I’ve got 3 x 4.8kwh LifeP04 batteries rated at 80% DOD over 6000 cycles - that is over 16 years of cycles and my warranty period is only 10 years.

When they were installed, my installer set up the inverter to drain down to 30% DOD  overnight before switching to grid and leave 10% in the batteries to cover for load shedding and power outages in the middle of the night.

Now nobody likes paying for insurance and I suggest nobody likes leaving 10% of battery on the table every day for the odd occasion that Eskom goes boom while the sun is not shining. 
So here is my idea. For the odd occasion that I need the extra battery, I’m comfortable draining the battery to 90% while keeping essentials running. I’m not going to be doing this every day - perhaps less than 10 times a year, or 160 cycles in the supposed life of the battery.

So I'm rather going with a 80% DOD every night and in an emergency allowing the battery to drain to 90% before switch off. This reduces my payback period from 7.3 years to 6.4 years instantly.

What do you guys think? Anything wrong with my logic.

 

2 hours ago, RichardZA said:

I’ve got 3 x 4.8kwh LifeP04 batteries rated at 80% DOD over 6000 cycles - that is over 16 years of cycles and my warranty period is only 10 years.

Firstly, there isn't a lot of peer reviewed research for LiFePO₄, but the amount there is clearly shows that you will not get 16 years.  Age is significant factor in LiFePO₄ batteries (as with all battery chemistries but more so than it is for Lead Acid, in other words, Lead Acid can outlast, in terms of age, if you assume no cycles).

People like pointing out how things work on Solar Panels or Lead Acid.  However each battery chemistry is exactly that, an ongoing chemical reaction.  Solar Panels aren't chemical reactions (photons hitting a solid substrate to release electrons is nothing like a battery which is on-going reaction and has bazillions of electrons constantly migrating regardless of use).  Nor are two chemical reactions comparable just because from the outside it looks like a battery.  I think you may need a healthy dose of realism in that respect.

Secondly, for a LOT of battery manufacturers their warranty isn't worth the paper it is written on.  The company closes down or they simply claim abuse and you have no recourse.  So I wouldn't put much stock in the warranty either.  Google is your friend, the evidence is there.  Battery manufacturing is cut throat low margins and the people running these companies see their customers as their enemy (once the product is out the door).

In a LiFePO₄ chemistry, the electrolyte is a mixture that should "block" electrons from moving from the anode to the cathode.  There is undesirable electron migration from the anode to cathode which damages the cells by making it harder for lithium ions to migrate through the electrolyte over time.  Three factors determine the speed of this degradation

1. Battery temperature.  The higher your temperature, the faster your batteries will age.  This makes perfect sense as chemical reactions get more vigorous at higher temperatures.  But LiFePO₄ also shouldn't be charged below 0℃ so there is a balance, 25℃ is generally accepted balance.
2. State of Charge/or cell voltage.  The higher this is the higher the electron migration probability is (Which is why Li-ion batteries last longer at lower states of charge).  Again this makes perfect sense because you have significant number of electrons at a very high energy state that want to move to a lower energy state.  And because of uncertainty principle, you will ALWAYS have electrons somehow making their way between the anode and cathode.  More charge, more probability.  We are talking on the scale of trillions of trillions of trillions btw.
3. Discharging, each discharge does actually cause degradation also.  This is because a barrier forms between the anode and cathode that blocks lithium ion migration.  The more you discharge (not per cycle, just discharge in general), the more that barrier forms.

Depth of drain is not such an important factor for LiFePO₄, in simple terms, the amount of Wh the battery can charge and discharge is finite (in total).  Wether you reach that amount by a large or small number of cycles you'll get similar results.  But because of aging it makes more sense to deeply discharge your LiFePO₄ battery to get the most out of it.  Obviously I'm not talking about beyond 90% DoD, then other factors come into play that will significantly reduce age.

Lastly because age is a significant factor in battery age (due to the electron migration problem), you should go for 90% DoD to get the most of out of your battery.  You aren't going to get 16 years, I can guarantee you that.  I would also be surprised if you get 10 years.  And in 10 years battery technology will have moved on significantly.  6 years is a realistic time frame for batteries up to 8 years I would say.

Small print: I'm not a chemist.  I love physics (especially quantum physics), electronics and find chemistry fascinating (but only have enough knowledge in that area to be a danger to myself if I were let loose in a lab).  So take anything I tell you or anyone else on the internet tells you with a grain of salt.  Research papers is the way to go and above is what I've gathered so far.  There isn't a lot of long life studies out there because LiFePO₄ has only existed since around 1996.  Especially don't trust manufacturers, they are trying to sell you something and their testing is not long term.  They make predictions based on short term testing which will be fully covered by their T&Cs saying that they believe their data is accurate.  But above is based on most of my reading and I believe it to be accurate.  If it isn't call me out on it, we can all learn more

Edited December 30, 20213 yr by Gnome

Thanks, Gnome, most informative.

 

Great Post! Promoted to the top.

Awesome!!

Thanks @Gnome great detail and very helpful. 

19 hours ago, Gnome said:

Firstly, there isn't a lot of peer reviewed research for LiFePO₄, but the amount there is clearly shows that you will not get 16 years

19 hours ago, Gnome said:

You aren't going to get 16 years, I can guarantee you that.  I would also be surprised if you get 10 years.  And in 10 years battery technology will have moved on significantly.  6 years is a realistic time frame for batteries up to 8 years I would say.

@Gnome I am very interested in this research you found - do you have any links?  I have spent many hours looking for data on calendar based aging for LiFePO₄ batteries but have not found much of use to me.  The biggest issue is that they generally perform the tests at far higher temperatures than the 20-30ºC that my batteries experience.  

Another question: I am curious how some vendors can sell "2nd life" cells of unknown (to the customer anyway) calendar age if they indeed may last only 6 years from new.  Do you have any thoughts on this?  It is my understanding that those cells would already be 5+ years old when sold....

@Coulomb, from your signature you are running "ex-EV" LFP cells.  Any comments on age / performance / capacity decrease?

 

Edited December 31, 20213 yr by Calvin

Please see my other thread - we need to start crowdsourcing real world data I think

 

13 hours ago, Calvin said:

@Coulomb, from your signature you are running "ex-EV" LFP cells.  Any comments on age / performance / capacity decrease?

It's difficult. Some of mine are from an EV where there was some sort of problem with the BMS installation, and there was a high current event that basically ruined the cells for EV use. One of them had a partly melted post, for example. So I don't know how much life that took off them.

I think the cells are all from 2009, so about 12.5 years of calendar age. The mixing of 100 Ah and 160 Ah cells also complicates matters. Most of my "cells" are 2P of 160 Ah cells, but one is 4P of 100 Ah, and one is a 160 Ah in parallel with 2P of 100 Ah. But it was inexpensive, hence the very unusual composition.

I have about 55% SOH now (45% degradation), although that's not real-world tested. It's just that when I plug in 176 Ah into my software, it seems to track the SOC of the battery the best. The SOC gets corrected at very high and very low battery voltage; 176 Ah (as of now) minimises the size of the steps I get in SOC. 176/320*100% = 55%.

I have other cells ready to parallel with this battery, but I'm slack and haven't installed all the cabling required to install them. These ones are ex-EV too; from a failed bus conversion project.

I don't know that we can lean much about LFP longevity from my energy system.

5 hours ago, Coulomb said:

I don't know that we can lean much about LFP longevity from my energy system

Well, it certainly suggests that the picture may not be quite as dire as @Gnome suggests.  Given the early-life abuse that your cells appear to have experienced, 12.5 years and counting is very reassuring.

4 hours ago, Calvin said:

Well, it certainly suggests that the picture may not be quite as dire as @Gnome suggests.  Given the early-life abuse that your cells appear to have experienced, 12.5 years and counting is very reassuring.

Well most literature consider the battery dead after reaching 70% of design capacity because failures can become unpredictable at that point (what the literature states, not me).  @Coulombhas your batteries had any unexpected side effects running past this 70% barrier or do you find it to be quite reliable?

Again, I'm only pointing out what I've read. It is good to get some data and as I explicitly called out, it is a good thing to call out my post with better data if you have it!  I'm not offended by any stretch   If you can't accept your statements are wrong you can't learn anything.  Also why I said take anything I say with a grain of salt

On 2021/12/31 at 2:34 PM, Calvin said:

@Gnome I am very interested in this research you found - do you have any links? 

 

Sure I'll go through my search history.  The research did show at various temperatures if I recall, not just high.  But at high temperatures it did fair pretty terrible in terms of life.  I was quite shocked actually.

On 2021/12/31 at 2:34 PM, Calvin said:

Another question: I am curious how some vendors can sell "2nd life" cells of unknown (to the customer anyway) calendar age if they indeed may last only 6 years from new.  Do you have any thoughts on this?  It is my understanding that those cells would already be 5+ years old when sold....

Well I bought "2nd life" cells.  I emailed the company and they claim the cells have never been cycled but they are older.  2nd life cells I think can mean anything so it is hard to draw conclusions.  But typically I suspect 2nd life cells would be sold after reaching a certain capacity, not age.  So for example once it reaches 80-90% design capacity they resell them as "storage" batteries.

Edited January 1, 20223 yr by Gnome

On 2021/12/30 at 4:45 PM, RichardZA said:

I’ve got 3 x 4.8kwh LifeP04 batteries rated at 80% DOD over 6000 cycles - that is over 16 years of cycles and my warranty period is only 10 years.

When they were installed, my installer set up the inverter to drain down to 30% DOD  overnight before switching to grid and leave 10% in the batteries to cover for load shedding and power outages in the middle of the night.

Now nobody likes paying for insurance and I suggest nobody likes leaving 10% of battery on the table every day for the odd occasion that Eskom goes boom while the sun is not shining. 
So here is my idea. For the odd occasion that I need the extra battery, I’m comfortable draining the battery to 90% while keeping essentials running. I’m not going to be doing this every day - perhaps less than 10 times a year, or 160 cycles in the supposed life of the battery.

So I'm rather going with a 80% DOD every night and in an emergency allowing the battery to drain to 90% before switch off. This reduces my payback period from 7.3 years to 6.4 years instantly.

What do you guys think? Anything wrong with my logic.

 

What type of inverter charger are using which gives you this level of flexibility and does this come with a cost to match?

12 hours ago, Gnome said:

@Coulombhas your batteries had any unexpected side effects running past this 70% barrier or do you find it to be quite reliable?

Well, I do get "the clunks" now and then. That's when my last defence BMS action applies, and disconnects the source contactors (5 of them) with a resounding clunk, because of overshoots by the Axperts. What seems to happen is that one of the usual suspects (it's not always the same one, but one of just a few) shoots up in voltage. It would not be a problem if the Axperts didn't overshoot. I suspect that it's temperature related. One does or does not get hot from the heat of the bypass resistors, and the hotter one(s) has(have) (a) higher effective capacity, and it's the lower capacity one that goes over-voltage.

But I seem to have had this problem for a long time, so I don't think it's something that has happened only when some SOH percentage is crossed. I just reduce the absorb and increase the float voltage for a day or two, they all bypass and equalise for longer, and the problem goes away. I think perhaps a slight redesign of the system would fix it, but I can't be bothered.

11 hours ago, Moffat said:

What type of inverter charger are using which gives you this level of flexibility and does this come with a cost to match?

Deye 8kW inverter around R30k.

On 2021/12/30 at 4:45 PM, RichardZA said:

So I'm rather going with a 80% DOD every night and in an emergency allowing the battery to drain to 90% before switch off. 

I have the same inverter but with 2x5.1kWh batteries and I use the same battery settings. To me the batteries are required for loadshedding / power failures at awkward times but now that I have them I sweat them to maximise the economic benefit of the system. I've actually considered dropping to 10% every night, but feel safer to stay clear of the stated specs. Speaking of which, I still don't get how manufacturers claim 80, 90 and even 100% DOD with the same battery chemistry

1 hour ago, Scubadude said:

I have the same inverter but with 2x5.1kWh batteries and I use the same battery settings. To me the batteries are required for loadshedding / power failures at awkward times but now that I have them I sweat them to maximise the economic benefit of the system. I've actually considered dropping to 10% every night, but feel safer to stay clear of the stated specs. Speaking of which, I still don't get how manufacturers claim 80, 90 and even 100% DOD with the same battery chemistry

I have a question, how do you determine the DOD is it based on battery KWh or battery A/h or individual cell voltage

In other words if your battery is 5.1 KW/h 100 A/h   bringing battery down 20% DOD would 1.02 KW/h or should one rather work out on Amp/hour 20 A/h

Edited January 2, 20223 yr by Antonio de Sa

5 hours ago, Antonio de Sa said:

In other words if your battery is 5.1 kWh 100 Ah   bringing battery down 20% SOC would 1.02 kWh or should one rather work out on Amp·hour 20 Ah

Good question, though I had to correct some of your units and terms.

Ideally, it should be energy (kWh), but since that's rather awkward to calculate and LFP have a fairly flat battery voltage, it's usually done on coulombs† remaining (Ah). Ah are easier to calculate.

If the battery voltage was perfectly constant, it would amount to the same thing. But obviously, all batteries have a slightly different voltage when full and empty, and when charging and not, and there is hysteresis. Because of all these factors, it's difficult to determine what the remaining energy in kWh actually is. Also, LFP batteries are almost perfectly Ah efficient; the number of Ah available on discharge is very close to the amount put in when charging. So the inefficiency comes mainly from the fact that the battery voltage charges at a higher voltage than it discharges, at the same SOC (as determined by Ah). Clear as mud? 🥴 [ Edit: So that makes the Ah based SOC figure more accurate; you don't have to fudge it for Ah inefficiency. ]

† Technically, coulombs x 3600; coulombs x 3600 = Ah.

Edited January 2, 20223 yr by Coulomb

13 hours ago, Coulomb said:

Well, I do get "the clunks" now and then. That's when my last defence BMS action applies, and disconnects the source contactors (5 of them) with a resounding clunk, because of overshoots by the Axperts. What seems to happen is that one of the usual suspects (it's not always the same one, but one of just a few) shoots up in voltage. It would not be a problem if the Axperts didn't overshoot. I suspect that it's temperature related. One does or does not get hot from the heat of the bypass resistors, and the hotter one(s) has(have) (a) higher effective capacity, and it's the lower capacity one that goes over-voltage.

But I seem to have had this problem for a long time, so I don't think it's something that has happened only when some SOH percentage is crossed. I just reduce the absorb and increase the float voltage for a day or two, they all bypass and equalise for longer, and the problem goes away. I think perhaps a slight redesign of the system would fix it, but I can't be bothered.

Two questions:

1. What kind of voltages are you using for your "float", "equalize/bulk" and "shut-off"?
2. What did you do re-custom BMS?  I've been building a "digital variac" which uses a stepper motor, CT (current transformer) and VT (voltage transformer) for measurements.  But while building this I did a lot of experimenting with shunts and hall effect sensors.  Ultimately I enjoy it quite a bit.  I would never, ever consider running "production" with a home BMS.  I just don't trust my software enough without extensive testing (years in the software industry made me wary).  I'm far too scared that the software I write locks up or I experience an edge case I didn't anticipate and doesn't cut out when it should.  And last thing I want to do is spend months testing, I'm too lazy 😛  But I have considered building my own voltage, current measurement side of the BMS (at least starting there).  Personally I mostly code on STM32 and their dual 12bit ADC is much more stable and much faster than the crap they are using in these bought BMS.  Curious what you've done in this aspect

23 hours ago, Scubadude said:

Speaking of which, I still don't get how manufacturers claim 80, 90 and even 100% DOD with the same battery chemistry

I think a lot of it is marketing. The battery manufacturer looks at the curve and says I will market 6000 cycles @ 80% rather than 3000 cycles @ 100%. They're not saying you will blow up the battery if you take it to 90% or 95% DOD. Nothing on my batteries says the warranty is void if it take it to 95%. I will ask the installers after the break.

 

So basically I think they say do I want to market this at 100,000 cycles at 25% DOD or 3,000 cycles at 100% DOD. Of course there would be minor differences in technology around the battery, but the fundamental chemistry of all these brands is the same.

This paper which compares LFP with NCA and NMC cells, gives real world results. Their problem is by the end of the study most of the LFP batteries were still going strong! "Most of the LFP cells had not reached 80% capacity by the conclusion of this study for the NCA and NMC cells, and their longer-term degradation will be reported in a later work."

https://iopscience.iop.org/article/10.1149/1945-7111/abae37

Edited January 3, 20223 yr by RichardZA

18 hours ago, Gnome said:

1. What kind of voltages are you using for your "float", "equalize/bulk" and "shut-off"?

I run 16S LFP, with 53.7 V float, 55.2 V absorb/bulk, and a 50.8 V cutoff. The latter is really so that my patched firmware will effectively treat the back to grid voltage as 50.8 + 0.5 = 51.3 V. I find 52 V too high and 51 V too low, so this hidden/undocumented interaction is useful. (Non-patched or non-LFP patched firmware has a 2.0 V offset, not 0.5 V).

18 hours ago, Gnome said:

2. What did you do re-custom BMS?

It uses a TI 16-bit microcontroller, in the MSP430 family. We use the same chip but in a 20-pin variant (from rusty memory, maybe it's 16 pins) in the Cell Management Units on top of each prismatic cell (or cell buddy pair/triplet/quad). Communications is serial, switching to industrial optic fibre at the end of a row of cells, and to/from the BMS. All coded in assembler (with macros for if/then/else, switch, loops etc). We use a current shunt and a bunch of contactors. A BeagleBone Black (similar to a Raspberry Pi) for data gathering and ethernet comms.

Quote

I've been building a "digital variac" which uses a stepper motor, ...

Whoa! How does that fit with a BMS? Variacs, current transformers etc are all AC only, right? It sounds intriguing.

18 hours ago, Gnome said:

I would never, ever consider running "production" with a home BMS.  I just don't trust my software enough without...

Heh. Weber and I had ideas of selling our BMS as a production system, but then the thought of maintaining all those systems with buyers of varying levels of competence, and just the variety of corner cases etc etc helped us come to our senses. It's all open sourced and open hardware, so anyone can copy it, but we don't offer support. I don't believe that many have used it.

On 2021/12/30 at 6:47 PM, Gnome said:

Firstly, there isn't a lot of peer reviewed research for LiFePO₄, but the amount there is clearly shows that you will not get 16 years.  Age is significant factor in LiFePO₄ batteries (as with all battery chemistries but more so than it is for Lead Acid, in other words, Lead Acid can outlast, in terms of age, if you assume no cycles).

People like pointing out how things work on Solar Panels or Lead Acid.  However each battery chemistry is exactly that, an ongoing chemical reaction.  Solar Panels aren't chemical reactions (photons hitting a solid substrate to release electrons is nothing like a battery which is on-going reaction and has bazillions of electrons constantly migrating regardless of use).  Nor are two chemical reactions comparable just because from the outside it looks like a battery.  I think you may need a healthy dose of realism in that respect.

Secondly, for a LOT of battery manufacturers their warranty isn't worth the paper it is written on.  The company closes down or they simply claim abuse and you have no recourse.  So I wouldn't put much stock in the warranty either.  Google is your friend, the evidence is there.  Battery manufacturing is cut throat low margins and the people running these companies see their customers as their enemy (once the product is out the door).

In a LiFePO₄ chemistry, the electrolyte is a mixture that should "block" electrons from moving from the anode to the cathode.  There is undesirable electron migration from the anode to cathode which damages the cells by making it harder for lithium ions to migrate through the electrolyte over time.  Three factors determine the speed of this degradation

1. Battery temperature.  The higher your temperature, the faster your batteries will age.  This makes perfect sense as chemical reactions get more vigorous at higher temperatures.  But LiFePO₄ also shouldn't be charged below 0℃ so there is a balance, 25℃ is generally accepted balance.
2. State of Charge/or cell voltage.  The higher this is the higher the electron migration probability is (Which is why Li-ion batteries last longer at lower states of charge).  Again this makes perfect sense because you have significant number of electrons at a very high energy state that want to move to a lower energy state.  And because of uncertainty principle, you will ALWAYS have electrons somehow making their way between the anode and cathode.  More charge, more probability.  We are talking on the scale of trillions of trillions of trillions btw.
3. Discharging, each discharge does actually cause degradation also.  This is because a barrier forms between the anode and cathode that blocks lithium ion migration.  The more you discharge (not per cycle, just discharge in general), the more that barrier forms.

Depth of drain is not such an important factor for LiFePO₄, in simple terms, the amount of Wh the battery can charge and discharge is finite (in total).  Wether you reach that amount by a large or small number of cycles you'll get similar results.  But because of aging it makes more sense to deeply discharge your LiFePO₄ battery to get the most out of it.  Obviously I'm not talking about beyond 90% DoD, then other factors come into play that will significantly reduce age.

Lastly because age is a significant factor in battery age (due to the electron migration problem), you should go for 90% DoD to get the most of out of your battery.  You aren't going to get 16 years, I can guarantee you that.  I would also be surprised if you get 10 years.  And in 10 years battery technology will have moved on significantly.  6 years is a realistic time frame for batteries up to 8 years I would say.

Small print: I'm not a chemist.  I love physics (especially quantum physics), electronics and find chemistry fascinating (but only have enough knowledge in that area to be a danger to myself if I were let loose in a lab).  So take anything I tell you or anyone else on the internet tells you with a grain of salt.  Research papers is the way to go and above is what I've gathered so far.  There isn't a lot of long life studies out there because LiFePO₄ has only existed since around 1996.  Especially don't trust manufacturers, they are trying to sell you something and their testing is not long term.  They make predictions based on short term testing which will be fully covered by their T&Cs saying that they believe their data is accurate.  But above is based on most of my reading and I believe it to be accurate.  If it isn't call me out on it, we can all learn more

Great stuff thanks