Battery Bank State-of-Charge Calculator

[bzrqmy]

I have another idea bouncing around in my head.  I want to develop a calculator to provide my battery bank's state-of-charge.  In addition, I would like to possibly calculate the bank's true Amp Hour rating. There are a lot of state-of-charge charts available, but most of these are no load charts and are meaningless to me as my bank is rarely at a no load/no charge state.  I know that measuring SG is an accurate way to measure SOC, but I have DOC Wattson's on my input and load.  I know Voltage, current and power at any instant.

At a minimum, this calculator should have the following programmable input parameters:

*    Nominal System Voltage

*    Battery type FLA/SLA/AGM  lets start with FLA as that is what I have

*    Actual battery bank voltage

*    Estimated bank amp hour rating

*    Battery bank temperature

*    Load on battery bank

At a minimum, the calculator should provide the following output:

*    Battery bank State-of-Charge in %

*    Low Voltage Disconnect cut-in voltage based on input parameters

Optional features:

*    State-of-Charge charts on the above input parameters

*    Battery bank amp hour calculator based on a known load over a period

        of time to 50% SOC

*    State-of-Charge calculations based on charging current/power input at any

        given time

*    Runtime for a given load before Low Voltage Disconnect

My plan is to use Excel to do the charting/mathematics.  What I need help with, and probably will get no consensus on, is the formulas and voltages to use.  I can set up a mini FTP server so anyone can access the current version of the program.  Heck, if anyone wants to contribute, that would be great too.

Anyone interested?

[DamonHD]

Not to rain on your parade, but it seems very hard to do well to the point that Morningstar (at least in my SHS and MPPT controllers) does not attempt to do more than measure battery voltage, probably (I don't know) correct a little for temperature and current in or out of the battery at any given moment, and add a little hysteresis.

Rgds

Damon

[bzrqmy]

Not sure how electrical vehicles are doing it, of course they are going to use lion, but seems that if you cannot calculate/estimate the SOC of the battery bank, then someone is going to be walking. Seems like we should be able to do the same for RE.

[taylorp035]

I think what you should do is have a amp hour meter that resets itself every time you top of the batteries, and also have a load meter, so you can calculate the power left or power needed.

Lets get this straight in my head

Meter's needed:

1. Power used
   
   
2. Power generated
   
   
3. Something that resets when you get to 100%
   
   
4. The efficiency of the system

   
   
   

I think this will be close enough if the goal is to know your current power percentage left.

[taylorp035]

Your other option if you wanted the voltage under load would to make a data logging system and track the voltage and amperage of the batteries 24/7 for a couple of weeks. Then you can use the data.

Unfortunately my idea would much harder in reality.  You would need a 3D array and some programming experience ( it can be done )  

[wooferhound]

Some charts down this page that show Charge vs Temperature

http://www.powerstream.com/SLA.htm

A chart down this page that shows state-of-charge: charging, at rest, discharging

http://solarjohn.blogspot.com/2007/03/measuring-battery-state-of-charge.html

[scottsAI]

Bzrqmy,

What you want to do is possible. Not easy.

Of course you checked out:http://www.batteryuniversity.com/partone-16a.htm

Good list, missing:

Peukert number (Not just impedance): http://www.evdl.org/uve/battery.html

Battery charging efficiency based on SOC. Varies greatly.

Battery discharge efficiency based on SOC. Drops as the battery is discharged.

Battery age.

Deep cycle batteries are rated to 20% SOC, consider estimating “time remaining” based on what discharge level they are comfortable with.

Glad to help, email me.

Have fun,

Scott.

[DamonHD]

I do indeed use SJ's chart myself, but take a very conservative view because of the big differences in voltage for the same SoC depending on whether the battery is charging/discharging.

Rgds

Damon

[frackers]

You'll need to count charge in and out in real time. That's what I do with my controller a couple oif times a second with a catchall for over and under volts.

Check out what I've done to see if it lines up with your skills and requirements.

I also get to make a fancy plot of what the mill is doing as well :-)

[Flux]

The only really practical scheme uses an Ah meter. I don't think you will get there using half baked figures from any other form of metering.

Even using Ah you have to include Peukert and make allowances for battery ageing. The tricky bit even then is keeping the thing on track. The only real reference point you have is full charge and these Ah based schemes reset to 100% charge each time the volts have been above a chosen point for so long and the current has fallen to a set figure. They work very well if you meet this requirement but if you don't get the batteries fully charged often enough they drift off.

Building an Ah meter is not that difficult and if you want real results I see no other way. I wouldn't waste any time on calculating things from any other data source unless you want a bit of fun and aren't too bothered if you get the wrong answer. Again if you reset your calculations from full charge you will be much closer to the mark.

Once you have got a reliable Ah meter running then you can always check your bank occasionally with a deeper than normal discharge to make sure it is still meeting its claimed capacity and you will be sure that you are working normally within a reasonable % discharge.

Flux

[scottsAI]

Happy New year Flux,

I'm hoping Bzrqmy is working on the next practical step in AH to a true SOC.

Navy, 1Mw battery had AH accumulator. NO sign. We controlled power in/out for the most part. Knew SOC relatively well. Including 4 hour discharge curve for it, we were willing to kill it if necessary. Never went below 70% SOC, people starting freaking out around then.

Solar / wind charging is never a complete charge or discharge cycle.

By modeling the battery based on SOC, temp, charge/discharge rate, time, age etc. the SOC can be more accurately determined than using AH.

4 years ago started out on this road, needed to estimate SOC for automotive battery.

Signed charge accumulator and Peukert number were a good starting point.

Vehicles battery is usually fully charged, simple starting point each cycle.

Tossing in Temperature, Peukert numbers gave us acceptable SOC. Project completed. Wanted to extend the work to RE (personal project), worked on it for a couple days, realized did not know enough to make it work, to busy to work on it further. Collecting info since to understand how it could be done. I like to model things, got my wind turbine model to correctly predict power into the battery.

SOC battery indicator for RE is more complex.

AH may be 20% accurate over a shorter time, longer charge/discharge cycles less accurate. Quickie AH on discharge side is rather accurate.

Peukert is a first level calibration factor, over time will not cut it for RE.

Surprising how little recent information about LAB modeling out there.

Considering LAB (of some type) represents 60% of the battery market.

If battery is modeled correctly expect it to tell you when to water it! When to turn on the fan.

Performance of the battery is dependent on battery construction.

Manufactures do not spell it out, any attempt at SOC must figure this out over time.

SOC indicator must be bonded to a battery. Connecting it from the start will be as good if not better than AH.

AH is SWAG. SOC better than 5% long term is possible. (I believe)

Long term small errors grow, destroying accuracy. By modeling the battery correctly, can nullify the errors based on measurable conditions: Current, voltage, temperature and time.

Not easy, else it would have been done. Someday will take this project on myself, many other projects in front of it. #1: Extracting water from air.

Willing to spend time helping someone else do it. Testing is the big time sink.

Not heard from Bzrqmy, maybe he was scared off or does not believe I can help!

Have fun,

Scott.

[madlabs]

I've been working on the same thing. One question I have is why do folks use amp hours instead of watt hours? It seems that watt hours would be more accurate.

Jonathan

[DamonHD]

Tradition is ... traditional.

Yes, Wh would indeed be more uniform and would help avoid all sorts of confusion, but that would be too simple...  B^>

Rgds

Damon

[scottsAI]

Madlabs,

Simple, while checking the load on your battery do you look at a watt meter or amp meter?

Scott.