LiFePO4 Batteries: Discussion Thread for Those Using Them as House Banks

[deckofficer]

Back to cycling, but now with an Amprobe. Found another heater to use that has a 500 watt low setting, which the Amprobe showed 40 amps on the DC side. Much closer to the 0.5 C rate that T1 Terry says is how the cycle life is determined. Ran for 2 hours and 45 minutes or 112.75 a-hr used. Cell voltages at this percentage of discharge in the same order I've always given them...

2.96, 2.91, 2.75, 2.98

For future cycles to be fair to the cells, I'll end the cycle at 100 a-hr instead of when the red voltage LED remains on on the inverter. So will run at 40 amps for 2 1/2 hours instead of 2 3/4 hours. Will share those cell voltages at 2 1/2 hours, but I'm sure they all will be 3.00+.

[deckofficer]

Discharge resting
2.96, 2.91, 2.75, 2.98 (11.6)
Charged resting
3.38, 3.39, 3.48, 3.35 (13.6)

As it has always been, cell #3 goes higher and lower, charge vs discharge.

[senormechanico]

Imho, #3 has the least capacity and/or the highest internal resistance.

[diugo]

Originally Posted by deckofficer

Back to cycling, but now with an Amprobe. Found another heater to use that has a 500 watt low setting, which the Amprobe showed 40 amps on the DC side.

Did the Amprobe show the same 40A draw at both the beginning of the test and at the end? Probably not, since battery voltage changed dramatically. So check if your Amprobe can measure average current. If not, then averaging the two currents should be pretty close.

I am very interested in the Peukert coefficient of your batteries, since we both have LFPs that are several years old. At your earliest convenience, if you could perform and time another drain test, this time with a much different load current, we can easily calculate the coefficient using the formula n = (logT2 - logT1) / (logI1 - logI2). We already have T1=2.75 and I1=40.

[T1 Terry]

Originally Posted by diugo

I have had my 260Ah "Balqon Special" set for almost a year now, and routinely avoid both knees of the charge curve on a daily basis. My last bottom balance was ten months ago, and the voltage spread from high to low cell never exceeds 30mV and averages about half that. In order to maximize cell longevity, I only charge my bank to 80% SoC---so I cannot top balance.

if you don't top balance, how do you determine you have reach 80% SOC and stop charging? What cell voltage and technique did you use to determine the bottom balance?

T1 Terry

[T1 Terry]

Originally Posted by deckofficer

Discharge resting
2.96, 2.91, 2.75, 2.98 (11.6)
Charged resting
3.38, 3.39, 3.48, 3.35 (13.6)

As it has always been, cell #3 goes higher and lower, charge vs discharge.

When you do your next capacity test, the rested charged voltage should be 3.4v or better, I generally settle for 3.5v after 2 hrs. it is a fiddle getting all the cells to hold above 3.5v but it will give you a genuine full indication.
Then, perform the 40 amp discharge, I'm guessing the 40 amps is at around 13v at the beginning and climbs around 45 amps at the end of the test, a roughly constant 520w load. Set an alarm for any cell dropping to 2.8v under load, stop when this cell voltage drops further to 2.78v. the number of ah drawn during this test is the useable Ah capacity of that 4 cell battery as this is the point that the weakest cell is fully discharged. That will be the significant figure and will be interesting to see after the thrashing you have given those cells. You have pushed them way beyond any correctly sized house battery would suffer so your results will be those of extreme test conditions.
I think my next capacity test will show just how much punishment these cell can handle and still live to tell the tale. When I first ventured into hands on testing these batteries I would have declared that what I've done to these cells would have surely killed them, but they are still appear to be functioning as if nothing had happened.

T1 Terry

[diugo]

Originally Posted by T1 Terry

if you don't top balance, how do you determine you have reach 80% SOC and stop charging? What cell voltage and technique did you use to determine the bottom balance?

T1 Terry

I count amp hours charging and discharging, which I admit is quite an inexact science---since my 3 y/o LFPs have a coulombic efficiency less than 1.00, and one that seems to vary wildly with cell temperature.

My meticulous bottom balance procedure involves discharging the series string at C/25 down to 12.00V, then slowly discharging one cell at a time at C/100 until each reaches 2.800V---virtually dead as a doornail. I then hook all the cells together in parallel for an hour to equalize the miniscule remaining charge between them. So when separated, the cell voltage spread is only a few millivolts. I quickly recharge 200Ah to get them up to near 80% SoC.

[deckofficer]

Doug and Terry,

40 amps was the average, starting at 39 amps and ending at 41 amps. The last cycle was a capacity test as I ran it to 11.6 volts pack and 2.75 lowest cell. 100 a-hr rated cells delivered 112.75 amps.

Future cycles will removed 80 a-hr, then charge. I'll do another capacity test at 50 cycle intervals.

It is going to be a long time for me to catch up with the cycles you and Maine Sail have, but doing a capacity test on the 50th cycle till I hope 2000 cycles should show a good cycle vs capacity vs age results.

[Maine Sail]

Originally Posted by diugo

Did the Amprobe show the same 40A draw at both the beginning of the test and at the end? Probably not, since battery voltage changed dramatically. So check if your Amprobe can measure average current. If not, then averaging the two currents should be pretty close.

I am very interested in the Peukert coefficient of your batteries, since we both have LFPs that are several years old. At your earliest convenience, if you could perform and time another drain test, this time with a much different load current, we can easily calculate the coefficient using the formula n = (logT2 - logT1) / (logI1 - logI2). We already have T1=2.75 and I1=40.

I would love to discuss this more because I find it does not always mesh... I have run those calculations for many of the branded batteries I use and the Peukert calculations rarely matches up with the Peukerts Constant the battery manufacture calculates.

For example some of the available Peukert calcs out there, including the Victron calculation, suggest a Peukert of 1.39 for a T105 yet Trojan Battery engineering department says it is a 1.24...???? That would be a huge difference in counting errors to a battery monitor.

I ran another battery I use and that is a US Battery L16 HC. It comes up with a Peukert of 1.16 yet the VP of Engineering at US Battery says it has a Peukert of 1.26562. (a very calculated number if I do say so myself)

If we take a 450Ah bank of Trojan T105's at 1.39 and apply an average load of 6A with a calculated Peukert of 1.39, we would get approx 750Ah's out of that bank with a 6A load.(if we discharged to 10.5V)

Now take the same battery and apply the Trojan supplied 1.24 Peukert, at the same 6A load. The bank now becomes 618Ah's at the 6A load....Again a battery monitor is going to get very confused over this...

How is it that battery manufactures calculate a Peukert that is very, very often different than the Peukert calculations get you to..?

Does anyone know the actual method battery manufacturers use to calculate a Peukert? It seems to be different from the standard calculations? I have tried a number of Peukert calculators, and calculations run out manually, yet I find it very hard to match the manufacturer supplied Peukert?

That said I have capacity/time tested the manufacturer supplied Peukert on a number of occasions and found them to be pretty accurate.....

I suppose this could be another thread entirely....

[diugo]

I performed a half dozen drain tests, all at different currents, and got quite good correspondence with calculated Peukert coefficients all between 1.05 and 1.07 and averaging 1.059.

Each test was carefully performed. In all cases, the battery was fully bulk charged to 13.9V, then the voltage was held there until absorption current fell to exactly 5.0 amps. During each test, discharge current Id was kept as constant as possible, with resistive loads periodically added as battery voltage fell to the termination voltage Vt. The hardest part was finding the correct Vt, as it is a function of Id. The goal here is to have the battery display the exact same resting voltage Vd one hour after every test. This required quite a bit of experimentation. For example, at C/100, the test might be terminated at Vd=11.23V. But at C/10, this might require taking the battery down to 10.77V. The higher the discharge current, the lower the termination voltage.

As to why manufacturer Peukert values vary from real world observations, my guess is that the values were either calculated theoretically, or with prototypical sample cells under ideal lab conditions. In other words, just like cycle life and all other boastable battery parameters.

[Maine Sail]

Originally Posted by diugo

I performed a half dozen drain tests, all at different currents, and got quite good correspondence with calculated Peukert coefficients all between 1.05 and 1.07 and averaging 1.059.

Each test was carefully performed. In all cases, the battery was fully bulk charged to 13.9V, then the voltage was held there until absorption current fell to exactly 5.0 amps. During each test, discharge current Id was kept as constant as possible, with resistive loads periodically added as battery voltage fell to the termination voltage Vt. The hardest part was finding the correct Vt, as it is a function of Id. The goal here is to have the battery display the exact same resting voltage Vd one hour after every test. This required quite a bit of experimentation. For example, at C/100, the test might be terminated at Vd=11.23V. But at C/10, this might require taking the battery down to 10.77V. The higher the discharge current, the lower the termination voltage.

As to why manufacturer Peukert values vary from real world observations, my guess is that the values were either calculated theoretically, or with prototypical sample cells under ideal lab conditions. In other words, just like cycle life and all other boastable battery parameters.

I am working with a company in California to get a battery discharge tester that holds the load constant and is specifically designed for this.. They build them for Concord battery but are geared towards the Aircraft requirements in terms of reporting. The owner is tweaking the reporting software so that 20 hour, 10 hour etc. can be performed and printed. While it won't be an inexpensive tool, I really need for what I do and want more accuracy than my home built capacity tester.. I hope it will be ready in a few months.

In the mean time I have my own capacity tester but as you mentioned holding current at a constant is tough. Still my tester counts Ah's and if I keep it close I can get a decent approximation.

I plan to spend more time this winter deriving a more accurate Peukert for my bank... This past season I had my batt mon set to 1.01 and it was fairly accurate.


P.S. You may be interested to know that the new "Lead Crystal Batteries" coming out of China have a Peukert of 1.0......

Lead Crystal Battery Discussion (LINK)






.

[OceanPlanet]

Originally Posted by Maine Sail

I am working with a company in California to get a battery discharge tester that holds the load constant and is specifically designed for this.. They build them for Concord battery but are geared towards the Aircraft requirements in terms of reporting. The owner is tweaking the reporting software so that 20 hour, 10 hour etc. can be performed and printed. While it won't be an inexpensive tool, I really need for what I do and want more accuracy than my home built capacity tester.. I hope it will be ready in a few months.

In the mean time I have my own capacity tester but as you mentioned holding current at a constant is tough. Still my tester counts Ah's and if I keep it close I can get a decent approximation.

I plan to spend more time this winter deriving a more accurate Peukert for my bank... This past season I had my batt mon set to 1.01 and it was fairly accurate.

Looking forward to sneaking into Maine Sail's shop and tweaking the current on that tester when he's not looking...

[darylat8750]

So in planning an easy but safe installation.
If I balance:
" Top Balance Made Simple:

Wire in parallel let sit for a day or two or until current stops moving between cells. If you don't have the equipment to test this just give it more time. As voltages converge between the cells the movement of current gets very, very slow....

Now charge the parallel pack to 3.8V = DONE"


Control like this blatantly plagiarized (by me) but beautifully built system :

Quote:
Originally Posted by Maine Sail
BTW This is what I have into my BMS/Battery protection system:


Tyco EV-200 Contactor For LVC = $79.00
House Power BMS Board = $75.00 (Clean Power Auto LLC.)
Cell Modules = $54.00 (Clean Power Auto LLC.)
Solar and Alt HVC relays = $8.60 (Cole Herse 70A)
Bus Bar = $4.80 (Blue Sea)
Switches, Alarm Buzzer etc. $44.00
Terminals = $9.50
Job Box = $13.00
TOTAL = $287.90

That is small potatoes compared to toasting $2000.00 worth of cells due to a mishap..

This was shot during the load and charge bus re-wire and install for LiFePO4...
This image has been resized. Click this bar to view the full image. The original image is sized %1%2.

I already owned a Balmar MC-614 and a fully adjustable solar controller."


I assume (and you know what ass u me does ) that I would need to take the cells to the bottom knee once to make sure each cell is good. Is it necessary to take them to the bottom knee more than once with a different load to figure Purket or is it safe to just use a Purket of 1? Or is it even necessary to take them to the bottom knee even once with the House Power BMS?

[senormechanico]

Originally Posted by Maine Sail

Does anyone know the actual method battery manufacturers use to calculate a Peukert? It seems to be different from the standard calculations? I have tried a number of Peukert calculators, and calculations run out manually, yet I find it very hard to match the manufacturer supplied Peukert?

That said I have capacity/time tested the manufacturer supplied Peukert on a number of occasions and found them to be pretty accurate.....

I suppose this could be another thread entirely....

Probably like this.

[deckofficer]

Using the specs for a 80% DOD, doesn't seem like much of a test. Specs say my batteries are 100 a-hr but we all know these cells give about 15% more than their spec sheet. I took 80 a-hr out on this cycle and these are the cell voltages....
3.26, 3.25, 3.25, 3.25 (13.01)

Quite a ways from the discharge knee, but to be fair to Winston, this will be the cycles for the future tests.