SiriS

It could be a bad cell best have someone  connect there laptop up to the battery to monitor individual cell to get to the root of the problem .
An question how are those 2 batteries wired to each othere and to the inverter .
 

When the  battery gets charged, is it from grid (it would appear so) or from solar?
LiFePO4 batteries usually report 100% SoC well before they are really full, then they would still consume lots of amps to balance and actually get the cells full, look at my charge for yesterday, the BMS reported 100% at 15:03 app. but the batteries still took current until an hour later, by which time they were pretty much full...

Thanks everyone for all the very instructive replies. It seems like shading might be the culprit after all - the parapet is slightly higher than the bottom of the solar panels, so it is a creating a small shadow area.
Very small, but substantial impact!

Back to the drawing board to try to solve. 
Thank you 🙏
 
 
 
 

Just some pure speculation, which you could check with your installer.
To the right of the image, seemingly on a lower level, it looks like you've got a further 6 panels, but I'm going on the understanding that those are for a different purpose, not involved with this issue.
On your main roof section you're showing what looks like 18 panels, laid out in 3 strings of 6 panels each, and this is what I assume are the panels in question.
I think it's likely that you've got 2 of those strings in parallel on MPPT1, and the third string by itself on MPPT2. I say this because the Voltage on each MPPT is the same, while the current on MPPT1 is roughly double that of MPPT2.
At this point it would be interesting to know the voltage and current specs of your 545W panels to verify if that suspected layout makes sense. It might also help if you can confirm what combiner boxes you have installed.
What I suspect is that the discrepancy in power generation is more pronounced at this time of the year because of the lower sunlight in late autumn as we're going into Winter.
MPPT2 is probably outputting near to the peak power that it can manage in the lower available light - it can probably push closer to 10A during summer, but MPPT1 can continue to produce higher current because of the two strings of panels connected to it.
But as I said, this is some kansvattery, just some thoughts to be checked out.

Looks like your battery is fully charged which will result in the reducing of power being pulled from the panels. 
I assume that all your panels are facing the same direction and no shadding as you said. 
My guess is that you may have either a bad connection somewhere or a dodgy panel pulling down the amperage of the whole string. 
You could swop your strings around on your inverter to confirm it is a string issue and not the MPPT controller (not likely). 
If you have a Flir camera you can check the panels with it, a hot spot will show up pretty quickly, otherwise look carefully at the panels for small cracks. 

Set your times from 00:00 onwards  - else you'll get these oddities. 
 

That 1st graph shows nicely your base loads, so those spikes are normal and most likely your fridges switching on and off. Looks normal to me.
I only run 2 fridge/freezers so my spikes are not as pronounced, but they are still there.

If you shrunk that vertical axis, it wouldn't look half as scary. 
The range seemd to be just south of 500w to 150ish. I have 2 combos, deep freeze, security system and external lights running at night, plus all the stuff that's on standby, and I'm in the same ball park. So I don't see too much to worry about here. 
But you do well to keep an eye on this data. That's how we learn what's going on and figure out how we can optimise usage of the system. 

Hi, wondering what is causing these load fluctuations late evening/early morning (or probably the whole day but clearer to see as closer to base load at those times). Assuming one of the fridges, but is this normal or indicative of a failing motor perhaps?
1 x chest freezer (brand new)
2 x fridge/freezer combos in kitchen (5 years old or newer)
1 x wine fridge (5 years old)
1 x very old bar fridge
Thanks!
 
 
 
 
 

At roughly midday and if its a nice clear day then switch on big draw appliances such as microwaves etc and then see on your inverter screen or app what the combined pv output from both mppt is.

Thanks for everyone's comments/advice really appreciated! 
Still playing around a bit and getting used to different scenarios. Had a good test run last night/this morning - on batteries since about 7pm and went down all the way to 41% this morning including the 6-7am 4kW geyser run, then solar started to kick in and charged to about 50% already (and overcast atm). I should be able to manage LS in that period as well, as long as I can manage the beasts (geysers/pool pump etc).
Thanks again 🙏😎👍

Thanks for everyone's comments/advice really appreciated! 
Still playing around a bit and getting used to different scenarios. Had a good test run last night/this morning - on batteries since about 7pm and went down all the way to 41% this morning including the 6-7am 4kW geyser run, then solar started to kick in and charged to about 50% already (and overcast atm). I should be able to manage LS in that period as well, as long as I can manage the beasts (geysers/pool pump etc).
Thanks again 🙏😎👍

Don't waste excess PV get it into the batteries if your load doesn't need it. Just keep priority load ticked
 

Batteries having mini cycles is perfectly fine and better than charging to 100% and discharging to 15% over and over. 

Would it not use eskom from 9am before there is enough PV and batteries are below 80%? Would it not be better to have a setting of 30% during daytime? (to rather use PV and not make use of eskom). That late afternoon slot to 95% I agree is a good one.

Looks like a good starting point. As mentioned, you need to play around a bit. See how you run with these settings - and then tweak when you see that something does not work the way you want it
Correct
To me this is normal and to my understanding the batteries are built for this. Especially happens on days when you have overcast weather this happens. Like my day today:

I am using Sonoffs on my major consumers (heat pump, 2nd geyser, pool pump) to manage by major consumers only to come on when usually I have sufficient PV. A normal day for me looks like this:

I do not believe this is possible only for a specific time. 'Prioritize load' is the relevant setting - but if I am not mistaken this can only be set globally

Getting the hang of it...slowly 😀
Quick question re battery management/health. 
Batteries set to be charged from solar, after load. In practice this morning the following would happen:
Load 400W, Solar 900W, 500W charged to battery Then say load increased to 1500W, Solar 900W, 600W pulled from battery Then load reduced to 500W, Solar 900W, 400W charged to battery In that time slot where PV/sun not producing decent power yet the battery alternates quite a few times between charging and discharging depending on load change. Assuming this is not great for the battery/cycles?
Is there any setting to stop solar feeding the battery for a given time period? I would like for the solar panels to producer sufficient power to manage normal load before starting to charge the batteries (i.e. that 7-9ish morning slot).
Thank you
 
 
 
 
 
 

Txs @wolfandy
This setup is consuming me 🙃
I've rethought the approach a bit
1) use some battery in the morning to heat geyser
2) use some spare PV to heat geyser during day
3) limit battery usage for geyser in late afternoon
Intention still to minimise Eskom usage, maximise battery usage and get PV to charge. Barring some exceptions (few cloudy days, 4h loadshedding etc) I think 30% SOC should cover me for 2h LS.
Does setup look ok? If I understand correctly, the grid will only charge the batteries from 5pm-8m if SOC below 95%. Rest of day batteries will only be charged by solar. 
Thank you
 
 
 

Txs @wolfandy
This setup is consuming me 🙃
I've rethought the approach a bit
1) use some battery in the morning to heat geyser
2) use some spare PV to heat geyser during day
3) limit battery usage for geyser in late afternoon
Intention still to minimise Eskom usage, maximise battery usage and get PV to charge. Barring some exceptions (few cloudy days, 4h loadshedding etc) I think 30% SOC should cover me for 2h LS.
Does setup look ok? If I understand correctly, the grid will only charge the batteries from 5pm-8m if SOC below 95%. Rest of day batteries will only be charged by solar. 
Thank you
 
 
 

In my setup, if I tick the "limit to load", then the non-essentials does not get pv power at all, which in my case is only the geyser.  So I charge the battery normally and on sunny days it is fully charged before 12h00 and the washing is mostly done.  I then untick the "limit to load" and then switch on the geyser to run on pv until hot.  Sometime I will top up again later in the afternoon, after 15h00 whilst there is still some sun left to run it.  In the cases where I have excess sun, but it is less than 4kW, I still do it as I will rather take 2kW from pv and only 2kW from Eskom than all 4kW from Eskom.

Thank you for the comments, appreciate. Steep learning curve!
I definitely want to minimise Eskom usage, but also keep in mind having enough SOC to deal with loadshedding. That makes sense now and I will change SOC limits during the evening to 30-40%. Also untick Grid, as I want PV to charge the batteries. 
Just to confirm, is there no way to prevent battery going to geyser? The biggest issue is the 5-7am slot, cold geyser with 4kW element. Per above I can prevent the geyser from using battery by setting SOC to 100%, but that means I'll have to use grid for everything else too (essential)?
If I tick "limit to load" would that also disable the ability to send surplus solar to non-essential?
Thanks again!
 
 
 
 
 
 

I am not aware of such a feature.  However, the way I do it is to just switch it on and see how much it produces.  My other way in doing it is to change the setting to not discharge the battery and if the pv only gives for instance 3kW, then the balance of 1kW is topped up with grid.  Therefor in such cases it is not full pv, but also not full grid.  Still a saving.