How many cycles from a LiFePO4 cell?

[evm1024]

A recent paper published this year is quite interesting.

Capacity Fading Mechanism of the Commercial 18650 LiFePO4 Based Lithium-Ion Batteries: An in Situ Time-Resolved High-Energy Synchrotron XRD Study

https://pubs.acs.org/doi/pdf/10.1021/acsami.7b13060

This paper looks into capacity fade vs number of cycles for A123 LiFePO4 cells. Of course 1.1 AH A123 18650 cells are not 400 AH prismatic cells but there are some lessons to be learned.

First off the paper defines a cell to be failed when it retains 80% of its initial capacity. This is in line with industry standards. The cells still work it is just that they have lost 20% of their capacity.
Secondly, the paper uses 100% of the cells capacity in the test.

To reach 100% SOC they charge the cell at 1C (CC @ 1100mA) until the cell voltage reaches 3.6 VPC and then continue at 3.6 VPC (CV) until the tail current is reduced to 0.02 amp (20 mA). 20 mA is about 1.8% of C and thus for a 400 AH cell that would be 7.27 amps (12.7 amps for a 700 AH cell).
After the cell is fully charged they wait about 5 minutes then discharge it at 1C (1.1 amp) until the cell voltage reaches 2.0 volts. That is 100% DOD (0% SOC) for the test.
Besides the cell size there are some other differences between the study and how we use LiFePO4 in out boats. The first is that cycling between 3.6 VPC (100% SOC) and 2.0 VPC (100% DOD) is quite a bit more “aggressive” than we typically subject our house banks to. And that the study discharged at 1C which is much larger than I typically discharge my house bank at (0.04C or 28 amps).
On to the results.
At a 1C discharge rate they were able to get 2500 cycles out of their cells (remember that this is to 80% capacity retention). The cells at the 2500th cycle were charged and retained 80% of their initial capacity.
The loss per cycle curve is nearly linear at 1C which means that you lose the same capacity (% wise) in each cycle. Capacity retention at 1C was 99.4%. Interestingly as the C discharge rate went up the loss per cycle went up. Retention at 5C was 98.9%.
Now for the exciting results – the paper indicated that for a discharge rate of 0.1C (40 amps for a 400 AH cell) had a retention of 100% - no loss in capacity. Of course this study did not look into fractional C discharge rates so we should not read too much into this finding. Plus these are 18650 cells and not big prismatics.
The paper goes on to show that the loss of capacity was primarily due to a loss of active lithium ions. (trapped within the FePO4)
These folks are doing some really exciting work. I do have an email into the corresponding author asking if they would consider doing a study cycling between 3.45 VPC and 3.0 VPC at 0.1C (of if they knew of such an existing study).

What does this all mean for us? Well presuming that cycling between 3.45 and 3.0 VPC at 0.1C results in less capacity loss than shown in the study (in other words there is not another failure mode at work) then we can expect to see more than 2500 cycles (to 80% capacity retention) out of a LiFePO4 house bank.

They have another paper on over discharge failures and I will summarize. It is also very interesting.

Of course you will not get this number of cycles if you over charge or discharge. Just like you will not be able to sleep in your house if you burn it down. That goes without saying.

[john61ct]

That is a **very** destructive test protocol IMO, and I'm sure that is intentional.

I really wish someone would sponsor a repeat with, say 100AH prismatic cells, ideally comparing the top 5-6 makers that sell to consumers.

.2C discharge rate, maybe .4C charging.

2.95V at the bottom

3.45V at the top, endAmps .03C

My WAG would be you'd double the cycle lifespan, maybe 5x

Then keep going to 70% SoH, more realistic for individual consumers.

Probably add another 30-50% cycle lifetime.

And that's 100% DoD testing, before looking at only discharging to say 60% or 40% DoD, double again?

Pass them on to your grandkids, batteries no longer a consumable item 8-)

Actually calendar aging may become a factor long before cycling wears them out.

[evm1024]

Originally Posted by john61ct

That is a **very** destructive test protocol IMO, and I'm sure that is intentional.

SNIP!

Pass them on to your grandkids, batteries no longer a consumable item 8-)

Actually calendar aging may become a factor long before cycling wears them out.

LOL - very destructive testing and still get 2500 cycles to 80% residual capacity. How spoiled we (and I do have LiFePO4) have become!

It would be very interesting to get larger prismatics tested to limits more like we use on out boats (very conservative limits).

They tested for at least 2500 cycles and at 1C 1.1 AH cells 1 hour to charge and 1 hour to discharge. So at 2 hours per cycle we have the test running over 200 days.

With that 100 AH cell you want tested (so do I) we have 56 hours or more per cycle. 2500 cycles in 16 years..... Ouch!

Calendar aging - that was one of those other factors I was alluding to.

[john61ct]

Originally Posted by evm1024

LOL - very destructive testing and still get 2500 cycles to 80% residual capacity.

That has been the standard estimate for many years, surely not surprising to you?

How spoiled we (and I do have LiFePO4) have become!

> 2500 cycles in 16 years..... Ouch!

I get much faster than that, check your math.

And of course they have algorithmic accelerated tests they can extrapolate from, maybe not perfect.

But the industry has no incentive to encourage that care profile anyway.

[evm1024]

Originally Posted by john61ct

That has been the standard estimate for many years, surely not surprising to you?

How spoiled we (and I do have LiFePO4) have become!

> 2500 cycles in 16 years..... Ouch!

I get much faster than that, check your math.

And of course they have algorithmic accelerated tests they can extrapolate from, maybe not perfect.

But the industry has no incentive to encourage that care profile anyway.

Think compared to FLA.... 2500 is 10x.... Spoiled indeed.

OK 56 hours per cycle * 2500 cycles = 140,000 hours

140,000 hours / 24 hours per day / 365.25 days per year = 15.97 years

OK my math works out there I must have blown the cycle time calculation. Let's try that.

100 ah cell
0.2 C discharge = 20 amps so 100 / 20 = 5 hours for discharge
0.4 C charge = 40 amps so 100 / 40 = 2.5 hours for charge

Ah there is my problem (how did I come up with 56 hours?) 7.5 hours per cycle.

So 7.5 hours * 2500 cycles = 18,750 hours

18,750 /24 hours in a day / 356.25 days a year = 2.14 years.

Thanks for catching my error!

Good numbers are good to have.

[john61ct]

Originally Posted by evm1024

Think compared to FLA

Great of course, but not surprising.

> 56 hours per cycle

Where did that 56 come from? I use 24 IRL, but could be 6 hours at .2C / 1C as I spec for testing, would obviously be automated.

Ah I see you got there. . .

[evm1024]

Originally Posted by john61ct

Great of course, but not surprising.

> 56 hours per cycle

Where did that 56 come from? I use 24 IRL, but could be 6 hours at .2C / 1C as I spec for testing, would obviously be automated.

Ah I see you got there. . .

I make errors and would rather be corrected. Not a problem. Of course sometimes my ego is bruised, oh well.

[john61ct]

And in fact I see your calc was more correct wrt my specs 8-)

I'd thought I'd spec'd 1C charge rate, but in fact scaled back to a gentler .4C, more IRL realistic as well.

[tanglewood]

I'd love to see any research that shows cycle life WRT charge/discharge rate, say over the 0.25 to C range.


And also showing cycle life WRT SOC range, say covering 100%, 90%, 80%, 70%


In the absence of this, I think all this talk about greater cycle life and how to get it is nothing more than pure speculation. Does it have ANY basis at all?

[john61ct]

The DoD depth vs cycle lifetime curve is well published and discussed at length in recent threads.

Similar relationship as lead, perhaps true for all chemistries I dunno.

But because lifetime is **so** long even at 80% average DoD,

up front acquisition costs are already **so** high,

and just KISS principles,

it is not usually a factor taken into account in care recommendations.

Maine Sail did recently give an update on his own personal experience with his test+production pack,

which he intentionally discharges to 80% DoD every cycle, getting into 4? years with zero sign of SoH declining.

As he stated, that is a sample size N=1, but there are plenty of such reports in the mother House LFP thread I've cited befire and others in the various forums, not just marine but many more off-grid and mobile House banks on land.

Anyone in Australia should contact the installer from Mannum, likely many hundred banks there, and he was (very stridently 8-) advocating avoiding the shoulders from many years back.

[tanglewood]

Originally Posted by john61ct

The DoD depth vs cycle lifetime curve is well published and discussed at length in recent threads.

But is it anything more than a finite ah/khw throughput capability? Pick your DOD & number of cycles and the lifetime ah/kwh throughput is the same no matter what point you pick on the curve?



I get that it's "consensus", but I'm looking for science. People who have done controlled experiments over thousands of cycles and shown concrete differences, then explained why, preferably including dissection of batteries.

[nebster]

The other factor that's left out here is temperature. It confounds the testing, such that we probably will never find a small research lab that actually has the resources to conduct a good parallel, controlled study with so many parameters that can be manipulated. But I suspect time-integrated exposure to heat might become the most significant detriment to lifecycle, once reasonable voltage boundaries are agreed upon.

[evm1024]

Originally Posted by nebster

The other factor that's left out here is temperature. It confounds the testing, such that we probably will never find a small research lab that actually has the resources to conduct a good parallel, controlled study with so many parameters that can be manipulated. But I suspect time-integrated exposure to heat might become the most significant detriment to lifecycle, once reasonable voltage boundaries are agreed upon.

I've got a MS thesis that I am reading through (not from my university) that does take temp into account. It is not as helpful as it could be due to the selection of SOC vs temp. But it does show how temp plays a significant role.

Once I've studied it enough to summarize it I'll post a new thread on it.

[tanglewood]

Originally Posted by evm1024

I've got a MS thesis that I am reading through (not from my university) that does take temp into account. It is not as helpful as it could be due to the selection of SOC vs temp. But it does show how temp plays a significant role.

Once I've studied it enough to summarize it I'll post a new thread on it.

I've seen what might be the same paper. The one I saw was looking at storage at full vs 50% SOC, and storage at 60C vs I think 25C. Temp was the bigger factor, but of course storing at full SOC wasn't good either. But the focus, as I recall, was on storage.

[evm1024]

Originally Posted by tanglewood

I've seen what might be the same paper. The one I saw was looking at storage at full vs 50% SOC, and storage at 60C vs I think 25V. Temp was the bigger factor, but of course storing at full SOC wasn't good either. But the focus, as I recall, was on storage.

That is the one. It shows (at least within the limits of the paper) that storing at 100% SOC or 0% SOC caused irreversible damage to the cells. More so at 100% SOC.

This disproves the assumption that keeping your cells at the lowest SOC is good for them.

And that temperature played a very significant role (more than SOC). Especially at temps above 50 C (60 C being very bad).

35 C was not so bad (95 F)