Awesome study / information on NCA battery degradation (Panasonic NCR18650PD)

First off, thanks to all in this forum for the endless discussions and information.
Now is my time to share back.

How to properly treat a cell and have as much use out of it is my favorite discussion topic here.
I stumbled upon a very interesting scientific study with well controlled parameters, with the focus on lithium battery degradation for EVs;
https://mediatum.ub.tum.de/doc/1355829/file.pdf
much of this information can be also applied for smaller applications like ebikes and even rc.

The cell tested is:
Panasonic NCR18650PD (NCA type)
I would expect the lessons learned here to at least transfer to other NCA cells but also other chemistries to some extent.

Some new takeaways for me (reading as layman) are:

• If sensibly used, NCA cells should last well over 10 years and over 1000 cycles with still around 80% original capacity left (WOOHOO)

• Higher charging currents (in this case starting already > 1a!) DRAMATICALLY increase degradation, everything below 1a has little further positive effect.

• At 10c cell degradation is much higher than at 25c, both in charge and discharge. Higher temps during charge and discharge are not detrimental

• Peaking of discharge current vs constant current draw degradation is similar; i.e. some load spikes when pumping an ebike should not hurt too much as long as it is not a constant state.

Already very well known but confirmed:

• Lower storage temperature and lower storage SoC decrease degradation

• Lower charge end voltage decreases degradation (pretty big differences also between 3.7, 4.0 and 4.1 volts)

• Lower depth of discharge increases life considerably.

From this I take the following lessons for my Ebike:

• I will set my charger end voltage between 4.0 or 4.1 volts and charge whenever I arrive home, less deep discharge is better and with a conservative end voltage the degradation due to storage is not so bad. Only for a big trip I charge it up to 4.2

• Leave the bike half discharged (ie. around 3.6v) if not in use for some time

• Use 1a charging current or lower (per cell, so for 3p, 3a max, etc) if i have the time

• In winter charge with even less amps (pack is outside), do not charge at all below 0c

Hope this was helpful and/or interesting! Looking forward to more discussion !

these conclusions are valid for every (3.6v) chemistry. as they match my tests with PF, GA, 29E and many others.

A derated battery is a happy battery...

Killer first post!!

geekmystique said:

per cell, so for 3s, 3a max,

Click to expand...

3p (not 3s)

It's number of parallel cells that matter for that, not series.

amberwolf said:

3p (not 3s)
It's number of parallel cells that matter for that, not series.

Click to expand...

Thanks! corrected

geekmystique said:

From this I take the following lessons for my Ebike:

• I will set my charger end voltage between 4.0 or 4.1 volts and charge whenever I arrive home, less deep discharge is better and with a conservative end voltage the degradation due to storage is not so bad. Only for a big trip I charge it up to 4.2

Click to expand...

Thanks for the interesting study. Just one remark to your conclusions : not sure why you want to charge whenever you arrive home. That is in direct opposite with the general knowledge and also with the conclusions in this study. You definitely increase the calendar aging.
Citation from the page 71, at the bottom : „Thus, the EV battery should be kept at a low temperature and at a low or medium SoC during nonoperating periods.“ We should consider all the time between the rides as a nonoperating period. And charge just in time, before the next ride. Or charge only partially, if really needed.

docware said:

geekmystique said:

From this I take the following lessons for my Ebike:

• I will set my charger end voltage between 4.0 or 4.1 volts and charge whenever I arrive home, less deep discharge is better and with a conservative end voltage the degradation due to storage is not so bad. Only for a big trip I charge it up to 4.2

Click to expand...

Thanks for the interesting study. Just one remark to your conclusions : not sure why you want to charge whenever you arrive home. That is in direct opposite with the general knowledge and also with the conclusions in this study. You definitely increase the calendar aging.
Citation from the page 71, at the bottom : „Thus, the EV battery should be kept at a low temperature and at a low or medium SoC during nonoperating periods.“ We should consider all the time between the rides as a nonoperating period. And charge just in time, before the next ride. Or charge only partially, if really needed.

Click to expand...

I can imagine a few reasons;

The next ride isn't necessasarily the following morning commute. If it was then starting charge with a timer in the early hours to be ready just before riding would be the most battery friendly. If use is more ad-hoc then it makes sense to keep topped up and rely on the slow charge rate & lowered max SOC for 'battery care'.

in theory that would work but just charging to 4.1 or lower works just fine. the few percent you might gain with delayed charging is nothing compaired to just having a lower SOC limit.

I found this thesis as excellent guide how to evaluate/test the battery cell. (By the way I wonder if the author reads this forum :wink: )

But..

as I have repeated several times this evaluation/results are valid only for this particular cell NCR18650PD. Even other Panasonic NCA cells have significantly different parameters, especially when evaluating cycle life. I have no problem with the author's conclusions how to extend the calendar/cycle life of the cell, but the most important question it is worth it?

For me are important measurements and author's conclusions on pages 87-89. Data clearly shows when the cell is cycled at 100% DoD (4.2-2.5V) it can do 750 EFC (in this case those are real full cycles) before capacity drops to 70% of its initial capacity. And the real benefit of using lower charging voltage (4.1V) for user will be shown not earlier than after 750 EFC. Here are author's quotations:

Comparing the curves for charging to 4.1 V or 4.2 V with 1 A or less in Figure 56a shows that the curves of the different charging voltages do not intersect before ca. 750 EFC. This means that the absolute available capacity remains greater for higher charging voltages, even if the relative capacity fade is larger, as it was shown in Figure 51.

Click to expand...

This means that a reduced charging voltage does not necessarily provide a better absolute capacity retention over the cycle life of the battery, although the relative capacity fade remains lower. When defining the charging voltage for an EV, the manufacturer has to find an appropriate balance between maximizing the driving range and minimizing the battery degradation perceived by the customer. When the maximum charging voltage is reduced, the relative capacity fade decreases. However, the vehicle provides a shorter driving range. When using the maximum charging voltage, a higher driving range is obtained. As the driving range corresponds to the absolute available capacity and as the absolute capacity curves do not intersect in Figure 57a, a higher charging voltage provides still a higher driving range for the aged cells. Yet, the capacity fade perceived by the customer is substantially higher since it corresponds to the fade in relative available capacity. Considering a vehicle age corresponding to 750 EFC for example, the absolute available capacity and, thus, the driving range is still ca. 10% higher for charging to 4.2 V than for charging to 4.0 V, although the relative capacity fade amounts to ca. 28% for the higher charging voltage and only to ca. 20% for the lower charging voltage.

Click to expand...

The same principle can be used for calendar aging evaluation in chapter: 4.4.4.2 Calendar Life Prediction on page 69. You can clearly see that higher resting/storage voltage means higher capacity fade. But you should consider, that if you set the charging voltage to 4.1V per cell for the whole battery life, you will permanently locked cca 10% of available capacity just on the very beginning of its calendar life. So you can expect the same behavior as for cycle life - the practical benefit of using lower charging voltage (4.1V) will be shown not earlier than ca 10 years.

I agree that we should be aware of cells differences.
However, our aim is flexible using of the applicable knowledge in the practice.

For example, for daily commuting is timer good solution. In this case charging to 4,0 - 4,2 V (accordingly to traveling distance) make sense, because cells will spend very short time at the high voltage.

Completely different example is my using of ebike. I live in the small village and use ebike in the summer as a very pleasant and effective local transportation equipment, travelling from 5 to 15 km daily. I usually cycle cells in the range 3,8 – 3,5 V, charging each few days. Sometimes I have longer trip, cca 30 km, so charge to 3,9 V. When I go to all-day trip, charge to 4,1 - 4,2 V.

Is it worth it ? For me, definitely yes.

https://www.researchgate.net/publication/328086745_Extending_Battery_Lifetime_by_Avoiding_High_SOC

View attachment 271 Cell SOC and longevity.pdf

geekmystique said:

If sensibly used, NCA cells should last well over 10 years and over 1000 cycles with still around 80% original capacity left (WOOHOO)

Click to expand...

LiFePO4 cells would do better even if used careless + they support charging/discharging at wider temperature range.

thunderheart said:

LiFePO4 cells would do better even if used careless + they support charging/discharging at wider temperature range.

Click to expand...

and they cost more, take up more volume, weighs more and is harder to determine SoC.
all factors that make it less usable in consumer products. even one of those points is enough to get it striked off as a option in a commercial or consumer product.

lifpo has its place, just not in mobile applications.

docware said:

Is it worth it ? For me, definitely yes.

Click to expand...

You're a special case

By the way Figure 3: Calendar aging which you posted only confirmed what I am trying to explain. This graph construction can be a little tricky. If one set the fixed charging voltage to 4.1V (90% SoC) then the aging course of usable capacity should also start from the 90% SoC value (here from 23.5Ah capacity), not from 100% as is shown in the graph.

In the real world condition you can expect ca 90% average SoC level for fixed charging to 100% (because there will be still plenty amount of time when your battery will sit on lower SoC due to its use). On the other hand, from my personal opinion it is unlikely that an average 15% SoC will be achievable in normal operation without limiting the user behaviour. I will expect ca 40-50% SoC average. If we add those two courses in this graph (100% SoC fixed charging represented by the 90% SoC course will start descending from 26Ah, the 50% SoC will start from 23.5Ah), we will see that the benefit of using lower fixed charging voltage (or keeping the cell sitting at 50% SoC) for lowering calendar aging will show up no sooner than 5-8 years. And of course it differs cell from cell so for many 18650 HE cells you will see this benefit after 10 years or even more.

So it is always good to have data of cycle and calendar aging for desired cell.

Interesting. So .3C or below charge rate is the goal for long life.

Pajda said:

You're a special case

Click to expand...

Maybe not so much, judging by existence of this study.

Well, in my opinion, if you look at two thousand EV users, you will see two thousand various ways of EV utilization. Any expectation may be wrong.
Systematic 15% SoC during parking seems to me in real life unrealistic also. But important is to limit time at high SOC, eventually operate at 50% SOC and bellow if possible, e.g. commuting TESLA owners.

Having aging and cycling data of specific cell would be nice. Unfortunately not possible.

It´s pity that your cycling use unrealistic 100 % DOD. Something like 4,2 – 3,2 V or 4,1 – 3,4 V, ...….., would be closer to real life. :wink:

geekmystique said:

• Lower storage temperature and lower storage SoC decrease degradation

Click to expand...

This is why I have turned my beer fridge into a Lithium fridge.

Little offtopic. I found original 100pcs paper carton with few forgotten LG HG2 cells (supplied by Nkon in 4/2017). The code on cells is P342L072A3 which says that they were manufactured in December 2016. They were supplied with 30% SoC (it was written on the carton and the voltage when arrived was 3.500V). The carton with cells was stored in cca 25°C all the time. This week I made a test measurement. Cell voltage was in the range from 3.492~3.499V.

The measured nominal capacity of HG2 (so after almost 3 years of storage, and at least two years at room temperature) was 2.94Ah / 10.76Wh. 10s DCIR test at 50% DoD shows 26.34mOhm.

I make the same test with Tesla Model S P85D cells from MY2016 (only 3083km driven) also stored at room temperature resting at 3.6V. And the results: 3.11Ah / 11.32Wh, 10s DCIR test at 50% DoD shows 40.68mOhm.

I buy used batteries from a reputable guy or I should say ive bought around 210 25r's that ive abused and I like. I like the amps.

I bought around 110? used ncr 18650pd about 6 months ago and made a 13s4p for my sons 20 amp bike, works fine but runs warm compared to 25r. Ive had the other cells sitting ever since, voltage was 4.18 now its 4.04. I thought I might make a 17s3p as a range extension pack on a phaserunner controller 30 amps continuously, these batteries don't have the amps
I would need a 6p pack of these but only a 3p pack of 25rs. so its either amps or ah.

mnke 26650 5000mah 20amp continuous $5.65 cad new/ 2p is 40amp 10ah $11.30
25r 18650 2500mah 20 amp continuous $5.75 cad new. / 4p is 80amp 10ah $23.00
the Panasonics 2900mah 10amp continuous $6.60 cad new/ 4p is 40amp 11.6ah $26.40

the 26650 weighs twice an 18650

I haven't looked to see if moogs done a test on the mnke anyone use them yet?

Random data point- cells from a 14s5p battery of Panasonic NCR18650A after 150 cycles of 90% to 30% showed almost no degradation of either capacity (approx loss 50mah) or internal resistance (identical according to charger). This was despite being used at 2C discharge with peaks of 3C, and a charge rate of 1/3C. So I believe this study