Evaluate battery capacity, again

Question

question

timbo asked May 8, '22

Evaluate battery capacity, again

There were some discussions on battery capacity evaluations but they did not really answer my question/concerns.

I have 16 CALB Lifepo4 prismatic cells 400Ah connected in series to get 48v. So, nominal capacity is about 20kWh. Battery is 7 years old (time flies!). Batteries promised 2000 cycles for 80% DD. This is a good manufacturer and their batteries were tested before to deliver this promise. 2000 cycles is about 7 years. But I never went that far down. My typical DD 25-49%, not more. So, I should expect some life in the battery

I want to expand my total storage capacity. I try to evaluate remaining capacity to understand: is it worth to add a similar battery pack in parallel or I need to build a new system with new batteries.

All methods with "charge to 100% and then discharge to 50% or 20%" don't work for me. They are impractical. I live off-grid from this battery charging from solar. I need energy for the house, and charging is not up to me - it depends on sun. It is not so easy to discharge 10-16kWh in one go, and if I do it then there is a chance I might not be able to charge it back as there is no sun.

Hence, the only way for me is a natural experiment and observations.

I assume that Multi derives SOC based on the voltage, and Multi knows that I use Lifepo4 battery. So, if I collect SOC and consumption numbers from VRM portal, I might be able to get some idea of the remaining capacity.

For example, SOC after sunset s1 = 98.5% and consumption so far is c1 = 4.9kWh. Then I check the same parameters next day before sunrise, s2 = 66% and c2 = (7.61+3.48) = 11.09kWh as there are two days.

Estimated capacity should be (c2 - c1) / (s2 - s1) = (11.09 - 4.9) / (0.985 - 0.66) = 19.0kWh

I tried this approach several times for different SOCs and time periods and I got numbers around 19kWh. If SOC is low or time interval is short, then estimation might be lower, e.g. 18kWh. Sometimes it is higher, e.g. 20kWh. But most time it is very close to 19kWh.

Could you please comment on this approach and reliability of my estimation?

19kWh looks good, it might make sense just to add another 16 cells in parallel and use them for the next 7 years.

Lithium Battery battery capacity

13

2 |3000

Attachments: Up to 8 attachments (including images) can be used with a maximum of 190.8 MiB each and 286.6 MiB total.

Alexandra ♦ commented · May 08, 2022 at 05:16 AM

@timbo

Are you using ess and keep batteries charged before beginning the test cycle?

With the multi you might see it charge up (using its internal battery monitor) even when at 100% on optimized with battery life.

And you would have to do the charge and discharge test without solar attached. But really without a shunt it is clever guesswork.

0 ·

timbo Alexandra ♦ commented · May 08, 2022 at 05:46 AM

No, I don't use ESS. I use a BMS assistant together with the PV inverter assistant (AC-coupled system).

The battery might be charged to 100% or it might not. It depends on the weather. I take numbers only after the sunset to eliminate sun charge effect. Then I have only consumption till next morning.

I take numbers from VRN portal, which gets them from Victron Color Control, which gets them directly from Multi. When all of them show 100% SOC, charging stops. There might be a bit plus or minus as this is a live system, there are always consumers of the energy and sun-inverter is connected. But Multi does a good job in general to balance generation and consumption. Charging stops at 100% SOC.

My main concern is how precise estimations of SOC and consumption I get from Multi. Obviously, there are rounding errors. For example, SOC is measured with 0.5% precision. It has its effect, especially when time difference and SOC difference (s2-s1) are small.

0 ·

snoobler timbo commented · May 08, 2022 at 06:41 AM

MP does not use voltage except to synchronize to 100%. It counts current in and out. It's only as accurate as its programming, i.e., you've programmed it for 400Ah, so it's X% of 400Ah. It in no way knows if your battery is deteriorated.

Because of this, your experimental method will yield very nearly 20kWh all the time.

The only way to reliably determine your battery capacity is to test it with a discharge down to 2.5V/cell following a charge.

0 ·

timbo snoobler commented · May 08, 2022 at 07:55 AM

Thank you. This is very important.

Discharge-charge is impractical or even dangerous in my case. I guess, I should try to calculate my own SOC based on the voltage and discharge curve.

0 ·

nickdb ♦♦ timbo commented · May 08, 2022 at 08:04 AM

It is impractical to use voltage as the measure of charge in lithium batteries.

0 ·

timbo nickdb ♦♦ commented · May 08, 2022 at 08:22 AM

I would really appreciate if you give me a better idea.

I can not gamble with discharge to 2.5 volt. It is like measuring fuel tank capacity on a flying airplane - possible but dangerous. I might not be able to refuel in time.

0 ·

Nev timbo commented · May 08, 2022 at 08:38 AM

I doubt there is a better way. Without a full discharge capacity test you may never know.

If you ever manage to achieve that better way you may just revolutionise LiFePO4 SOC calculation.

Our 4 cell 300Ah Sinopoly battery has survived 8 years of full time travelling in our RV.

I perform an annual capacity test each November. Constant 30A discharge and stop when any cell reaches 2.8V.

Capacity for the last test was 4008Wh - c305Ah. They are holding up well.

The battery both starts the 3.9TD diesel truck and powers our house behind.

I look forward to any easier capacity test method one day.

0 ·

Alexandra ♦ Nev commented · May 08, 2022 at 12:34 PM

@timbo

@Scubadoo has it nailed really.

You would need a known constant load (heating element) until low cell conditions (or to where you would consider end of discharge) to sort of have an idea of where you are at for capacity.

0 ·

timbo Alexandra ♦ commented · May 08, 2022 at 01:02 PM

My entire house sits on these batteries. I don't have a luxury of playing with constant load, discharge to 2.8v, leave battery at rest, charge them back at will.

I understand this method works. I'm ready to believe it is the best method to do the job. But I can not use it, I can not disconnect my house and play with batteries.

So, I need another method - it might be not the best, good enough would be enough for me.

0 ·

timbo Nev commented · May 08, 2022 at 12:37 PM

@Scubadoo could you please confirm if I got your post correctly. You had batteries for 8 years and according to your test they show no degradation at all - 300Ah nominal vs 305Ah estimated by the test.

This is important for my understanding. Thank you.

0 ·

Nev timbo commented · May 08, 2022 at 08:15 PM

The initial battery capacity on the then new battery 8 years ago was in Ah 315.

We have lost a little but we must be doing something right compared to a few others.

Our setup:

0-30°C

Maximums ever.

Alternator 85A Solar 50A Voltage from any source 14.1.

The battery is always full at 14.1V. To 14.6V is another Ah at most.

Absorption for a few minutes at times. "Float" 13.45V.

No cell balancing has proved neccessary since the initial top balance. I occasionally check cell voltages. Maximum deviation observed 32mV.

Used daily to start the truck, often multiple times daily and to power all the gear in the house behind.

Lowest SOC 38%

Typical draw overnight 90-110Ah

The sun or driving fills the battery most days.

Low SOC alarm at 20%. Victron Battery Protect 12.1V Never triggered.

0 ·

timbo Nev commented · May 09, 2022 at 09:21 AM

Thank you. Sinopoly is a good brand. Such low degradation is really encouraging.

0 ·

snoobler timbo commented · May 09, 2022 at 06:01 PM

I'm not arguing that it's impractical or very difficult for you to do this. It's simply how things are. You can't use your battery monitor and voltage as you have done to determine battery capacity with any reliability.

I would assume that any home so critically dependent on this system would include a backup generator. If you don't, you're very brave.

In the thread of "something is better than nothing," a discharge down to the first cell hitting 3V is very nearly 10% SoC. It's more reliable than shallow discharges because it's outside the flat portion of the voltage curve.

0 ·

question