An EIS asset[1] can give you hints to sense which is each one of the problem (from a robust set of possible problems, not, always, the same problem for each one of two cells). Among other things, a common invisible (microscopic) problem, in these tests[2], is the loss[3] of the contact, inside the WE[2], and the other interphases[4] (in the test-cells); it is, mainly, the electronic[3] part of the (WE internal) contact (due to the thermo-mechanical stresses), that oftentimes deteriorate akin re-assembled cells. So, new thermo-mechanical stresses built-in upon cell operation might form (local and moving) micro-bubbles, that act as merged interfuses in the interphases, resulting in such wayward cells' proceeding[6].

1. A descent full EIS (measurements, spectra) is needed, not only a spot (elementary) measurement of the (internal) resistance.

Your Quote(1): Cell: 1 The internal resistance in one cell was 0.3mOhm which is close to the average fresh cell's internal resistance.

2. Your curious quote(2): It can be due to loss of contact inside the cells.

3. For a high sensitivity of morphological contact loss you can adroitly measure in the high(er) freq. bands (>1kHz) of EIS spectra.

4. Effect of interphase percolation on mechanical...: A multiscale approach

https://www.sciencedirect.com/science/article/pii/S000926141500473X

5. Multiscale modeling of interphase in crosslinked epoxy nanocomposites https://www.sciencedirect.com/science/article/pii/S1359836816304383

6. Your quote(3): Cell1: The internal resistance in one cell was 0.3mOhm which is close to the average fresh cell's internal resistance. Cell2: Other cell had infinite internal resistance.

Dear Dr.Arthur,

I assembled the module with LFP cells. I did not make the cells by myself.

Dear Raam,

Lithium iron phosphate batterys could be damaged if overcharged beyond 3.65 volts per cell, or discharged below 2.5 volts per cell. You must incorporate a protection circuitry for avoiding the deep discharge on every cell.

Dear Rafael,

I used BMS with cell module, current sensor and a control unit. It takes care of the balancing and also SOC monitoring and voltage control in each cell.

Lithium metal batteries can withstand 4V2 fully charged. Is your battery management system specialized on LFP batteries? Have you manually monitored the tension levels during the charge discharge cycle? What manufacturer and model is your BMS? Have you got the data sheet?

Dear Rafael,

The BMS is configured for LFP. I did not monitor tension levels during the charge/discharge. BMS is from EMUS v.1.2.1. the data sheet can be found in the below link.

https://www.elektromotus.lt.../Emus_BMS_User_Manual_v.1.0.p...

I dont know whether I am right.

My prediction is that the failed cells might have lower capacity which could have terminated the charge/discharge quickly.

The cell is completely dead with 0V within few cycles. It can be due to loss of contact inside the cells.

But in LFP, the expansion is not more than 9% as far as I know. that's why i am confused.

An EIS asset[1] can give you hints to sense which is each one of the problem (from a robust set of possible problems, not, always, the same problem for each one of two cells). Among other things, a common invisible (microscopic) problem, in these tests[2], is the loss[3] of the contact, inside the WE[2], and the other interphases[4] (in the test-cells); it is, mainly, the electronic[3] part of the (WE internal) contact (due to the thermo-mechanical stresses), that oftentimes deteriorate akin re-assembled cells. So, new thermo-mechanical stresses built-in upon cell operation might form (local and moving) micro-bubbles, that act as merged interfuses in the interphases, resulting in such wayward cells' proceeding[6].

1. A descent full EIS (measurements, spectra) is needed, not only a spot (elementary) measurement of the (internal) resistance.

Your Quote(1): Cell: 1 The internal resistance in one cell was 0.3mOhm which is close to the average fresh cell's internal resistance.

2. Your curious quote(2): It can be due to loss of contact inside the cells.

3. For a high sensitivity of morphological contact loss you can adroitly measure in the high(er) freq. bands (>1kHz) of EIS spectra.

4. Effect of interphase percolation on mechanical...: A multiscale approach

https://www.sciencedirect.com/science/article/pii/S000926141500473X

5. Multiscale modeling of interphase in crosslinked epoxy nanocomposites https://www.sciencedirect.com/science/article/pii/S1359836816304383

6. Your quote(3): Cell1: The internal resistance in one cell was 0.3mOhm which is close to the average fresh cell's internal resistance. Cell2: Other cell had infinite internal resistance.

Thank you Mr. Samaras. It was very detailed and informative. I will check those literatures and try to make impedance measurements.

Dear Raam,

Does every module consist of two cells in series and you cycle them by charging and discharging them by a charge controller where you charge them and discharge them by a constant voltage source? I think you have to monitor the current during the charging and discharging operation in addition to the temperature of the cells to get out the cause of failure.

Best wishes

Dear Mr.Zekry,

Every module consists of four cells connected in series. 10 modules were connected in series altogether. Each module was equipped with passive balancing circuit. It monitors temperature, voltage and current. All the cells were charged in CC/CV charge and CC discharge. The charging was regulated by a CAN charger.

Dear Raam,

welcome,

It is nice that you follow the answer which points out your interest to hear from the others. Let me greet you for this respected behavior.

I would like to divide the problem so get the cause of failure. I would propose to cycle very cell alone if this is possible, then cycle a module of two cells and so fourth. If The cells succeed to withstand cycling without failure, then it is is the interconnection which may be the source of troubles. Is these interconnects are pressure contacts?

If cells fails in the cycling, then it points out that the cells nay have fabrication defects.

Best wishes

Dear Mr.Zekry,

Thank you. Yes, it is possible to charge the cells separately. I charged all the cells to 100SOC before installing it in the module so that the difference in SOC can easily be balanced. I will try to cycle it separately to find out whether the problem is with balancing circuit or the cells.

I apologize for the delay, I was busy.

Indeed, your inquiry deal with two questions. In one hand you are looking for understanding, in general, the causes and chemical processes underlying the bulging of the cells. If you click this link you can find some insight. http://publications.lib.chalmers.se/records/fulltext/202184/202184.pdf

In the other hand your aim is to know what went wrong in your setup or your charge and discharge procedure, so that your cells were damaged early. I can enumerate you some possibilities:

1. Low quality cells. Try another manufacturer, model and/or supplier, as Mr. Zekry pointed out.

2. Review there are no wrong connections, and that settings on the configuration page in the control panel, not overpass the manufacturer maximums ratings.

3. Mechanical stress on cells terminals, if contacts are by pressure, or long term high temperature, if they were soldered, already was suggested as a damage cause by Mr. Samaras. With soldered terminals, a higher power solder will shorten the high temperature span, and it should be safer.

4. Disbalancing on accuracy among current sensors on different cells. (BMS Manual 6.2 : "Recalibration of current sensors")

5. LFP cells balancing must be based upon control of total charge introduced in every cell.( current x time integral). The monitoring of voltage levels doesn´t show the actual SOC of the cell. So your passive balancing circuit could be one origin of the problem.

6. Moreover, you afirm in your penultimate post, you connect four cells in a module with ¿ONE? passive balancing circuit which monitors V, I and T. (3.1.2.) "Emus BMS cell module is a specially designed controller board that mounts directly on the terminals of a SINGLE CELL...". So, the term "module" on the BMS manual means a printed circuit board(pcb), not a cluster of cells. If you install one unique controller board for a group of four cells, it can´t monitor the actual voltage value on every cell, so one of them could be in a overvoltage or undervoltage not detected condition. One "BMS module" is needed for every cell, not for every "module of four cells".

Dear Mr.Villalobos,

First of all, I would like to thank you for taking time to write such a detailed answer for the queries.

Like you said, I will cross check the configuration on the control panel and Termination conditions by testing the single cells.There is a small misinterpretation, may be the word 'module' is misleading. Actually, Emus cell monitor was mounted on each single cell in the pack (serial) and all the single cell monitors were connected to a controller.

I also think that

1.The balancing current is too low for cell of 300Ah so it takes so much time for the cell balancing.

2. I found that there were some bad cells with low capacity and high internal resistance which makes them age quickly.

Thank you for your suggestion and I will recalibrate the accuracy of the current sensor.

"Is there any literature on the influence of mechancial stress and welding/soldering on the performance of prismatic cells?".

Raam, be conscious that voltage is NOT a good indicator for SOC. If you charge the four cells of a module in serie with the same electrical current, and then connect the four cells of the module to your passive balancing circuit, and later you discharge individually every cell, you will obtain surely four different values for the total capacity of every cell. But if you discharge individuadlly every cell directly after charging, without balancing, the total capacity must be more homogeneous among cells, so in LFP cells you must avoid balancing circuits which goal is to equal the voltage in the cells.

As Rafael Villalobos. Also, a balancing circuit is, commonly, optimized for a (near) DC balancing.

However, at high(er)[1] frequencies (kHz to MHz, regimes), their uncomprehending balancing (and their active feadback) might distort[2,3] the nominal function (as a principle), for a more complex spectrum for the "same"[3] (discharge or charge) mode(s).

1. A Switch-Mode Power Supply is an injudicious source for a balancing issue, at high frequencies (MHz).

2. If possible, avoid (minimize), the number and the big and fast current changes (0, iMax). Scratch of the "same" electrical current switching mode(s), during (inside) a discharge or charge phase mode.

3. At high frequencies (MHz). one pf the (in series) connected samples (a single battery) might revise their function, acting as a temporal active Load for the other samples (batteries).

Rafael Villalobos Dear Mr.Villalobos, it does the balancing by using the charge not with respect to Voltage. But I have one more question, how to determine the balancing current needed? how to choose a right passive balancing circuit, if we know only the capacity of the battery. I do not know how much is the difference in capacity without checking each cell assembled in the module.

I agree with Mr. Samaras, discharge regime, average DC current and high frequency components, indeed do influence the actual capacity of the cells, so it's a good (and bulky) practises incorporating low pass filtering.

Raam,

It is not in the fields of my specialty but I can recommend that you contact a professional from CALCE (https://calce.umd.edu/) that specializes in batteries.

I am quite sure that they will be able to direct you and possibly give you insights in your quest.

Best,

Alon

Dear Raam:

1. If you discharge a module of several in series cells and your controller only supervise the voltage of the pack, but not the individual voltage of every cell, the undervoltage condition that disconnect the charging process on the module could be produced, in a not balanced cluster of cells, when the lowest capacity cell has reached a forbidden discharge level and that cell gets damaged. With a balanced group of cells, they evenly decrease their voltage while discharging and noone gets a dangeous low voltage when the undervoltage event of the module is detected, but this has a drawback as the charge in every cell equals the lowest capacity cell, so in every cycle the pack delivers less power than it potential maximum.

2. You can assure all cells have got the same electric charge using an in series charguer, if all cells are previously completely emptied, the charge finish when the lowest capacity cell is full. So you get automatic balancing of all cells.

3. If your charguer is a in paralel model you need calibrate the different paths so that all them gives exactly the same electric current and again disconnect all the cells when the lowest capacity cell gets his maximun voltage level. You need a very stable current source in every track, so that variations among paths on voltage or temperature doesn´t affect the delivered current. This is the second proceeding for autobalancing the cells, but with the same accuracy as the match of the current on the paths.

4. The optimum strategy when you works with lithium iron phosphate cells is to couple a controller with individual voltage monitoring on every cells, so first you can charge every cell to its maximun capacity, an later when discharging and the lowest capacity cell gets empty, the controller first disconnec this cell and then shortcircuit this path so others cells can give all their plenty power but with a lower voltage supply. With this assembling you don´t need at all balanced cells