-downs at work driving the pack temps down to a target temp of 26C, that Tesla is picking up this charge behavior as well as the pack's idle / sitting temp while in the garage at work, as well as the lower temps when I drive back home. As we all know, temperature is a huge factor in an ESS's health and degradation rate as well as the breakdown in the cell's chemistry. And when this "new" behavior is fed back into Tesla's algorithm, the the CAC gets recalculated and as a result, the CAC improves since the given behavior does not contribute to battery degradation as much as before, such as without the cool-down at work and other good rules of thumb in order to keep the pack in a healthy state.

Lasty, I can also monitor my pack's lowest brick voltage since the Vmax, Vmin, and Vavg of each brick is displayed in real time in Diags. To get your weakest brick, have the appropriate screen up with the V's and punch it. No need to have a high or low SOC for this test. The brick that drops to the lowest voltage is your weakest brick. And I'm happy again to run a test with anyone who would like to charge the comparing cars to the same SOC and compare our voltages with this test. Note there will be some variances in resistance and load here, such as with the motor, PEM, etc. But its a close approximation to compare the CAC between the cars and add another data point for discussion.

 

[smorgasbord]

I would pay more for the same car being older with more miles if the CAC has climbed for that I have confidence in the Tesla electrical engineering team for that they know and understand what their doing with Lithium Ion battery packs in many respects, including their CAC algorithm computations.

I'm screaming inside my head quite loudly.

 

[wiztecy]

I'm screaming inside my head quite loudly.

 

[hcsharp]

I would pay more for the same car being older with more miles if the CAC has climbed for that I have confidence in the Tesla electrical engineering team for that they know and understand what their doing with Lithium Ion battery packs in many respects, including their CAC algorithm computations.
...

Umm... This contradicts what you said in post #46.

All the nice things you do to your pack to preserve its life are great, but they're not bringing your cells to better health than they were a few months ago. All you did was bring the CAC up, not the cell health. You also prevented the cells from degrading as much as they would have without those measures.

- - - Updated - - -

I'm screaming inside my head quite loudly.

LOL.

Maybe somebody will lend you some "mothers' little helper." (I didn't really mean that).:smile:

 

[wiztecy]

Umm... This contradicts what you said in post #46.

All the nice things you do to your pack to preserve its life are great, but they're not bringing your cells to better health than they were a few months ago. All you did was bring the CAC up, not the cell health. You also prevented the cells from degrading as much as they would have without those measures.

Correct, because I have more information from people who have a better understanding on batteries than I and my best guess from post #46. That's part of life if you you have an open mind, to take in new information, respect those who have a higher understanding but asking questions on why things act a certain way given a set of inputs and outputs. At least that's how I try to live day by day.

So to restate what I believe is happening here. The CAC is our and Tesla's best gauge for overall health of the ESS. Tesla's proprietary algo takes a set of given inputs or behaviors and conditions and crunches this data to give the best projection over time on the health of the ESS. This number can go up and down, but will settle in on a number that is the best guess on the health of the pack. I'm not saying that when this number dives low, that that's degradation or not. What I can say is that it can be a trend that's pointing to degradation down the road if the inputs / behaviors don't change. It appears the argument that's happening here is that some believe it is, that a dive is purely degradation of chemistry and the CAC no-way can jump back up, that this number is useless. My point is that I disagree with that statement. When the number comes back up, an action, input, or condition has changed and gets recalculated into the algo and a new curve or number is generated. I don't know about others, but my CAC always gets recalculated constantly. This occurs after every cool-down and also after every drive after a drive, no matter where the SOC is.

I'm happy to hear others join in with their opinion and add in their expertise like Mark and djp did, otherwise we're not moving forward in understanding our Roadsters better.

Note that any temp that's 30C and higher is considered an elevated temperature for a lithium ion cell. When you can change the behavior of a pack that was sitting at 33-40C for over 10 hours, but instead, drove the temp of the ESS down to 24-26C for that 10+ hour duration, that new input is fed into the algo and recomputed to come up with a new curve or trend of the CAC.

I still have not heard back from anyone regards the head-to-head CAC milage-athon road trip challenge either. That is the real world test for which we can come up with a valid and concrete conclusion with the CAC. So it would be a challenge with an ESS that I believe is at 149 CAC (call this Roadster A) vs an ESS that dove down to 146, but recovered back up to 153 (call this Roadster B). If the Roadster A keeps driving and the Roadster B runs out of range, we have a valid result. Also if Roadster B passes up Roadster A in this test, again we'd have a valid "real" world result and a better understanding the CAC value.

 

[smorgasbord]

The CAC is our and Tesla's best gauge for overall health of the ESS.

Incorrect. A complete log file history provides the best information. Heck, a single log file provides better information and is easily obtained.

It appears the argument that's happening here is that some believe it is, that a dive is purely degradation of chemistry and the CAC no-way can jump back up, that this number is useless. My point is that I disagree with that statement.

No-one has made the argument as you state it. You're creating your own strawman to take down.

I still have not heard back from anyone regards the head-to-head CAC milage-athon road trip challenge either.

For me, that's because I consider it pointless.


What I have repeatedly said is that the CAC number is simply not accurate. We have in wiztecy's own Roadster a case where his battery at 146 was better than his battery at 153. So, CAC values have to at least be greater than 7 Amp-Hours to possibly make a noticeable difference in battery health, but we don't have enough information yet to know just how much bigger than 7 the value has to be to actually point to a true battery chemistry health difference.

It's too bad what I'm saying isn't penetrating, but as sellers of Roadsters you should take advantage of people's ignorance to coddle your battery to raise its CAC number. Even though you may have to wait a few months, some people will think it's worth more when it's older and has more miles on it.