Low SOC better for Lithium

[CroDriver]

I thought that the Lithium batteries in our cars/bikes will last longer if we don't discharge them to a low SOC. ThunderSky and other manufacturers have some graphs that prove this assumption.

Tesla is having a different approach:

The other significant factor that affects calendar aging is the charge state of the battery during storage. At higher charge states cells lose capacity faster. This is a second reason why we have limited our maximum state of charge to 4.15V/cell instead of 4.2V/cell. We also offer the driver the option of charging to only 3.8V/cell (~50 percent) or 4.10V/cell (~90 percent) to further extend calendar life if the full vehicle range is not needed on the next few trips. We advise and encourage a full (4.15V/cell) charge only when it is needed.

OK, Li-ion and LiFePO4 are not the same chemistry but maybe this approach could be better for LiFePO4 too.

I'm sure that Tesla Motors has tested these cells very detailed in a variety of conditions.

 

[Velocipede]

I also often read that.
For winter storage, I put my Ping at 50% SOC and I think many here do the same. (and maintain charge every month - 2 times during winter - so I don't have a dead battery in the spring).

 

[cell_man]

Hi Mate,

I read that the Chevy Volt which used A123 prismatics :wink: limited the DOD to something like 85% and also kept the maximum charge level low. I really should have the article in my favourites. Using this method they were able to achieve lifecycles of 10-15years in less than ideal environments and even longer in moderate climates. Obviously these can only be predictions based on rigorous testing. I must do some checks to compare the capacity of a cell charged to 3.60V as opposed to 3.65 and different discharge voltages. The best way to determine the DOD is just to log the energy used as a Cycle Analyst would do.

However, the biggest thing to effect the lifecycle is temperature AFAIK, attached is a doc which compares Li Po and A123 cells and it clearly shows that excessive temperature is the biggest killer of cells. To complicate things further, the internal resistance reduces with temperature and that is why you will generally see the discharge characteristics for cells at very high discharge levels shows the voltage actually increase as it discharges, rather than steadily drop as is normal for lower discharge levels. It's also why in race EV they often heat the cells before a run. IMO there are likely internal parts of the cell that are damaged beyond a certain tempertaure but there are other parts that maybe work better at those elevated temperatures. In an ideal world it would be best if the cells can be maintained at a specific, elevated temperature and kept within a certain temperature window, but not so easy to achieve and cell surface temperature can only ever give you an approximation of the internal temperatures. At least for the cells I'm supplying they are very thin so lend themselves quite well to cooling, or heating due to the large surface area and IMO the differences in internal temperature and surface temperature will be smaller than for a cylindrical cell or a thicker cell. Probably why they were made so thin

 

[fechter]

There is quite a bit of literature on this for cobalt based cells. I think Lipo and LiFePO4 will have similar characteristics. Calendar life is adversely affected by time spent at a high state of charge, and it gets really significant at or near the charge cutoff voltage. A lower cutoff voltage will help there. Temperature is very important also. They don't like heat. Above a certain temperature, damage can be immediate. Otherwise, the time weighted average temperature will greatly affect calendar life. Overheating due to excessive discharge rate can easily push the innards of the cell into the instant damage range.

Cells can be immediately damaged by overdischarging them as well and using the full capacity (deep cycle) is bad for the cycle life.

Ideally you'd want to store cells around 50% soc if you were going to leave them for a long time. Of couse you'd need to charge before you go anywhere.

 

[liveforphysics]

This isn't LiFePO4, but LiMn based LiPoly.

Characteristics of cycle life based on the Depth of Discharge (D.O.D)
Items Criteria Calculated Cycle Life Description Remarks
DOD100%¹ Until 80% of Initial Discharge Capacity >1,400 cycles - Based on real test
DOD80%² >2,500 cycles Fitting function : y = -(0.98E-7)x2-(5.46E-3)x+100 Expected Cycle times
DOD20%³ >11,500 cycles Fitting function : y = -(1.697E-3)x +100.07 Expected Cycle Times

 

[liveforphysics]

Found a few things with LiFePO4, but it's for 1C 100% DOD only, not a SOC compare.

Effects of LiFePO4 and temp:

Here is more good stuff, but sadly for LiCo not LiFePO4.

" The practical voltage limits are a consequence of the onset of undesirable chemical reactions which take place beyond the safe working range. Once all the active chemicals have been transformed into the composition associated with a fully charged cell, forcing more electrical energy into the cell will cause it to heat up and to initiate further unwanted reactions between the chemical components breaking them down into forms which can not be recombined. Thus attempting to charge a cell above its upper voltage limit can produce irreversible chemical reactions which can damage the cell. The increase in temperature and pressure which accompanies these events if uncontrolled could lead to rupture or explosion of the cell and the release of dangerous chemicals or fire. Similarly, discharging a cell below its recommended lower voltage limit can also result in permanent, though less dangerous, damage due to adverse chemical reactions between the active chemicals. Protection circuits are designed to keep the cell well within its recommended working range with limits set to include a safety margin. This is discussed in more detail in the section on Protection . Cycle life estimations normally assume that the cells will only be used within their specified operating limits, however this is not always the case in practice and while straying over the limits for short periods or by a minor margin will not generally cause the immediate destruction of the cell, its cycle life will most likely be affected.

For example continuously over-discharging NiMH cells by 0.2 V can result in a 40 percent loss of cycle life; and 0.3 V over-discharge of lithium-ion chemistry can result in 66 percent loss of capacity. Testing has shown that overcharging lithium cells by 0.1 V or 0.25 volts will not result in safety issues but can reduce cycle life by up to 80 percent."

 

[cell_man]

Thanks for the great info Luke.

What in your opinion would be a good charging voltage and a good DOD %age? I would have thought that a simple LVC is not really enough to ensure a particular DOD as cell voltage is heavily reliant upon load so IMO a LVC of say 2.0V to protect against excessive load and discharge as well as some kind of intelligent monityoring device like a CA to indicate remaining charge would be the best combination to ensure long cycle life. A simple circuit that could be calibrated for any pack size, that then gave you an LED indication or LCD display of remaining charge %age would be nice. There is a model of ebike that has exactly that.

For temperature I'm thinking a max of 60deg C or maybe 55deg C would be better combined with some cooling/heating method for high discharge levels. I think with these bases covered you could get the vast majority of the available performance from any particular cell and yet still stay within the safe boundaries and have a very long cycle life.

 

[liveforphysics]

For many lithium cells, keeping with the 80% of capacity bounds seems to bring a lifespan extension somewhere in the 5 to 10times cycle number increase area vs using 100% DOD. The other LiFePO4 nano-phosphate cells have proven to show very minimal loss of performance or capacity after very high cycle numbers when used in an 80% DOD range. With A123 claiming very significant cycle life increases from the older 26650 cells (which all ready had perhaps the very best life cycle numbers of any cells out there), I think we can reasonably expect to see a least a few thousand cycles of high current use. All info shows the shallower the DOD the greater the cycle life, often showing a logarithmic trend between DOD and cycle life. I've also found some research from LiCo based cells that showed the optimal method for extending cycle with an 80% DOD was to take 10% off the top in a lower HVC, and 10% off the bottom in a higher LVC. I had once thought taking all 20% from the top was most optimal, but I was wrong.

I have a cell of yours that's in the 140 cycle's range now just sitting on the cell cycling charger for weeks. It's still slowly gaining capacity. lol


Of course the trick with any cell is to keep the resting voltage always between the HVC and LVC points.

The voltage under load does not matter for chemical cell damage issues, only for thermal issues. So, if you're pulling the cells well below your resting voltage LVC under high loading, the chemistry inside the cell is not adversely effected. It's only damaged when the resting voltage, which is the only external indicator of the chemical SOC of the cell goes below the safe LVC point.

This makes it darn tough to do LVC's on high current packs, and I think your idea of measuring the capacity used sounds like a great idea.

 

[cell_man]

For many lithium cells, keeping with the 80% of capacity bounds seems to bring a lifespan extension somewhere in the 5 to 10times cycle number increase area vs using 100% DOD. The other LiFePO4 nano-phosphate cells have proven to show very minimal loss of performance or capacity after very high cycle numbers when used in an 80% DOD range. With A123 claiming very significant cycle life increases from the older 26650 cells (which all ready had perhaps the very best life cycle numbers of any cells out there), I think we can reasonably expect to see a least a few thousand cycles of high current use. All info shows the shallower the DOD the greater the cycle life, often showing a logarithmic trend between DOD and cycle life. I've also found some research from LiCo based cells that showed the optimal method for extending cycle with an 80% DOD was to take 10% off the top in a lower HVC, and 10% off the bottom in a higher LVC. I had once thought taking all 20% from the top was most optimal, but I was wrong.

I have a cell of yours that's in the 140 cycle's range now just sitting on the cell cycling charger for weeks. It's still slowly gainingg capacity. lol

Thanks luke and the last comment is pretty encouraging So what would you think is a good way to achieve the 10% off the HVC?

I wish I had a bit more time to do some more testing and researching but unfortunately I'm spending too much time cutting up bits of board and looking for nice enclosures.... Drop me a line sometime I'd like to have an MSN or Skype chat to hear where you are up to with your tests etc

 

[CroDriver]

Hi Luke

The voltage under load does not matter for chemical cell damage issues, only for thermal issues.

Are you sure about that?

I'm abusing my ThunderSky's at over 10C for several seconds at very cold temperatures (below 0 celsius) and the voltage drops well below 2V. Yesterday I even saw sam sparking out of the cell tabs (don't know how this could have happen - all screws seem to be tighten) and one cell is making noise when I push the throttle hard - sounds like it's venting a little.

 

[liveforphysics]

Hi Luke

The voltage under load does not matter for chemical cell damage issues, only for thermal issues.

Are you sure about that?

I'm abusing my ThunderSky's at over 10C for several seconds at very cold temperatures (below 0 celsius) and the voltage drops well below 2V. Yesterday I even saw sam sparking out of the cell tabs (don't know how this could have happen - all screws seem to be tighten) and one cell is making noise when I push the throttle hard - sounds like it's venting a little.

Boiling the plates from Ri induced heating on the surfaces of the current collectors. That's going to F-them up, but it's not F-ing them up because of non-reversible chemical reactions happening from too low of voltage. The sparks thing sure seems like there must be a wimpy connection somewhere in the group though doesn't it? If you were to pull some hard current through the pack, and then check the temp of each strap with an IR-temp gun, I bet somewhere you've got a poor connection. Weird things happen sometimes. You could have the bolt all properly torqued and looking perfect from the top, and yet a couple grains of sand or fibers, clear enamel or a zillion other things could be in between the surfaces causing a high resistance connection. Weird stuff happens like that sometimes when you've got loads and loads of connections.

 

[fechter]

Here's an interesting article: http://www.che.sc.edu/faculty/popov/drbnp/WebSite/MSA-calendar.pdf

Boils down to this graph:


As you can see, keeping the cells on float (voltage constantly applied) is worse than letting them sit open circuit. The effect of temperature is also shown.