24V solar powered system - Question 1

[NeillHog]

Hello!

We have built a communications repeater station and it has been working for over 18 months.
It is set at 1700 meters in the Alps and supplies Internet and telephone services for a village and a few ski lifts.

https://plus.google.com/u/0/photos/...6105691627253201778&oid=103827427036815679772

There are four 12V 125Ah batteries - two by two giving a 24V output. There is a load of 20W permanently connected.
We can measure the system voltage via a GSM modem and I automatically get the actual system voltage sent to me before sunrise and an hour after sundown.
It works great but I would like to improve the monitoring system and have a question.

All the values that relate voltage to capacity assume that the battery has no load connected. But we always have the 20W being drawn. Is there an equation that allows us to take this load in to account?

Thank you and greetings from the snow covered Alps.
Neill

 

[waross]

You cab subtract the 20 Watts from the charging energy during the charge hours. You may have to estimate the condition during periods of low charge when the 20 Wtt load is partly fed from the solar cells and partly from the batteries.
Thanks for sharing the beautiful pictures.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[Sparweb]

What a view!

http://forum.solar-electric.com/showthread.php?25160-24v-Battery-Bank

They are copied out of a magazine (Home Power #36, August- September 1993)

I have referred to these curves on several occasions to estimate the state of charge of a battery, either under discharge or recharge.
It does not take temperature into account, however. At room temperature I've found them accurate, outdoors in the winter - not so much.

Charge controllers (you should be using one between the solar panels and the battery) have a temperature compensation coefficient built into them. If you have installed a temperature sensor on the cells (I highly recommend this!) then the controller will automatically adjust the float voltage up by 0.03V / degree C below 25C. Hence at 25C the float voltage may be set to 27.6V, and at 0 degrees C, the float voltage would be:
(27.6V) + (0C - 25C)*(0.03V/C) = 27.6V - 0.75V = 28.35V

You may want to explore this link:

http://batteryuniversity.com/learn/article/discharging_at_high_and_low_temperatures

But not all of it is true. Statements like "At –20°C (–4°F) most nickel-, lead- and lithium-based batteries stop functioning..." are false if taken absolutely. The capacity degrades steeply at those temperatures, but they don't "stop". I can start my automobile at -35C just fine.

So-
After the sun has gone down, and the load is still on, find the suitable curve on the graphs above for the load, then plot the voltage you measure MINUS the temperature compensation factor, to locate the approximate state of charge.

STF

 

[IRstuff]

Since the load is so small, you might entertain the notion of including a current monitor, whereby you can determine how much charge has been consumed to increase the accuracy of your SoC estimate.

TTFN
faq731-376

Need help writing a question or understanding a reply? forum1529

Of course I can. I can do anything. I can do absolutely anything. I'm an expert!

 

[itsmoked]

Spar; Thanks for those charts. I've often had NeillHog's problem but ultimately just muddled through.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[Sparweb]

Keith and Neill,
Now that I'm at home, I can upload the original article for you.
I must correct what I said about temperatures (I shouldn't be so eager to write this stuff off the top of my head)
The coefficient to use depends upon the battery type (flooded, AGM, gel) and the voltage of their connections.
I gave you 0.03V/C but that factor applies to a 12-volt set of cells. For your 24V system the coefficient scales up by 2x.

There is something else I'd like to add: The resting voltage of a cold battery is lower than a warm battery cell, but not 0.03V/C lower. Maybe it can't be pinned to a simple coefficient. My bank of AGM batteries (yes, I have a 24V battery bank, too) tends to rest overnight at 25.2V when it's cold (like today) and 25.6V when it's warm (in the summer). This doesn't follow the coefficient that applies to re-charging them.

20W / 26V = 0.8 Amp
0.8A / 250 A-h = 0.003
In battery parlance, this is a "C/300" discharge rate. A very light load.


STF

 

[itsmoked]

Ah! The actual article. Thanks.

I'm working on a project now that will probably benefit from this new knowledge.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[NeillHog]

@Bill. Thank you for the suggestion about subtracting the power being drawn from the current being put back in. Unfortunately we have no way of measuring the power being supplied by the PV system. With the system running at 20W, we do not want to start adding measuring systems with 1-2W.

@SparWeb. Thank you for the link to the 24V curves and the link to the original article. That was exactly what I was looking for.
Just to be sure I do not have a mistake in my thoughts. We have two 12V 125AH batteries in series and this set up twice in parallel so we have 24V x 250Ah = 6000Wh. We are drawing 20W so we are at C/300. If I am right then I will need to find the curve for C/300 but at least this shows that it is possible.

Do I understand you correctly that the float voltage should be increased at lower temperatures? You wrote 27.6V - 0.75V = 28.35V which has me confused. And yes I understand that for a 24V system the coefficient should be 0.06V/C.

The information about your 24V AGM battery bank is very interesting (we also use AGM batteries). If I understand you correctly, then a fully charged bank will show less voltage when cold than when warm. Can you give me the temperatures that you describe as “cold” and “warm”. A few hours after sunset on a sunny day I am seeing our battery bank showing 25.4V volts. 12 hours later this will have dropped to 25 or 24.9V. At a constant temperature this suggests that the battery has lost 10% of its charge. For this to be the case, then the system would be drawing twice as much power as the 20W we know it uses. But the temperature at night is also dropping and tends to be about -5°C in the morning. So if you are suggesting that maybe 0.4V (roughly) loss is not due to the discharge but rather to the difference in temperature, then that is very interesting.

Your answer and the documents that you have attached suggest that by measuring voltage and temperature and knowing the load, we may be on our way to predicting when we need to get the skis and snow shoes out again. Thank you!

@IRstuff. Thank you for your suggestion but at the moment I am trying to avoid extra measuring equipment, complexity and battery usage.

Glad that you all enjoyed the pictures. It is a tough job up here in the mountains but some one has to do it Being honest, I was up there recently at -15°C during a blizzard and ended up being rescued by a piste bully – but we don't talk about those days.

 

[NeillHog]

Just a note on a test I am running today.
We had a fully charged AGM battery on the shelf at 19°C. It showed 12.90V.
I took it outside and sat it in the snow. The air temperature is 0°C.
7 hours later it is still showing 12.90V.
I was expecting to see a change in voltage based on the observations of SparWeb. I am not seeing that yet.
I'll leave it out all night and post the results again tomorrow.
Neill

 

[Sparweb]

Battery banks VS snow banks!

Thanks for the follow-up info Neill, that give me a better picture of your set up.

Not knowing at first what type of batteries you were using, I made a guess of "flooded" type. Since you have AGM's I'll address the differences.

It sounds like your batteries are new. My set is rather old, but with care I can keep them going. Making comparisons between my measurements and yours should not cause you undue worry. Your bench test with a battery resting at 12.9V for about 7 hours, even in the snow, is normal for a brand-new battery. Once it has been in service for a year it may not do this. Old batteries can become very sensitive to temperature, but, their age is also reducing their total capacity. My old batteries sit outdoors, in an unheated building. They are enclosed in an insulated box, and due to their mass they keep the temperature inside the box about -10C when the daily temperatures can vary from 0C to -20C. I accept that this is not ideal because I cannot move them into the house (many reasons for this, just ask my spouse).

AGM batteries should not be overcharged. Their cases are sealed, and if they lose water, they permanently lose capacity. It is very important to use a charge-controlling device to get long life out of AGM batteries. Connecting them directly to the output of solar panels can reduce the battery life dramatically. There can be confusing information on this point. People connect solar panels to their car batteries all the time, so what's the problem?

Automobile batteries are designed to withstand a rather brutal life cycle; they tend not to be re-charged completely every time, temperatures fluctuate rapidly, vibration, etc. So any advice that applies to automobile batteries and recharging them should not be trusted for deep-cycle AGM batteries. To use an analogy you might find amusing - they are as different as downhill skis are from cross-country skis.

There is a process called "equalizing", which over-charges a battery deliberately, to use a small amount of heat and electrolysis bubbling to break away deposits of sulfate on the plates. This is normally done for "flooded" lead-acid batteries. It does not damage them; in fact it is good to do this every month or so, as long as the fluid level is topped up with distilled water. None of the acid is lost in equalizing, just hydrogen and oxygen gases, which can be replaced by opening the covers and refilling with water. But your AGM batteries are sealed!

For my AGM batteries, I permit a maximum 28.2 VDC when they are being bulk charged, even in the cold. I have a charge controller that regulates this (Tristar TS-60). Once the bulk charge phase is complete, it regulates the solar cell current to reduce it slowly, then holds the current steady at a level that keeps the batteries at a float voltage of 27.2V in the winter. If the sun never set, the charge controller would never stop holding 27.2! But it does, and there are loads that drain my system, so every morning it goes back into a bulk-charge cycle again. Sometimes I allow the system to up to 29V, but only once every 2-3 months. The batteries get noticeably warm when this happens - they are being heated by the current flowing in them. This would not be healthy if they were already warm indoors, but they are outside, and in the winter a little heat is beneficial.

If I had flooded lead-acid batteries, I would use different float voltage and bulk voltage settings (higher). With AGM's, I don't strictly follow the 0.06V/C rule. In the summer, temperatures are warm enough that the temperature compensation (an automatic feature in the charge controller) doesn't raise the voltage up that high. I forget the exact settings, but at 25C it may be 27.6V / 26.8V. It rarely ever gets that warm here (Calgary) so I rarely find them at these set points.

I could give you a plot of their voltage for January, if you'd like.

STF

 

[NeillHog]

Hello SparWeb,
Thank you for taking the time to answer me.

Between the solar panels and the battery bank we have a charge controller - Solarix MPPT 2010 (

http://www.steca.com/index.php?Solarix-MPPT--MPPT-1010--MPPT-2010537c9a802b50e).

This controller has an internal temperature measurement and is mounted in the same enclosure as the batteries. It automatically compensates for temperature. Nice is also that it only uses 10mA.

I'll get back with more information later but I just wanted to say thanks and ensure you we are not trying to kill the batteries.
Neill

 

[Sparweb]

Neill,
I went back you your original picture.
It looks like you have 1/2 square meter of solar panel.
If so, then winter sun at low angle on your vertical panels will give 4-5 hours of charge at about 200 Watt.
That adds up to about 1000 Watt-hours of energy per sunny day.
Drain the batteries with 20 Watts every 24 hours, is a load of 480 Watt-hours per day.
So your system is in balance, roughly.
With battery capacity of 250 Amp-hours at 26V then the storage capacity is 6000 Watt-hours, or enough to last 2 weeks with absolutely no sun.
Accounting for temperature, which reduces battery capacity substantially, you don't want a whole week to go by without some serious re-charging.


STF

 

[NeillHog]

Hello Again SparWeb.

The system has the two solar panels. They are mounted close to vertical so that snow and ice fall off as fast as possible. The photo was taken after a night of snow with temperatures below freezing. Each panel has a rating of 245W. They receive sunlight from sunrise to sunset and there is nothing to cast a shadow on them at any point.
I measure the battery voltage a few hours after sunset each day and am seeing that, even on days when the battery voltage starts at 24.3V, it will be showing 25.4V (full charge) a few hours after sunset on sunny days. The charging side of things seems to be running well.

But, as you mentioned, we have a theoretical capacity of 6000Wh and a load of 20W so the theory says that the batteries should be empty after 300 hours or 12 days.
I measure at 18:30 and then the next morning at 06:30. In this time I know there is no power from the PV array and that the system is drawing 20W so I would (theoretically at 25°C) expect to see a drop of (20W x 12 hours) / 6000W = 0,04 or 4% of capacity.
I am actually seeing system voltage drops as follows
25.3 -> 24.9
25.4 -> 24.9
24.6 -> 24.4
25.2 -> 24.7
24.8 -> 24.6
Obviously measuring to +/- 0.1 is not optimal and we are back to my original question about correlation between voltage under load and capacity.
But if we assume that any error caused by the 20W load is constant then the five data pairs above suggest capacity loss of 14%, 17%, 9%, 22% and 9% so an average of about 14%. Even the best data pair show capacity loss of twice the theory.
This difference between the 4% in theory and 14% in reality is what I am looking for.

You mentioned that the temperature is going to reduce capacity. I agree with you on that. The data sheets (

http://www.lifelinebatteries.com/graphs.asp)

suggest that at -10°C (14°F) and c/20 the bank should still have 80% of its capacity available. So this would increase my theoretical 4% to 5% (4% / 0,8) which is still far from what we are seeing.

Lots of facts, lots of figures. But I am really keen to find my problem as working off grid relay stations revolutionize the Internet connectivity in rural areas.
Thank you and have fun
Neill

 

[Sparweb]

Your measurements are not quite what I expected, either.
Yesterday my bank was topped up by the wind turbine all night, but the night before I watched the voltage readings at various times:
18:00 PM

25.6V​

23:00 PM

25.4V​

7:00 AM

25.0V​

This is a typical voltage "sag" for my bank. It was about -5C. I usually see this range overnight when there are no loads running off of the batteries and there's no wind to charge them overnight. My bank is quite old, and larger than yours (17kW-hr). Your system sags by ~0.5V overnight, while mine sags by ~0.6V overnight, so this may not be the main difference. Mine seems to have a resting voltage much higher than your system does. That's what really bothers me.

Take a look at the Solarix charge controller when you are next up on the mountain, or maybe with your remote monitoring system you can observe the voltage trend numerous times through the day. I looked at the datasheet and the manual - I can't see any setting for battery type or voltage setpoint. The datasheet says there is an option to connect an external temperature sensor, but I see no evidence of it in the manual, nor any terminals for the wires.

One thing I found quickly looking at the Lifeline battery charts, there is no Voltage limit for temperature corrections - just keep increasing it as it gets colder. So that throws away my earlier suggestion that your AGM's may require a voltage limit. In fact, your system should have a resting voltage higher than mine, both because of their age, and because proper temperature correction should raise their voltage to a much higher range than mine. For example, at -10C, the float voltage should be about 29V. They will still sag back down to about 26V overnight, but never down to 24.6 as you have seen.

So my first guess is that the charge controller is unable to determine the temperature of the battery cells, and is regulating voltage at far too low a level to complete a proper charge of the batteries. It may be designed to assume they are stored at 25C, or worse yet - a temperature sensor inside the case of the charge controller is used, which is worthless data.

Take a look at the manuals for charge controller such as Morningstar's Sunsaver or a Xantrex/Schneider C40 and you will see that proper regulation of the batteries must account for their type and their temperature, and ideally will also give them separate bulk, absorption, and float charge cycles to prolong their life.


STF

 

[NeillHog]

Hello SparWeb.

Do I understand you correct? That you are seeing your "sag" even with nothing connected? My loss during the night is (I think) because I have the 20W connected. The test I did on the fully loaded battery showed that it lost nothing with nothing connected even after 24 hours and this agrees with the technical specs.

Page 30 of the manual (

http://www.lifelinebatteries.com/manual.pdf)

suggests that with a small load connected the battery should begin by showing 12.69V. I have two batteries so they should be showing about 25.4V when they are 90% charged. I think that yours is showing more as you have no load connected so you are measuring the open terminal voltage.

The Solarix controller has an option for an external temperature sensor but we use the internal one as controller and batteries are in the same small enclosure. The manufacturer confirmed that this is OK when I spoke to a lady at their factory.

The Solarix controller can be programmed with bulk, absorption, and float charge voltages but our technicians din't do it correctly. This is not a good thing and will be corrected during the next visit to the top of the mountain.

But I think that my original question can be answered with a definite "yes". By knowing the charge being drawn and the battery voltage, I should be able to calculate the remaining capacity. The table on page 32 of the manual is a good start but I need a curve for the 300 hour rate. I assume that if I had that, and applied the temperature compensation tables, then I would be able to estimate capacity based on voltage and temperature.

Neill

 

[IRstuff]

Your manual tells you how to do that on page 33

TTFN
faq731-376

Need help writing a question or understanding a reply? forum1529

Of course I can. I can do anything. I can do absolutely anything. I'm an expert!

 

[Sparweb]

Yes, there is a sag, even with nothing connected. It is important to repeat that these batteries are very old - 20 years old. I am their second owner after a commercial building was finished with them in their UPS.
Their resting voltage will remain at 25.0V for a very long time with no loads on them (months). Sagging down to below 2.1V per cell is no surprise for such old batteries.

Your new ones should not be going below 25V overnight. The load you have on them (C/300) is trivial and should not cause a sag down below 25V in your case. I suspected the kind of error in the charge controller that you describe. Once the program is set correctly, you will see much better overnight resting voltages.

Not having an external temperature sensor is still compounding the problem. As the MPPT and PWM functions of the charge controller require electronic switching, there are transistors called MOSFETS or IGBT's that manage the current flow. Even though the device itself is not converting excess energy into heat, the MOSFETs still warm up when they operate. Furthermore, the plastic case of the CC in the enclosure has a low thermal mass, which allows it to warm up during daily operation. Conversely, the batteries contain thick plates of lead and represent a thermal heat sink. They are probably much colder than it is inside the case of the charge controller - even though they are in the same compartment. On this point I disagree with Solarix. The effect is significant, perhaps misleading your controller by 10 degrees. In my system, I measure the temperature of the batteries themselves, the air in the building where they are located, and the temperature outdoors. The difference between all of these temperatures is very wide, especially on sunny days that heat the building. The heat doesn't penetrate the insulated box that encloses my batteries quickly.

IRstuff is right, you will need the Peukert formula to estimate capacity at C/300 discharge rate.

STF

 

[mikekilroy]

[There are four 12V 125Ah batteries - two by two giving a 24V output/i]

just an side note... consider changing to 4pc 6v deep discharge 250ah golf cart batteries for better life. batts do not work optimum in parallel.

http://www.KilroyWasHere<dot>com

 

[ScottyUK]

Hi Mike,

If changing the cells were an option then flooded NiCad would be a far better choice for the operating temperatures involved. With that said, the AGM type seem to be doing reasonably well in SparWeb's application so this is proving an interesting thread.

 

[itsmoked]

Indeed Scotty.

I totally agree with Spar's "Use a proper external temp sensor not the lame internal one."

I also would add you should do the correction trip to the mountain ASAP. Every day under charging is taking a cycle-life toll on the batteries. On one of our private rail cars, poor complete charging just roached an $8,0000 battery bank. If only I'd been in the loop. :/

Keith Cress
kcress -

http://www.flaminsystems.com