LFP SOC vs Pack Voltage

How do you store and manage your electricity?
Post Reply
Paul9
Groupie
Posts: 269
Joined: Mon, 20 Apr 2009, 02:10
Real Name: Paul Coggiola
Location: Sydney

LFP SOC vs Pack Voltage

Post by Paul9 » Mon, 22 Jul 2019, 21:09

My question deals with estimating a battery bank’s State of Charge (SOC) from the voltage (OCV) of that bank. I understand that the difficulty lies with the relationship between SOC and voltage in lithium batteries being very non linear. Going back maybe 8 to 10 years this topic was discussed. At that time, graphs and tables were put together giving a rough relationship between cell voltage (for a 3.2v nominal lithium cell) and SOC. I remember a graph of SOC vs OCV but try as I might I cannot find the thing! I have managed to find a table which the author said came from Weber and Coulomb’s MX5 thread. It is only brief so I’ll put it here:

OCV SOC%
3.25 25%
3.28 40%
3.30 60%
3.33 80%
3.35 90%

I do remember that the graph I cannot seem to find lined up ROUGHLY with the above figures. The graph also showed that SOC vs OCV did not vary much at all at reasonable operating temperatures.

I am hoping that someone remembers the graph I am talking about and would appreciate them posting it to this thread. What I would like to do is combine the graph and the table above into one graph which could give me a range of SOC’s for an OCV?

As the graphs/tables I am talking about were from years ago, when lithiums were considerably “newer” than they are now, could any other tables or graphs of more recent origin also be posted to this thread so I can maybe combine them all into one graph?

Now to explain why this SOC vs OCV thingo is a little bit important to me. I regularly do a trip of 44.1 kms. This trip is about every week and, ok, I’ll admit it, it is to my girlfriend’s place. What I need to know is can I get back to my place if, say, my charger konks out and I can’t recharge at her place (has happened!). Can I get back to my place if I say something I shouldn’t have said and I get kicked out (hasn’t happened but has got close!). We men never get anything right!

I have a battery bank consisting of 40 x 3.2v x 100ah CALBs. I have kept a couple (well 4 actually) trip records which show the following:

AH used during trip averages 32.2 AH and varies between 31.9AH and 32.7AH
SOC as per the TBS Pro (1 hour after arriving at her place) averages 66% and varies between 65.5% and 66.5%
Per cell voltage (1 hour after arriving) averages 3.2925/cell and varies between 3.28v and 3.3125v

The issue is that an average cell voltage of 3.2925 indicates a SOC, as per the table above, of just over 40%. The SOC as per the TBS Pro shows a SOC of 66%. The difference between 40% and 66% is a fair bit! If 66% is correct then me getting kicked out (whether or not I said/did or did/didn’t do the wrong thing) means I can get home without a recharge. If the actual SOC is just over 40% (as per cell voltage in table) I need to find a charging station where I can both charge the car and lick my wounds!

All assistance, advice, and posting of tables and graphs will be appreciated.

Thanks
Paul
Last edited by weber on Tue, 23 Jul 2019, 14:24, edited 1 time in total.

User avatar
jonescg
Senior Member
Posts: 3043
Joined: Thu, 21 Jan 2010, 23:05
Real Name: Chris Jones
Location: Perth, WA.

Re: LFP SOC vs Pack Voltage

Post by jonescg » Mon, 22 Jul 2019, 21:14

These are no-load resting voltage right? It's a very rough guide, but yes, that's pretty close.
If you load the cells up they will flatten out at 3.2 V per cell and more or less stay there till they fall off a cliff at 3.1 V.
AEVA National Secretary, WA branch chair.

Paul9
Groupie
Posts: 269
Joined: Mon, 20 Apr 2009, 02:10
Real Name: Paul Coggiola
Location: Sydney

Re: LFP SOC vs Pack Voltage

Post by Paul9 » Mon, 22 Jul 2019, 21:48

Thanks jonescg.

When you say these are no load resting voltages - yes that is correct.

When you say "that's pretty close" do you mean the cell voltages as per the table are pretty close to the SOC's I have determined or that the TBS Pro's SOC would be pretty close to the actual SOC? I am trying to determine if the SOC's I have deduced from the cell voltages are accurate or the TBS Pro's SOC are accurate because they are wildly different figures!

Thanks

User avatar
brendon_m
Groupie
Posts: 367
Joined: Sat, 28 Oct 2017, 11:00
Real Name: Brendon McCarrol

Re: LFP SOC vs Pack Voltage

Post by brendon_m » Mon, 22 Jul 2019, 22:21

Not that it answers your question and all scenarios are different but I have a little diahatsu that has 44 x 3.2v 100ah cells that are a bit rough and I've managed 98km on a charge and most of the drive was at 90/100kmh with a/c but that was absolutely scraping the bottom of the barrel.
I generally go with 60km being my limit, but 80km is OK and I can do 100km in an emergency.
If you know the efficiency of your charger you could use a 240v power meter and see how much power you are using to recharge and then what is getting to the pack

antiscab
Senior Member
Posts: 2490
Joined: Mon, 26 Nov 2007, 05:39
Real Name: Matthew Lacey
Location: Perth, WA

Re: LFP SOC vs Pack Voltage

Post by antiscab » Mon, 22 Jul 2019, 23:09

The TBS pro Ah counts to get the soc value. Assuming the battery capacity programmed into it is accurate, the soc reported will be far more accurate than trying to work it out by voltage.
Matt
2017 Renault zoe - 25'000km
2007 vectrix - 156'000km
1998 prius - needs Batt
1999 Prius - needs batt
2000 prius - has 200 x headway 38120 cells

User avatar
weber
Site Admin
Posts: 2626
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Re: LFP SOC vs Pack Voltage

Post by weber » Tue, 23 Jul 2019, 10:13

Hi Paul9,

I note that what you say about the difficulty of estimating SoC from OCV does not apply to all lithium-ion chemistries. It's only LFP that has this problem. And it's not because it's non-linear, it's because it's flat.

Here are the graphs you want:

See figure 2 on page 3 of:
http://www-personal.umich.edu/~hpeng/DSCC2013_Weng.pdf
See the graph on the left of page 15 of:
LiFePO4 voltage vs SoC.pdf
(4.21 MiB) Downloaded 36 times

This paper shows that the voltage is unaffected by temperature in the range -10 °C to 40 °C:
https://www.bestgobattery.com/technolog ... -test.html

This post tells how to account for internal resistance to estimate OCV in normal operation:
https://forums.aeva.asn.au/viewtopic.php?p=59994#p59994
And this thread talks about accounting for hysteresis:
https://forums.aeva.asn.au/viewtopic.php?f=31&t=4866
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).

User avatar
Richo
Senior Member
Posts: 3737
Joined: Mon, 16 Jun 2008, 00:19
Real Name: Richard
Location: Perth, WA

Re: LFP SOC vs Pack Voltage

Post by Richo » Tue, 23 Jul 2019, 12:32

weber wrote:
Tue, 23 Jul 2019, 10:13
... it's because it's flat.
Yep I agree determining SoC from OCV is a waste of time.
Use a current integrator thingo...
So the short answer is NO but the long answer is YES.
Help prevent road rage - get outta my way!

T1 Terry
Senior Member
Posts: 993
Joined: Thu, 30 Sep 2010, 20:11
Real Name: Terry Covill
Location: Mannum SA

Re: LFP SOC vs Pack Voltage

Post by T1 Terry » Tue, 23 Jul 2019, 15:34

The true 100% SOC rested voltage of a cell is greater than 3.45v after a few hrs or a few days, as long as you don't have those boards across the tops of the cells that continually draw current. A fully charged cell is 3.40v, but it requires a min of 0.05v to put further charge into the cell. This means if the cell voltage is 3.45v to 3.6v, there is a surface charge over and above what the cell can accept, it is full. Any voltage below 3.4v means the cell can still accept more charge.
If you charge the cells to 3.6v and have no parasitic loads on the cell and it won't hold 3.45v over night, it just isn't full yet and probably never has been full or needs a "memory charge" clear to allow the cell to absorb all the available current until saturated and only then will the voltage climb above the 3.45v mark.

T1 Terry
Green but want to learn

Paul9
Groupie
Posts: 269
Joined: Mon, 20 Apr 2009, 02:10
Real Name: Paul Coggiola
Location: Sydney

Re: LFP SOC vs Pack Voltage

Post by Paul9 » Thu, 25 Jul 2019, 13:09

Thanks guys much appreciated!

I have read your posts, and the references in the posts, and the references in the references in the posts, and obviously far greater minds than mine have grappled with the problem.

brendon_m, your diahatsu and my suzuki swift would be similar weights and sizes I guess. Your appear happy with 80km range and I have once done 81.2km. Thanks for the info.

Thanks antiscab, I have a tendency to believe my TBS Pro as it's AH used on a trip is very similar to the AH used according to EV power BMS readout for the same trip. According to both instruments I should be able to do the whole 88km trip and have a little left up my sleeve.

weber, thanks for the references. I printed out the graphs and they all seem to vary slightly. I am not going to get hysterical about hysteresis or variations due to temperature. Unfortunately the graphs vary the most at exactly the SOC and OCV values that concern me the most. The various graphs seem to have OCV "falling off a cliff" as jonescg said at very differing OCV's.

Thanks richo I will google "current integrator thingos" though I doubt I would know how to use one if I could find one.

T1 Terry thanks for the explanation. My charger charges my pack to 141v (3.525v per cell) and the pack voltage then gradually settles to about 133v (3.325v per cell) overnight. Maybe I am not charging right up to 100%?

Thanks again
Paul

User avatar
weber
Site Admin
Posts: 2626
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Re: LFP SOC vs Pack Voltage

Post by weber » Thu, 25 Jul 2019, 14:46

Paul9 wrote:
Thu, 25 Jul 2019, 13:09
Unfortunately the graphs vary the most at exactly the SOC and OCV values that concern me the most. The various graphs seem to have OCV "falling off a cliff" as jonescg said at very differing OCV's.
I suspect you mean "at very differing SoCs". Well spotted. The reason is that it is somewhat arbitrary what OCVs you call 0% and 100%. Different researchers have made different choices. I think you will find that the curves agree with each other, if given a small horizontal offset and scaling.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).

Paul9
Groupie
Posts: 269
Joined: Mon, 20 Apr 2009, 02:10
Real Name: Paul Coggiola
Location: Sydney

Re: LFP SOC vs Pack Voltage

Post by Paul9 » Thu, 25 Jul 2019, 16:53

Thanks weber, my bad!

User avatar
weber
Site Admin
Posts: 2626
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Re: LFP SOC vs Pack Voltage

Post by weber » Sat, 27 Jul 2019, 08:27

Terry, Although I disagree that there any such thing as a true 100% rested voltage, I agree that 3.45 V (when the last thing it did was gently charge) is an excellent choice. It must be, since it's what Coulomb and I use. :D And I agree that this corresponds to about 3.40 V when the last thing it did was gently discharge.

3.45 V is a good choice because it's about as low as it can be while still allowing you to be certain (despite possible measurement errors) that you're not merely somewhere on the upper plateau between 80% and 95%, and hence allowing you to properly balance the cells. It's good to be as low as possible because the cells age more rapidly at higher voltage.

We choose 3.15 V as our 0% rested voltage (when the last thing it did was gently discharge) because it gives our BMS a simple formula to resync our coulomb counter when the battery is below 30% (not only when it is at 100%). If the estimated open-circuit voltage (compensated for internal resistance times current) is less than 3.25 V per cell, and the current is between zero and -C/16 (a gentle discharge), and these conditions have existed continuously for more than 10 minutes, then we calculate

SoC = (OCV - 3.15 V) × 0.3%/mV (for OCV ≤ 3.25 V, on discharge)

and set the coulomb counter to that. So when we treat 3.15 V as 0%, we get 3.16 V as 3%, 3.17 V as 6%, 3.18 V as 9% etc. up to 3.25 V as 30%. You can see this nicely linear region on those published curves above.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).

User avatar
weber
Site Admin
Posts: 2626
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Re: LFP SOC vs Pack Voltage

Post by weber » Mon, 29 Jul 2019, 13:17

I had a question by email asking "What's special about 30% SoC?" which I choose to answer instead as "What's special about 3.25 V rested?"

The answer is analogous to what I wrote about 3.45 V (100%) above.

3.25 V (when the last thing it did was gently discharge) is a good choice because it's about as high as it can be while still allowing you to be certain (despite possible measurement errors) that you're not merely somewhere on the lower plateau between 70% and 40%.

When you look at the curves in the papers, on the discharge side of the hysteresis loop, you'll see a fairly linear (and neither flat nor precipitous) region between about 3.28 V and 3.18 V, and if you extend that region's line-of-best-fit to the vertical axis (SoC = 0%) you'll see it hits at about 3.15 V. And you'll see that both authors have chosen 0% and 100% voltages such that 3.25 V is approximately 30%. So my formula is justified.

The only region where my formula gives answers that disagree significantly with the charts, is for voltages below 3.18 V (SoCs below 9%). But it errs on the side of caution.

Rested   Formula  Chart
Voltage  SoC      Soc
-------------------------
3.18     9%       9%
3.17     6%       8.7%
3.16     3%       8.4%
3.15     0%       8.1%

So if our BMS disconnects the battery at 3% SoC according to the formula, it's really disconnecting it at 8.4% SoC according to the charts.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).

Post Reply