Collection of studies about LiFePo4 degradation
- Thread starter founder
- Start date
newbostonconst
Solar Enthusiast
- Joined
- Sep 24, 2019
- Messages
- 1,007
Thanks
Most of the studies looks fairly recent, good info. Several of my takeaways (previously found and backed by other research), is that no matter how you store LFP packs, they will degrade as time passes. The best case scenario of low temp and low soc is still 0.5% per year, and goes up from there. So those planning to get 20 years from a pack by very lightly using it, may do some math to establish the viability and value of such a proposal.
The PV usage chart for 25C shows a 5% loss per year with regular cycling. I think that's a good starting place, though lower might be reasonable if less cycles are accumulated.
Having done a little research into standby and backup power systems, I think a design life target of 5-6 years is a good trade off to maximize value. That would net 1,825 daily cycles.
I think its important to note that we aren't designing around 3,000+ cycles, even though the battery may be rated for that in lab conditions. The cycle life being that high is really just an indicator of degradation from cycling. It DOES NOT take into account calendar aging which is just as significant. All that figure tells us, is that we can expect a certain capacity loss per cycle in certain conditions.
For example end of life is 80% of new capacity. Lab cycle rating is 3,000 cycles. So 20% / 3000 = 0.0066% per cycle, or 150 cycles per 1% loss.
That is about 2.4% per year cycling daily. Now we bring in the calendar loss, which is around 1-4% per year. Which yields a yearly net capacity loss of 3.4-6.4% per year for daily cycling applications.
So 20%/3.4 = 5.88 years before end of life if cycling daily in modest temps. If temps are above 30C regularly, this will be a fair bit shorter.
The PV usage chart for 25C shows a 5% loss per year with regular cycling. I think that's a good starting place, though lower might be reasonable if less cycles are accumulated.
Having done a little research into standby and backup power systems, I think a design life target of 5-6 years is a good trade off to maximize value. That would net 1,825 daily cycles.
I think its important to note that we aren't designing around 3,000+ cycles, even though the battery may be rated for that in lab conditions. The cycle life being that high is really just an indicator of degradation from cycling. It DOES NOT take into account calendar aging which is just as significant. All that figure tells us, is that we can expect a certain capacity loss per cycle in certain conditions.
For example end of life is 80% of new capacity. Lab cycle rating is 3,000 cycles. So 20% / 3000 = 0.0066% per cycle, or 150 cycles per 1% loss.
That is about 2.4% per year cycling daily. Now we bring in the calendar loss, which is around 1-4% per year. Which yields a yearly net capacity loss of 3.4-6.4% per year for daily cycling applications.
So 20%/3.4 = 5.88 years before end of life if cycling daily in modest temps. If temps are above 30C regularly, this will be a fair bit shorter.
You are just guessing at calendar ageing.
I personally know of dozens of house banks that are over a decade old and still over 80% capacity.
It is impossible to lab test for calendar ageing without accelerating cell degradation.
I don't doubt you, but how many of those packs use cells like CALB which have 2-5% higher than rated capacity when new? How many were cycled daily? Those are obvious caveats. If a pack doesn't see daily cycling (say only 150 cycles per year), then it could see 1% less capacity loss per year. Which allows for 2% capacity loss per year, puts the pack around 80% in 10 years. If it had excess capacity when new, it would still easily test above 80%.
While I don't want to get into a shouting match, I have found many folks will overstate, or state their "guess" at capacity as fact. Kinda like when people report their fuel economy. Though if you have tested these packs yourself that's another matter. I can also point to numerous packs which have 10% capacity loss in 4 years of daily cycling. Supposedly with no abusive conditions, but I can't prove that without a doubt.
Of course pack temperature plays a major role, which needs to be addressed by the designer. If your in the far north or have a temperature controlled environment, that can make a big difference. For example a pack which operates at 10-15C year round, is cycled only a few times a week, and can be stored at low SOC when not used? Its reasonable to design around a longer calendar life. But those are not very common applications. Of course DIY users run the gamut of applications, from full time to seasonal.
Pointing to packs with ideal conditions for life span and excess than nominal capacity when new doesn't contradict the summation I posted above. In some cases the exceptions can prove the "rule" or at least the science. We do a similar analysis on the lifespan on aircraft parts at work pretty regularly. The MFGs (and other airlines sometimes) point to a handful of corner cases as evidence that our analysis is wrong. For example airlines which operate only in cold areas, or areas without dust, etc. Some only fly a few cycles a day while others do a dozen. Accounting for these types of variables to produce a mean life project which is accurate enough for million dollar parts is important to management.
Cell designs have been improving though, so its a moving target to say the least. When the long term lab testing finishes on those cells we will have a better picture of the trends.
Last edited:
For those of us putting LiFePO4 into an RV, I have to wonder if we can expect a life at the far end of the curve. I say this since we don't cycle our batteries down very many times a year.
Why is always 80% assumed as limit? Ok, when it is an electric car and range is important.
But at a house weight and space does not matter.
So I am designing for 12 slots for 16S 280 Ah LiFePo4 cells. At the start maybe only 3 slots filled.
Every time, more capacity is necessary, this can be caused by more demand or degradation of the celös,
Just put one more 16S 280 Ah parallel.
Even when a pack has only 30% of original capacity, it still helps over all performance and decreases the stress on new packs.
* Charging and discharging currents are lower
* Depth of discharge is lower.
Only when all 12 slots are used up, the oldest pack will go to recycling. Maybe in 30 years.
For this strategy is a special BMS necessary, able to hold a pack use able,
when the differences between cells increases after many years.
BTW, I made this PDF as I got different offers for 48V 200 Ah packs:
US$ 2400 with 6000 cycles
US$ 1600 with 2000 cycles
Ok, when I would just design a ferryboat with 3 cycles a day, I would sure take the 6000 cycles cells.
But here it was much more complicate.
40 kWh with 2000 cycles for US$ 6400
30 kWh with 6000 cycles for US$ 7200
What is after 15 years? 24 kWh remaining capacity
would be 80% at the 30 kWh pack,
but 60% at the 40 kWh pack.
But at a house weight and space does not matter.
So I am designing for 12 slots for 16S 280 Ah LiFePo4 cells. At the start maybe only 3 slots filled.
Every time, more capacity is necessary, this can be caused by more demand or degradation of the celös,
Just put one more 16S 280 Ah parallel.
Even when a pack has only 30% of original capacity, it still helps over all performance and decreases the stress on new packs.
* Charging and discharging currents are lower
* Depth of discharge is lower.
Only when all 12 slots are used up, the oldest pack will go to recycling. Maybe in 30 years.
For this strategy is a special BMS necessary, able to hold a pack use able,
when the differences between cells increases after many years.
BTW, I made this PDF as I got different offers for 48V 200 Ah packs:
US$ 2400 with 6000 cycles
US$ 1600 with 2000 cycles
Ok, when I would just design a ferryboat with 3 cycles a day, I would sure take the 6000 cycles cells.
But here it was much more complicate.
40 kWh with 2000 cycles for US$ 6400
30 kWh with 6000 cycles for US$ 7200
What is after 15 years? 24 kWh remaining capacity
would be 80% at the 30 kWh pack,
but 60% at the 40 kWh pack.
This is only a risk with only 1 string of cells.
At all parallel and more then 4 strings, the risk is no problem. The remaining can still do the job.
When the house computer controls all BMS, the software can tell which cell is the problem.
Maybe it is possible to identify risk cells before the become a problem.
When the effect follows a logarithmic curve and it is 5 years to 80%
5 years 80%
10 years 60%
15 years 51.2%
20 years 40.96%
25 years 32.77%
30 years 26.21%
It's not very logarithmic if I recall. Almost exponential near the end. Once you get under 60% cell failure rates get very high, imbalance issues become prominent as well. At some point the anode starts to degrade at an accelerated rate, which can result in wild swings and internal resistance.
Each Operator can assess the performance of the pack. There's no point in removing a pack that performs acceptably for the application, as long as the failure risk is tolerable.
Each Operator can assess the performance of the pack. There's no point in removing a pack that performs acceptably for the application, as long as the failure risk is tolerable.
Here a post how this happens https://diysolarforum.com/threads/shenzhen-basen-280ah-issue.14327/#post-159611
One cell is a little bit weaker.
This weaker cell has more stress, because more discharged.
So the weaker cell degrades faster.
A BMS with only top level balancing can not save this cell from faster degradation.
But when the BMS monitors for problem cells and
reduces stress for the problem cell in all phases,
it should be possible to hold a pack much longer.
Maybe we know it 2040.
I replied to you in the other thread and I will reply here with more detail. If the cells are kept within the knees your concern becomes invalid.
I only have a 5mv to 10mv difference while charging and discharging between the knees. I do not think there will be many that will be charging until a cell reaches 3.65 volts and discharging until a cell is 2.50 volts except when testing the capacity of the pack.
When charging my 8S pack my highest cell was 3.650 and the lowest 3.567. The cells started to drift more than 10mv's when the packs voltage was 27.24. When discharging my highest cell was 3.007 volts and my lowest cell was 2.5 volts.
I am discharging the pack now and will watch to see when the cells start to drift more than 10mv's and record the voltage of the pack. Right now it's looking like keeping the pack between 24 and 27.24 volts is the sweet spot between the knees. This equates to 3.00 and 3.405 volts per cell.
Next time I cycle the pack I will stay within the knees to see how many ah's I get out of them.
Everyone needs to keep in mind the EVE cells are not capacity matched and I am sure the same will be true of the Lishens and most other "value" cells forum members are getting. The cost of capacity matched cells is more and probably not worth the price difference. If needed one could take the difference in the price and order more cells and have much more capacity.
So all is about how to keep "value" cells working for a long time.
I think the solution for this is a better more intelligent BMS and
the possibility to have older and newer packs parallel, until the oldest pack is very much down, maybe to 30% if possible.
D
Deleted member 9967
Guest
For those who don't understand how they can test for "age". they don't. They test for cycles.
At 1C discharge and 1C charge they can do 12 cycles a day.
in one year that is 4,380 cycles.
So if they test at a certain amount of discharge [say to 10%] and charge [say to 90%] and find that it can go for 6,000 before it reaches the "magic number" of 80% efficiency. They they also know that if you discharge and charge daily it will last 6,000 days.
So they test for cycles not for an actual number of months or years.
If you only charge and discharge [or cycle] once a week. They it will last you 6,000 weeks.
I hope this clears things up a bit.
At 1C discharge and 1C charge they can do 12 cycles a day.
in one year that is 4,380 cycles.
So if they test at a certain amount of discharge [say to 10%] and charge [say to 90%] and find that it can go for 6,000 before it reaches the "magic number" of 80% efficiency. They they also know that if you discharge and charge daily it will last 6,000 days.
So they test for cycles not for an actual number of months or years.
If you only charge and discharge [or cycle] once a week. They it will last you 6,000 weeks.
I hope this clears things up a bit.
