Charging LiFePo4 in a hybrid system

JJH3rd

I am using a solar/battery energy system that is made up of: 6.2KW of solar panels, 800AH of LiFePo4 batteries (15S2p of CALB 400AH), Solark 8K controller, 10 circuit transfer switch).  Ironically, due to the storm system that came through here yesterday, our home was without grid power for more that 8 hours and the 6 circuits supplied by the solar/battery system provided power through the transfer switch as designed, so for a time I was "off-grid" and on batteries&/solar power.

Here is my question: Can I charge individual batteries if their voltage is below a defined value while the batteries are still connected in the string?

I asked SolArk about charging the bank while the controller was connected to the batteries and they indicated that was not a problem.  In the charge-phase, the SolArk does, in fact charge the entire bank.

My questions are: 1) could I charge an individual cell in situ, 2) would there be any benefit to charging the individual cells?

mcgivor

Charging individual cells, or cell blocks within a string is not something that is usually done, the rationale being that the cells should have been ballanced prior to commissioning and doing so would cause imbalance. 

JJH3rd

The first time I measured, (before they were put into service), the average voltage of the 30 batteries was 3.246, the maximum was 3.28, the minimum was 3.2, StDev was .022.  My concern is that as I measure the batteries, in service, while under charging conditions, one or more measure 3.1 volts and they are not at the first or last position in the string.

Right now it looks like more measurement is in order.

mcgivor

Sounds like they were not ballanced prior to series connection, hopefully you have a BMS to prevent individual cell voltage runaway. When nearing the fully charged or disharged state the cell voltage deviation increases dramatically, without a BMS serious damage could occur.
What charging voltage is being used?

JJH3rd

From the SolArk documentation: 

continuous battery charging output:190A

3 stage with equalization

charging voltage 54.4V

mcgivor

Most users use a lower charge voltage of ~3.400 volts per cell as there is little advantagein attempting  to achieve 100%, it is also considered that  limiting the voltage increases cycle life expectancy. Charging 15 cells at 54.4V represents 3.63 VPC, using this high a setting without a BMS which limits current as the voltage nears this value, would be something not recommend.

The normal practice is to disable equalization as it is not needed, my questions are, do you use a BMS and did you ballance the cells prior to assembly ?

JJH3rd

no ans yes

mcgivor

Without a BMS your charging voltage is too high, IMHO. Without one there is no way to limit the current  and individual cell voltage may exceed 3.650V, is the charging voltage adjustable to a safer voltage say ~51V maximum? 

Raj174

@JJH3rd
I have a setup similar to yours, although my LFP bank is one fourth the size, and I definitely agree with mcgivor. A 51V to 51.4V absorb setting would be the highest safe settings for your bank without a BMS. With that charge setting it should work well if you have already balanced the cells.

JJH3rd

Here's my basic quandary.  When I examine the BMS landscape, I see that they work with the battery bank as a whole.  This approach may be alright when there are a small number of batteries, less than 10 for example.  The number of small voltage LFP batteries needed to get to a workable voltage exceeds that number by a large margin.  In my case, to get to ~48 V & 800 AH, I have 30 batteries (15s2p). I would like to believe that a BMS I would install would provide more features than the Sol Ark controller to which the batteries are attached.

At this point, I would settle for information on each cell, such as voltage and SOC for instance.

I may have to construct an Arduino based monitoring system.

Raj174

Here are two BMS that do cell voltage monitoring with cell and battery protection capability. They do not do active cell balancing. One is a Beaglebone/Raspberry Pie project, information here:

https://github.com/simat/BatteryMonitor

and the other is the Chargery BMS16 (works with 2 to 16 cells in series or series / parallel configurations) available here:

https://www.aliexpress.com/item/32850502583.html

JJH3rd

Thanks for the links

mcgivor

JJH3rd said:

Here's my basic quandary.  When I examine the BMS landscape, I see that they work with the battery bank as a whole.  This approach may be alright when there are a small number of batteries, less than 10 for example.  The number of small voltage LFP batteries needed to get to a workable voltage exceeds that number by a large margin.  In my case, to get to ~48 V & 800 AH, I have 30 batteries (15s2p). I would like to believe that a BMS I would install would provide more features than the Sol Ark controller to which the batteries are attached.

At this point, I would settle for information on each cell, such as voltage and SOC for instance.

I may have to construct an Arduino based monitoring system.

There appears to be a little confusion on your behalf, a BMS looks not only at the total voltage but also individual cell, or cell block voltages. Your battery consists if 15 cell blocks in series, a cell block being 2 cells in parallel, each 400Ah CALB  unit is not a battery, it is a cell.

Attached is a pdf of the BMS I currently use, it is an 8 series but illustrates how it is connected, in your application the only difference would be the addition of 7 cells to meet the required voltage. They are available in many configurations depending on personal needs such as charge/discharge current and voltage. For around  $100 I considered it cheap insurance considering the overall cost of the cells.

JJH3rd

I sent a message to the Chinese manufacturer of the BMS available through IC GOGOGO and then sent links to a how-to-install and also a link to Will Prowse reviewing it:

How-to: https://www.youtube.com/watch?v=im4ggvoni5M

WP review: https://youtu.be/xIBlmvQWbXY

This BMS looks very promising.  I just wonder about which amperage board I would need and its connection cable size.

JJ

JJH3rd

how about this for all 30 cells?

https://www.aliexpress.com/item/4000533823272.html?spm=a2g0o.cart.0.0.4d3c3c00VkYZ2G&mp=1

mcgivor

As you have 2 cells in parallel per cell block you need a 15 S  BMS, the needed information would be maximum charging and dischargeing current. When I ordered mine I supplied all relevant information, the engineering department recommended one based on the figures.

Raj174

As mcivor said, a 15S BMS is called for because the parallel block doubles the amp hour capacity while the voltage remains the same. I'd also recommend increasing the BMS current capacity to better handle surges. Take a look at this one.

https://www.aliexpress.com/item/32997552090.html?spm=2114.12010612.8148356.2.42c43de2I8pAHH

JJH3rd

OK. Here my plan. What are your thoughts?

Above is the existing circuit.

Of course in my case the batteries are 3.2V LFP4 15S.

Is this the correct placement for the BMS boards in the circuit?

mcgivor

Whilst there is nothing wrong with the configuration using 2 strings of 15 cells each with its own BMS there are some considerations that need to be considered. The maximum current would need to be limited to the maximum each BMS can process, for example if 190 A is the maximum both would need to be capable of handling that value to prevent overloading of the BMS  should one disconnect for any reason as LiFePo4 will absorb extremely high current, usually 1C or greater.

The conventional method is to connect  two, or more cells in parallel first then series, this would mean only a single BMS would be required as the multiple cells in parallel  become one cell block. This is how small cells not much larger than an AA cell are configured, often with hundreds in parallel in automotive applications. With larger prysmatic cells it's common to parrallel 100Ah cells in parallel in order to increase capacity  rather than use large 400Ah as it's reputedly a better method. Naturally doing this will require only a single BMS, there may be some perceived logic of redundancy with a dual BMS as shown, but having all the cells as a common battery keeps things simple

.

JJH3rd

Thank you both for your help and information.  I'll have to wait until the Chinese economy gets over the COVID-19 virus.  I'll let you know how it works in practice.

Battery_Man

JJH3rd said:

I am using a solar/battery energy system that is made up of: 6.2KW of solar panels, 800AH of LiFePo4 batteries (15S2p of CALB 400AH), Solark 8K controller, 10 circuit transfer switch).  Ironically, due to the storm system that came through here yesterday, our home was without grid power for more that 8 hours and the 6 circuits supplied by the solar/battery system provided power through the transfer switch as designed, so for a time I was "off-grid" and on batteries&/solar power.

Here is my question: Can I charge individual batteries if their voltage is below a defined value while the batteries are still connected in the string?

I asked SolArk about charging the bank while the controller was connected to the batteries and they indicated that was not a problem.  In the charge-phase, the SolArk does, in fact charge the entire bank.

My questions are: 1) could I charge an individual cell in situ, 2) would there be any benefit to charging the individual cells?

It's ok to charge an individual cell to make it more balanced with all other cells. But for safety issue, you are doing that under system power off.

jwrgorman

just a question about BMS and battery monitors - do you knnow if there are there any devices that expose state of charge from a LiFePO4 battery via a modbus register?