question

sieade245 avatar image

Mixing battery types and capacities - my experiences

Hi guys. I've asked a lot of question on here and in the previous Victron community forums which people have helpfully answered to get me up and running so I thought I'd spend 5 minutes to give a bit back and talk about what I have found during the setup of my frankenstein battery bank which I use.

I'm not an expert on battery physics. You might not agree with some of my points however all I can say is these are ACTUAL results, not theory, based on the performance of my own kit.

Some background. I install Victron equipment at work: phoenix and Skylla chargers, multiplus and quattro inverters. We don't use CCGX or ESS we are 100% off grid using solar power to charger batteries which then power our loads with diesel generator support for the winter. Our banks can range from 200AH to upwards of 5000AH.

When our batteries are 10-15 years old, we scrap them and replace them. We have used wet cells OPZs which require water maintenance in the past but recently we have moved to AGM/ OPZv deep cycle gel batteries. We normally use 2V cells normally using 24V although we also use lower capacity 12V gel batteries in some installations as they are lighter and easier to move by hand. For installations which have mains power, our batteries are constantly held at float charge unless there is a power cut which is extremely rare (UK). Out stations with no mains power are recharged daily by solar power and used at night however they are massively overspecced with normally at least 2 weeks operational standby time designed in incase the backup generator fails (need a helicopter to get to most of them). We inspect the batteries annually and try and do a 50% load test if we have enough time on site.

I realised that most of the AGM/ Gel batteries we were scrapping from the mains stations were still serviceable for use in most cases so I asked if I could start to recycle them instead of scrapping them.

I started with a recycled bank of powersafe V AGM 2V 460aH cells. They were 12 years old. I sourced a used Multiplus 24/1600 unit and bought a new CCGX and a grid meter. I already have a 4kw solar PV installation. I set the bank capacity at 440AH in VEconfigure with a small 5% reduction in capacity due to their age. I set max DoD to 35% (so 65% SoC in CCGX ESS). The batteries are then cycled daily if the solar power is good enough to recharge them. That was just over a year ago now. I'm using 35mm cable with temp comp but no voltage sense, my 1600W unit doesn't have it.

One thing to mention first is that I have an issue with my multiplus measured battery voltage being 0.4V lower than the battery terminal voltage under load. This is not just down to volt drop which I have calculated and measured at 0.2V with 35mm cable. There is a measuring issue with the multiplus. I have never resolved it, there are no voltage sense terminals which would likely correc this. This meant I had to play around with the dynamic cut off voltages to stop the inverter switching off on low voltage early.

The system has worked fine with some caveats. I noticed that the battery terminal voltage dipped into the high 23V/ low 24V when the inverter was operating at max output 1600W close to 65% SoC. I could work around this but I had to manually change the dynamic low voltage cut off curve in VE configure and lower the limits. I put this down to increased internal resistance of the cells or possibly one of more dodgy cells which was dragging the rest of the bank down. To offset this and also to protect my bank from early failure, I limited the max inverter power to 1200W in CCGX ESS settings. This seems to have fixed the problem and the bank worked OK for the last year usually ending up around 24.3V under load. Resting voltages increase to around 24.6-7V which would suggest around 60-70% SoC (AGM cells). This is pretty impressive for 10 year old AGM cells I think.

Next, last summer I added another 12 powersave V cells. These were the same battery series but only 400aH and this time they were 12 years old. I tested them on their own first and noticed that at least one cell was a bit dodgy with the terminal voltage dropping off under 1200W load to the low 24V after around 30% DoD. I found the dodgy cell easily by checking the cell voltages under load. However...these are free batteries so I'm not going to chuck one.

I linked this bank in parallel with the first and set the total battery capacity relatively conservative at first to 660AH due to the dodgy cells. What I found though, is that the terminal voltage under load of the whole battery was much improved from before when they were individual banks. It didn't drop below 24.5V now under load, and resting voltages at 65% Soc were high 24. I put this down to two things:

1. The banks support each other. So the "good" first bank maintains the terminal voltage by propping up the dodgy cell in the second bank as they are in parallel.

2. The load on each bank is effectively halved. So before at 1200W, there may have been a discharge current of around 50A running through a bank, but now its 25A through each bank as they are in parallel. This is similar to limiting inverter power to 600W and changing the roughly C10 discharge rate to C20. This also increases the effective capacity I imagine. It also has the same effect on the charging rate which is lower and better for the AGMS. So less stress on each bank.

I increased the capacity in Veconfig to 700 and then to 800AH, running week long tests in each case. All was still good with slightly lower resting voltages at 65% SoC.

I next did what several websites told me not to. I added 3 more banks. But this time, these were gf-12 gel batteries. These were also 12 years old (estimated lifespan 5 years) I connected 2 in series to make a 24V bank and had 6 in total for 3 banks of 24V GF-12. These are rated at around 75Ah per unit, so 75AH per bank of 24V. These are all connected in parallel with the first 2 banks. Now, I know what you're thinking. That doesn't sound right I've paired 460/400/75/75/75. The 75s are going to be charged quickly first and then be overcharged while the 460/400s banks are charged which take much longer? I haven't found this to be the case at all. The terminal voltage of all banks is always the same during charging. So what is happening is that in fact all banks are charged at the same relative speed with the 460/400 having much higher relative charge current and the 75 banks are charged with much lower current. Tested with clamp meter. I think this is because they are all connected in parallel again, similar to before they balance each other. My max charge current is 40A with the 1600W multiplus but after losses I only get around 34A total. With discharging, the same appears to be the case. The majority of the current is taken from the 460/400 AH banks with lower current from each 75Ah bank with an equal terminal voltage across all banks.

I increased the battery capacity first to 880Ah, then to 960aH and left it on this for the past 6 months. What I noticed is again, the overall terminal voltages under load and resting are much improved on before with only the 2 banks. Under load, they didn't dip below 24.6V and resting voltages were normally in the high 24V low 25V (harder to test as half this period it was winter so I used weekly scheduled charging in the CCGX to cycle them). Also I was able to remove the 1200W limit on inverter power without any negative effects on terminal voltage. Now I hit a problem though, my battery capacity had grown so large that my multiplus couldn't charge it fast enough during the day to get it back up to 95% every day. Especially because it was winter, sometimes there was some sun in the morning and then none in the afternoon etc. Also 34A is only a third of the C10 charge rate I was aiming at....although now its not as simple to work this out because of what I mentioned above, the 400aH bank and the 460aH bank take the majority of the current.

I upgraded to a refurbished 24/3000 unit. This was an old unit which had the new CPU retrofitted for ESS usage with a max charge current of 70A/ 64A real world after losses. This was an easy straight swap and I copied my settings across. Under charge I measured around 25A to the 400/460 and around 5A to each 75A bank in bulk charge phase. These are nice safe charging rates for all the batteries, at around C15. Now what I haven't tested is the inverter max output of 3000W. I instantly limited it to 1600W for a couple of reasons:

1. This is the max normal load I have in the house unless the cooker/ shower is running. This is not meant to be an off grid system its supposed to save me electric and I can save more of it by running under powered for increased effective battery capacity. The cooker can use 3000W, the shower 9000W. The batteries will still supply some of this. The washing machine and tumble dryer we don't use unless its sunny...or unless I'm offshore and the wife uses them anyway :-)

2. I'm still concious that these are 12 year old batteries. Even with the parallel banks I don't want to push them to an early? earlier death! They're only going to last so long until I can recycle some newer ones.

NEXT (yes the saga continues are you still reading) I added ANOTHER 4 GF12 in 2x2 24V configuration for a new total of 460/400/75/75/75/75/75. These were only 5 years old but had been left to discharge for at least 4 years so they should be knackered. But after a recon charge on a CTEK charger they live again. Now just to be clear here, I hand picked these from a total bank of 20 cells. 16 of them were in fact knackered with terminal voltages of 0.9V - 5V so they are dead, but 4 of them were recoverable.

The max theoretical capacity (assuming brand new and C10) of all of this is now 1235Ah but I've limited this to 1160 for a 5% decrease in capacity due to their life..offset against the fact that now the C rate of charge and discharge decreases once more due to the increased parallel banks. Here is a shot of the system at the moment. It charged to around 94% today with a good solar day:



So all looks good, with a discharge current of 37A split across all the banks, this results in a very low discharge current per bank and low stress to each cell so I've got quite a high load terminal voltage of 25V.

One thing to mention is I haven't yet looked if my new multiplus has the same measurement error I mentioned above so this could also be a factor in my increased DC voltage reading, I'll have to get around to that.

Needless to say, this is easily going to power the house all night, loads are high now because the wifes watching celebrity love island kitchen idol on ice downstairs and I'm on the office PC. Overnight loads are around 200-300W. Unfortunately, yesterday as you can see was a no solar day so the graph doesn't show overnight performance very well but I've been getting to around only 70-75% Soc each night before the solar charges them back up again the next day.

SO. That's a lot of writing. It probably goes against a lot of battery dos and don't but I did read a paper from a chap on the internet who'd also tested similar setups with different battery types and capacities which largely the same results.

I think a KEY point is that all my batteries can operate with similar charge voltages. I'm using 28.8V bulk/ absorption with float at 27.6V but there's never time for it to get to float before the night cycle. I deffo am not suggesting its ok to mix wet cell and gel/ AGM. I certainly wouldn't try this battery matching of some sort is important. I don't think I can add any more without possibly upgrading the multiplus again to the 5KW version as its the charge current which becomes the limiting factor, I could add one in parallel but it can be hard to match them, I already had to sell one that I bought which I didn't realise had the "old" microprocessor so I don't think I'll go down this road. Having said that, adding more capacity and banks leads to even lower charge and discharge current per cell which increases battery capacity and expected lifespan also due to lower DoD required for the same AH output. The real test now is going to be how long they last and work as well as they have been working. I've also put the temp comp probe on the 460 bank as this what charges/ discharges with the most current but it means the other banks may be "unprotected" from a thermal runaway/ incorrect compensated charge voltage my garage is normally nice and cold though around 15 degrees with ventilation.

If anybody has any suggestions/ improvements or just wants to tell me what a bad idea this is please go ahead. Hope you enjoyed reading!

MultiPlus Quattro Inverter ChargerESSbattery chargingbattery capacity
1553984408828.png (52.2 KiB)
1553984447729.png (62.3 KiB)
10 |3000 characters needed characters left characters exceeded

Up to 8 attachments (including images) can be used with a maximum of 190.8 MiB each and 286.6 MiB total.

6 Answers
JohnC avatar image
JohnC answered ·

Hi Si. Thanks for that offering. I've played with multiple strings like this too, and my 'conclusion' is that they'll work ok for much longer if they're not worked too hard.

Imbalanced strings and failing cells will often show as temperature differences. I didn't have your menagerie, so just used a touch-test. But I almost bought one of those hand-held IR sensors, and I'd suggest it in your case.

A single hot batt/cell doesn't necessarily mean it's the baddie though - may be just the one doing all the work. Then it'll die from high temp.

But good luck with it. I've gotta buy my batts.. :(

2 comments Share
10 |3000 characters needed characters left characters exceeded

Up to 8 attachments (including images) can be used with a maximum of 190.8 MiB each and 286.6 MiB total.

John. What a good idea! I've got the IR readers at work. I never thought of doing that. I might borrow one for a few days and see what I'm getting at max charge/ discharge. Thanks

Best time to check them is after a fair stint of charge, say latish afternoon in a solar setup.

And the thing to really take in is that a hotter string will accept more A than a cooler one, so it becomes a bit of a 'thermal runaway'. Might be your best string too, but likely the first to die.

Stay on top of it..

dwmarine avatar image
dwmarine answered ·

That is seriously a cool story!

In the past, if one battery in a large bank had issues, you dug out the pitch top, drained the acid, cleaned it out and rebuilt it. So essentially it was a new battery in an older bank. Some of lead acid batteries in the telephone exchange my father worked in were near 20 years old. Great that you have the time and resources to experiment! Just make sure you have an air gap around each battery if possible.

Darren

Share
10 |3000 characters needed characters left characters exceeded

Up to 8 attachments (including images) can be used with a maximum of 190.8 MiB each and 286.6 MiB total.

tilo avatar image
tilo answered ·

Measuring the midpoint voltage with a BMV-702 might be useful here to detect in time a battery with problems. (In your particular case it might not make economic sense as the batteries are for free).

As charge and discharge current is controlled, a fuse for each group/line/bank of batteries according to their expected/measured working current might also help to detect if one group is taking over more work from other (failing) groups and thus be a candidate for thermal runaway. The fuses might be a problem in case of grid failure, when the inverter takes all the current it needs to cover a load.

A fuse per group/line/bank might also prevent discharging the whole battery installation in the rare case of a internal short circuit in one battery, which causes an instantaneous voltage drop of 2V in the affected battery.

If you are uncertain about the actual total capacity, is it possible to do a complete discharge to find out?

A question about the term battery "bank": Up to now I thought it was used for the complete battery installation. Here it is used for 2x 12V or 12x 2V batteries in series to make one "line" of 24V. As I am not a native English speaker, I would like to know which term is used for one "line" and for the complete set of batteries, just to avoid confusion.

Share
10 |3000 characters needed characters left characters exceeded

Up to 8 attachments (including images) can be used with a maximum of 190.8 MiB each and 286.6 MiB total.

Paul B avatar image
Paul B answered ·

Keeping the Draw current low across the banks is the key.

Good work and Thanks for sharing, all this work and examples do help


again Thanks

Share
10 |3000 characters needed characters left characters exceeded

Up to 8 attachments (including images) can be used with a maximum of 190.8 MiB each and 286.6 MiB total.

Boekel avatar image
Boekel answered ·

I don't understand you line here:

"Now, I know what you're thinking. That doesn't sound right I've paired 460/400/75/75/75. The 75s are going to be charged quickly first and then be overcharged while the 460/400s banks are charged which take much longer?"


Why would the 75Ah charge first?

Only thing that can be an issue is if you use flooded 2v cells and other types, as you have to charge the flooded cells to a higher voltage to get the acid properly mixed.
As long as your batteries are ok with the charge voltages there is no problem.

Problems can arise if you properly load the batteries, one bank providing more than the other (doesn't have to be a problem if you size everything accordingly).

In a 48V bank made with 2V Gel cells, usually the middle ones die first (they get warmer), charge the bank fully, than discharge and select the cells that hold best --> use to make 24V bank.

What would be interesting: measuring the charge efficiency, how many Ah and Wh go in, and how many come back out again.

4 comments Share
10 |3000 characters needed characters left characters exceeded

Up to 8 attachments (including images) can be used with a maximum of 190.8 MiB each and 286.6 MiB total.

Hi Boekel. Sorry I wasn't trying to suggest that they would charge faster.. instead I meant if you had separate 75Ah and 400Ah bank that you'd expect the 75Ah bank to charge faster (assuming the same charge current) but that of course it doesn't because they are connected in parallel. I realise this might be obvious to people who are experienced in battery tech but it also might not be so obvious to other people. I've read some of your other topics and I think you are definitely in the first group!

in practice the ones with the lowest internal resistance (vs capacity) will charge fastest, but probably not noticeable except for the last 10% SOC

On that subject, that last 15% is where I do struggle to get the charge in. ESS by default needs the bank to get to 95% otherwise it lowers the effective max DoD as part of the batterysafe algorithm. I can get the bulk charge to 85% done by 1pm normally...but its usually hovering around 93-94% at 5pm when the solar falls off. Won't be a problem as the days get longer and I have found a workaround to use VEconfig to change the 85% full assumption after bulk to 88/89% This normally ensures that 95% is hit even. Its very hard to determine how much actual capacity I have due to the age/ battery type mix and the dodgy cells so my 1100 ish AH capacity in the settings may also need some fine tuning. I could reduce it to get the 95% earlier. This was the main reason for upgrading to the larger multiplus...to get the bulk time lower so I have more leeway in absorption to get to 95% however something I forgot to mention in my post...

Although these batteries are 12 years old they are hardly ever used or cycled. I noticed that after the first month, the absorption phase was faster as the batteries were more readily accepting that last 15% charge possibly due to lower internal resistance caused by actually cycling them...similar to a new battery where the capacity is not at max until cycled a few times.

you really want to fully charge at least once a week of two weeks, otherwise your batteries will fail you soon.

kai avatar image
kai answered ·

interesting. I think you hit the nub of it when you said the the terminal voltages are all identical. As far as each bank (of different chemistry) is concerned, its just hooked up to a charger whose voltage might not be exactly optimised. Then again, these are free 10+ year old cells so optimisation is a relative term :P

I would be a little more hesitant about mixing chemistries in a 24v string though - better to have matched chemistry for a 24v string and then parallel up different chemistry strings.

Good job scoring the freebies


Share
10 |3000 characters needed characters left characters exceeded

Up to 8 attachments (including images) can be used with a maximum of 190.8 MiB each and 286.6 MiB total.