State of Charge

Started by zoneblue, October 28, 2013, 04:39:57 PM

Previous topic - Next topic

zoneblue

With WBJr comes SOC modelling. It doesnt take a brain surgeon to realise that quantifying SOC is pretty complex.  The first two results pages for google:algorithm battery state of charge amps counting, should be enough to squash any illusions to the contrary.

http://www.smartgauge.co.uk/sgvahrs.html and http://www.mpoweruk.com/soc.htm as examples both give a persuasuve enough argument  to adopt a hybrid model involving both voltage, and amp counting, and self learning.   

But i have no idea where to start. Are there any published/open algorithims out there? Do we have to start from scratch?
6x300W CSUN, ground mount, CL150Lite, 2V/400AhToyo AGM,  Outback VFX3024E, Steca Solarix PL1100
http://www.zoneblue.org/cms/page.php?view=off-grid-solar

TomW

Zoneblue;

Hmm. I suspect a person(s) could make an entire field of study for decades on this subject and maybe never find that perfect equation to predict / solve it.

I don't pretend to understand it and the more I read the more lost I got when I was trying to understand how to predict the actual S.O.C. without a sample of electrolyte and even then not sure you can totally rely on the numbers except in narrow, specific circumstances? Certainly tough on a system like mine that is never static but always under a charge or load and perhaps both simultaneously.

Interesting to watch.

Tom
Do NOT mistake me for any kind of "expert".

( ͡° ͜ʖ ͡°)


24 Trina 310 watt modules, SMA SunnyBoy 7.7 KW Grid Tie inverter.

I thought that they were angels, but much to my surprise, We climbed aboard their starship and headed for the skies

Westbranch

ZB said: adopt a hybrid model involving both voltage, and amp counting, and self learning. 

I think voltage at start of charge cycle and time to achieve desired end voltage/amps may help also.
KID FW1811 560W >C&D 24V 900Ah AGM
CL150 29032 FW V.2126-NW2097-GP2133 175A E-Panel WBjr, 3Px4s 140W > 24V 900Ah AGM,
2 Cisco WRT54GL i/c DD-WRT Rtr, NetGr DS104Hub
Cotek ST1500 Inv  want a 24V  ROSIE Inverter
OmniCharge3024  Eu1/2/3000iGens
West Chilcotin 1680+W to come

boB

In theory, it should be fairly simple to calculate SOC.

You have to know the actual amp-hour capacity of the batteries, the amp-hour efficiency (around 95%)
and when it is fully charged.  It ~should~ be fully charged when it goes to Float.  You ~could~ just
make the Absorb time longer than necessary to make SURE it is charged before going to Float and/or
measure specific gravity etc. or learn or know the Ending Amps.
Then, as the amp-hours come out of the battery, it is simple to tell what it's SOC is based on that...
Assuming a Peukert correction isn't necessary due to ~heavy~ discharging during that time.

Then, as it is charging, if the amp-hour efficiency is right on and the charging is done at the
proper "C" rating, it should be as simple as using that efficiency to tell how close you are to 100% full.
Then you could just go to Float at that time.  No Ending Amps or timer needed.

It's never that simple though, is it ?  This is where constant calibration is needed at transition to
Float using the normal timing or End Amps means... At least every few charge cycles.
I'm pretty sure that this is why you sometimes see a jump from 99% or something like that to 100% with
some  battery SOC meters.

The best reference is probably the Trimetric. There is some recent talk/research about using
Kalman filtering for battery SOC estimation.

Getting within a few percent of real SOC is probably pretty good I would think.

But testing of the actual amp-hour capacity will have to be done at least once to get an idea of
what the batteries are actually good for.  Also, this should be done once every so often because
of aging.  Then, the efficiency might also be found at those measurement times and the meter
updated accordingly.

All this sounds like too much fun, doesn't it ?

boB
K7IQ 🌛  He/She/Me

zoneblue

Quote from: boB on October 29, 2013, 01:49:26 AM
In theory, it should be fairly simple to calculate SOC. You have to know the actual amp-hour capacity of the batteries, the amp-hour efficiency (around 95%) and when it is fully charged.  It ~should~ be fully charged when it goes to Float. 
It's never that simple though, is it ?  This is where constant calibration is needed at transition to
Float using the normal timing or End Amps means... At least every few charge cycles.

I dont know if you know Northguy from NAWs forum, i asked how he does it, here:

http://www.wind-sun.com/ForumVB/showthread.php?21012-Battery-manager-and-generator-start/page4

His theory is that Ah in/Ah out  is 100% efficient, with the primary exception of absorb stage hydrolysis (self discharge being the other factor).

He also debates the topic at some length here:
http://www.wind-sun.com/ForumVB/showthread.php?20759-Internal-battery-function-%28from-another-thread%29

Frankly im still having a hard time understanding it, but the basic idea to not to try to compute SOC rather just the negative Ah from a known charged state. Seems to be a very good starting place. Its how to account for the bubbling stage that i need to go back and reread NG's post.

Quote
The best reference is probably the Trimetric. There is some recent talk/research about using
Kalman filtering for battery SOC estimation.

Yes interesting stuff!

Quote
Getting within a few percent of real SOC is probably pretty good I would think.

But testing of the actual amp-hour capacity will have to be done at least once to get an idea of
what the batteries are actually good for.  Also, this should be done once every so often because
of aging.  Then, the efficiency might also be found at those measurement times and the meter
updated accordingly.

I wonder if thats where internal resistance calcualtions could come in? Internal resistance reportedly increases reasonably linearly with age, and is realtively easy to calculate based on voltage sag with known currents. I forget the reference on that just now, one of the above refs i think.

Quote
All this sounds like too much fun, doesn't it ?
boB
6x300W CSUN, ground mount, CL150Lite, 2V/400AhToyo AGM,  Outback VFX3024E, Steca Solarix PL1100
http://www.zoneblue.org/cms/page.php?view=off-grid-solar

boB

Quote from: zoneblue on October 30, 2013, 04:57:58 PM

I dont know if you know Northguy from NAWs forum, i asked how he does it, here:

http://www.wind-sun.com/ForumVB/showthread.php?21012-Battery-manager-and-generator-start/page4

His theory is that Ah in/Ah out  is 100% efficient, with the primary exception of absorb stage hydrolysis (self discharge being the other factor).

He also debates the topic at some length here:
http://www.wind-sun.com/ForumVB/showthread.php?20759-Internal-battery-function-%28from-another-thread%29

Frankly im still having a hard time understanding it, but the basic idea to not to try to compute SOC rather just the negative Ah from a known charged state. Seems to be a very good starting place. Its how to account for the bubbling stage that i need to go back and reread NG's post.


I think I may have seen that thread before, at least the first part. I regard Northguy as a very bright guy.

I don't think I agree with almost anything electrical or electro-chemical being
100% efficient though.  Battery amp-hour efficiency may get  close to that at
times though.  If it were 100% (no exceptions) then SOC would be easy stuff.

I ~think~ that what might be missed there is mainly I^R losses.

And what discharge and charging C rate is that at ?  That will make a difference.

And  you can measure battery capacity.  You do have to get the
battery to almost 0% by measuring its resting voltage.  Deep cycle batteries are
made to be discharged all the way but you don't get nearly as many cycles
so you don't want to measure capacity very often because of that.
Just be careful when charging back up to make sure each battery's voltage
is as equal as possible.

But, battery capacity changes over the lifetime of a battery.  Some will get
better capacity with the first charge cycles, level off for a while and then
decline with more aging and A-Hour ins and outs.

If you do not know battery capacity, then you can only know when it is
at 100% SOC,  but not percentage of its actual capacity below that 100%
You might get lucky though and the batteries might actually have a similar
capacity to what their rating is ?


The Trimetric has pretty much been the standard in the industry and it
starts out at 94% A-Hr efficiency by default.  I wonder why ?
Ralph at Bogart (Trimetric) and others have been making SOC meters
successfully for many many years now and I believe they have been
through this all quite a bit.

There appears to be a LOT of great dialog in those threads, 100% efficiency
or not.  These guys bring up a lot of good brain exercise !

Thanks.

boB
K7IQ 🌛  He/She/Me

zoneblue

The take away for me was that rather than justing count amp hours in and out (using whatever efficiency), instead taking into account the charge stage/bubbling voltage.  Once the bubbling volltage is reached a higher portion of the amps in are wasted. If you just round the overall efficiency to 95%, then wont you get low error if the bank spends lots of time in bulk, and high error if it spends alot of time in absorb? I suppose if a bank averages a certain cycle routine, then an average efficiency might work. Dunno.

Quote from: boB on October 31, 2013, 02:27:26 AM
And  you can measure battery capacity.  You do have to get the
battery to almost 0% by measuring its resting voltage.  Deep cycle batteries are
made to be discharged all the way but you don't get nearly as many cycles
so you don't want to measure capacity very often because of that.

Quite. However a battery monitor cant really load cycle the bank for you, only observe the cycles as they occur and learn from the IV data it accumulates. I guess thats the first step to define and collect cycle data.

Quote
If you do not know battery capacity, then you can only know when it is
at 100% SOC,  but not percentage of its actual capacity below that 100%
You might get lucky though and the batteries might actually have a similar
capacity to what their rating is ?

Yeah i agree. For now i think just specifying 'SOC' as  minus Ah  from full, and worry about the actual capacity thing later. Much later!

Quote
The Trimetric has pretty much been the standard in the industry and it
starts out at 94% A-Hr efficiency by default.  I wonder why ?
Ralph at Bogart (Trimetric) and others have been making SOC meters
successfully for many many years now and I believe they have been
through this all quite a bit.

The question is, will he talk? Its my observation that the engineering world is a more closed shop than the coding world. Maybe we can soften him up over a few beers.
6x300W CSUN, ground mount, CL150Lite, 2V/400AhToyo AGM,  Outback VFX3024E, Steca Solarix PL1100
http://www.zoneblue.org/cms/page.php?view=off-grid-solar

mtdoc

#7
Great discussion!

One thing that I have found very helpful in monitoring how well my current battery meter (Outback Flexnet DC) is accurately reporting SOC is by monitoring end amps and how that correlates with reported SOC.

After a discharge - by using my current monitoring software (Wattplot) I can watch the end amp curve during charging and see exactly when it becomes asymptotic. If that is correlating with when the FNDC is showing batteries going back to 100% then I am  reassured that it is tracking SOC pretty accurately.

Since I have AGMs, the charge/discharge efficiency is pretty high and by playing with different charge/discharge efficiency numbers entered into the FNDC I can see how this changes the tracking of SOC.  The FNDC resets to 100% SOC regardless when the programmed "charge parameters" (end amps and voltage) are reached.  If that is jumping up from a much lower number,  or if the FNDC is showing 100% SOC long before the end amps curve becomes asymptotic, then I know that my parameters (charge/discharge efficiency or battery capacity) were not entered correctly into the FNDC.

Obviously if one has flooded batteries - using SG to monitor SOC can be a very good way of seeing if your battery monitor is accurately displaying SOC.

Not sure that I understand why charge efficiency would be better during bulk versus absorb. As far as I'm aware the charge/discharge efficiency number (eg 95%) is taking into account all losses on the round trip in and out of the battery bank.
In any case, it seems that what is important from a system monitoring/battery health point of view is that the overall effiiciency is correct to allow one to correctly track SOC.



Array 1: Sanyo HIT225 X 8 on Wattsun tracker. Array 2: Evergreen ES-E-225 X 12 on shed roof. Midnite e-panel with Outback GVFX3648, FNDC and Classic 150 X 2. 436 AH AGMs. Honda eu2000i X 2.

zoneblue

#8
Northguys theory is that if you measure input and output in amps (ie not in volts or watts), then amps are electrons, and if you put them in, they must come out.  However, two things change that:

1. self discharge, which he reckons is low enough to disregard (not sure i agree)
2. electrolysis bubbles off charge in the form of gases.

Hence the well known adage that absorb is less efficient than early bulk, and why some people try to keep their cycles in the 50-80% SOC range (with periodic absorbs).

This isnt to say than on a charge cycle average, a flat figure wont do the job. I think the only time you can check the SOC datapoint is when it returns to 100%, which by defition means it has completed 1 full cycle, one bulk, one absorb. So if its 80% efficient in absorb and 99% efficent in low bulk, then 95% average for the whole cycle works, sortof.

However if you do what Chris O and others do and dont absorb everday, using that average isnt gonig to be as accuarate.

Honestly i dont know how important any of this is. Maybe Bob is right and +/- 3% is good enough?

BTW mtdoc have you seen this, about floating agms:
www.intechopen.com/download/pdf/9712

They make a case that AGMs are usually floated at too high a voltage. Because of their low self discharge, they dont really need float at all. 
6x300W CSUN, ground mount, CL150Lite, 2V/400AhToyo AGM,  Outback VFX3024E, Steca Solarix PL1100
http://www.zoneblue.org/cms/page.php?view=off-grid-solar

vtmaps

Quote from: zoneblue on October 31, 2013, 08:01:55 PM
Because of their low self discharge, they don't really need float at all. 

Floating is very important to opportunity loads.  When I run a laundry in the afternoon, float allows my solar panels to power the laundry. If I turned off float, I would be using the batteries to do the laundry even though the sun was shining.    --vtMaps

zoneblue

well this is true! bit hard to use the battery when its disconnected.

The paper gets around this with a microcontroller that switchs out the controller.  They empirically confirmed extended battery health, but im not sure id want any kind of relay on my battery line...like they did. But you could turn the classic to standby mode...  lower float voltage should also help?
6x300W CSUN, ground mount, CL150Lite, 2V/400AhToyo AGM,  Outback VFX3024E, Steca Solarix PL1100
http://www.zoneblue.org/cms/page.php?view=off-grid-solar

mtdoc

#11
Quote from: zoneblue on October 31, 2013, 08:01:55 PM

Hence the well known adage that absorb is less efficient than early bulk, and why some people try to keep their cycles in the 50-80% SOC range (with periodic absorbs).

Are we mixing apples and oranges here?  I think most people who think of bulk being more efficient are really referring to the fact that more of the potential PV harvest is being utilized by the CC.    Dunno - maybe I'm wrong.

If you think of it from a thermodynamics/conservation of energy point of view were does the energy relayed from the CC towards the battery bank go? 

Obviously there are some wire losses, heat, etc (nothing electrical is ever 100% efficient as boB points out) and then there are the actual losses at the battery bank itself:  Heat production and, in flooded lead acid batteries there is also the off-gassing of hydrogen  (which we know contains energy).   Is there any other place energy is lost at the battery bank?

I would think the heat losses in wire or battery bank would actually be less during absorb than bulk since there is less current (ie less charge) moving.

As far as the bubbling/off-gassing of hydrogen in flooded batteries, my understanding (which may very well be wrong!)is that the off-gassing during absorb is fairly minimal and it is really only during equalization that there is vigorous bubbling.  It would be interesting to quantify the amount of hydogen gas that is output during various charging stages - then the exact amount of energy loss could be measured.  I'm sure this has been done by someone!

So  I dunno - but I am skeptical that the charging efficiency during absorb is that much different than durring absorb - even in flooded batteries.

Of course in AGMs there is no bubbling or off-gassing (unless you are doing something very wrong) - so no energy loss there.   This probably accounts for the slightly better overall charge efficiency of AGMs versus FLAs.  But again - this is only a small difference and one that would argue against there being a large difference in efficiency between bulk and absorb in FLAs.

Finally - if the bulk stage is that much more efficient than absorb then how would one account for the behavior of forklift/traction batteries?   They are overall much less efficient that standared FLAs and spend relatively more time in deeper discharge and bulk stage charging - I think - but I could be wrong on that...

Anyways it is an interesting topic to think about for sure!

Quote

BTW mtdoc have you seen this, about floating agms:
www.intechopen.com/download/pdf/9712

They make a case that AGMs are usually floated at too high a voltage. Because of their low self discharge, they dont really need float at all.

Thanks for the link.   :)

I haven't yet looked over it in detail but looking at their conclusions at the end I don't see them arguing for no float charge - just certain parameters for float.  Honestly I trust the battery manufacturers engineers real world experience the most in any case. My battery manufacturer has a very detailed Technical Manual that details optimal charging and float parameters.

Of course AGMs have a lower self discharge rate than FLAs as the paper you linked to described.  My batteries are supposed to have a shelf life at full charge of a couple of years but a float life of 10+ yrs!   So no float charging would likely be safe for the short term (with the exception of vtmaps point) but I'm not sure why it would be better?

In reality for my system,  which is grid tied and only occasionally has the batteries dischaged,  when I am in "float" during the day I am selling my full PV production (minus supplied house loads) and not sending any charge current into my batteries.   So I don't know how that scenario applies ... :o
Array 1: Sanyo HIT225 X 8 on Wattsun tracker. Array 2: Evergreen ES-E-225 X 12 on shed roof. Midnite e-panel with Outback GVFX3648, FNDC and Classic 150 X 2. 436 AH AGMs. Honda eu2000i X 2.

zoneblue

#12
I dont know mtdoc, which i had the time to learn more about it! Different sources say that agms float well, others say they often die early in float service. Possibly differences in design factor in there.

But the papers premise that low self  discharge requires lower float current makes basic sense to me. They are talking about micro amps is all that is required to maintain SOC. These last few days ive watched the agms here suck up between 0.2 and 0.8 amps, all though my long afternoon floats. And watched the "Ah In" datapoint steadily climb. Yet that energy is adding nothing to the SOC. Nothing, so late float efficiency is exactly zero. No?

Re: purpose of maintaining 50-80% SOC, what you say makes sense, but i do recall Chris O particularly saying that he measured charge efficiency at those SOC range. It was in that thread about optimum soc. Will try to find it.

Re I2R losses, true. Hopefully our cables are big enough, and with AGMS the internal resistance is very low which helps!
6x300W CSUN, ground mount, CL150Lite, 2V/400AhToyo AGM,  Outback VFX3024E, Steca Solarix PL1100
http://www.zoneblue.org/cms/page.php?view=off-grid-solar

Westbranch

Intriguing article.

I have been floating my AGM's for the last year now as the AC system is not finished yet, so have been keeping watch on the input to them and pondering if there is any issue(S) with keeping them WELL TOPPED UP?
Will keep in mind this concept when I reread this doc from my manufacturer.

http://www.cdtechno.com/pdf/ref/41_2128_0212.pdf
KID FW1811 560W >C&D 24V 900Ah AGM
CL150 29032 FW V.2126-NW2097-GP2133 175A E-Panel WBjr, 3Px4s 140W > 24V 900Ah AGM,
2 Cisco WRT54GL i/c DD-WRT Rtr, NetGr DS104Hub
Cotek ST1500 Inv  want a 24V  ROSIE Inverter
OmniCharge3024  Eu1/2/3000iGens
West Chilcotin 1680+W to come

zoneblue

WB, the whole book (its from) is quite interesting:

Trends in Telecommunications Technologies
Edited by Christos J Bouras, ISBN 978-953-307-072-8, 768 pages, Publisher: InTech, Chapters published March 01, 2010 under CC BY-NC-SA 3.0 license

http://www.intechopen.com/books/trends-in-telecommunications-technologies
6x300W CSUN, ground mount, CL150Lite, 2V/400AhToyo AGM,  Outback VFX3024E, Steca Solarix PL1100
http://www.zoneblue.org/cms/page.php?view=off-grid-solar