I wrote a runbook for the backup/emergency power setup and, of course, tested it under realistic conditions.
I also tested controlling the Hoymiles inverters via frequency increase in island mode. It’s a little rough, but it works.
There’s that other thread here where there are problems even with a Fronius and a full battery…
That is generally down to the system sizing. My system operates permanently in island mode and is not connected to the grid either. Frequency control with both Fronius and Hoymiles (yes, it’s rough) works without any problems. The only manufacturer I know of where it doesn’t work properly is Growatt. Their units start generating reactive power like crazy, which the MP2 doesn’t really like. If the battery is sized large enough (and as specified by Victron), then there are absolutely no problems here. The issue with the 20 ms is also a maximum. In practice, the switchover is usually significantly faster than that. I have never experienced any problems with devices connected downstream of the MP2 when switching to backup power or back again.
This is based on experience from several hundred installations I have built.
In that case, things are a bit different.
The “problematic” part is generally the moment of switching from grid to island mode, because the PV inverter’s energy briefly has to go into the (full) battery until the MP raises the frequency and the PV inverter reduces its output.
Although some PV inverters may also respond to the grid failure by shutting down and then have to restart first.
Rather than using a 15kW PV inverter on a 15kVA Victron system, I would also be more in favour of reducing the AC PV output and routing part of it to the battery via MPPTs.
But even this configuration cannot respond to a sudden grid failure without a delay; in this case too, energy briefly still goes into the battery.
That is why it is important that the battery is not too small (which would be the case here with just two Pytes V5 units), so that even at 100% charge it can still absorb energy for a few seconds without having to shut down immediately.
Why wouldn’t that be sufficient?
You can no longer use the electrical loads system without restrictions, but that is generally not the aim either.
Large loads such as an EV charger, instantaneous water heater or cooker/oven can be wired so that they no longer operate during a grid failure, and then the system is perfectly sufficient for the rest of the installation.
Of course, if you want to continue using the electrical loads system without restrictions during a grid failure, the system has to be planned and sized differently, which naturally costs considerably more, but it is possible.
Here too, it naturally depends on the system.
In a normal household, problems are probably relatively uncommon.
If in doubt, this has to be tested during commissioning, and an additional separate online UPS may need to be provided for extremely sensitive devices.
If I remember correctly, I think we encountered this once in a metalworking company, where the control system of a CNC machine was very sensitive and went into fault mode when switching over.
Not “absolute security”, but security that is entirely sufficient for the vast majority of applications.
Since it isn’t quite perfect, Victron also states in its warranty terms, for example, that the devices should not be used for life-support machines.
The new Microgrid might be interesting here, as the failure of one subsystem does not cause the entire system to fail.
Although the potentially widely fluctuating voltage/frequency could presumably cause problems here.
Island operation is a different matter altogether and makes many things easier.
There’s no need for either switching over during a power cut or for the MP2 to switch back to 50 Hz for synchronisation when the grid returns; the inverters connected to AC-Out will then see 50 Hz again as well.
With grid-following systems, it’s usually not disconnecting from the grid that’s the problem, but reconnecting to it.
But yes, things have become somewhat easier nowadays.
Indeed. I only wanted to illustrate the clean “control” aspect. It works quite simply. If I then also use the standard German grid code, it takes a while for the inverter to reconnect and deliver full output again. If the system is operating in MG50 mode, you simply need a slightly larger safety buffer in the battery than you would otherwise.
… so should I perhaps only charge the battery to a maximum of 95%?
I know that the battery isn’t supposed to “sit around” at 100% for hours on end.
I want to tackle this with Home Assistant.
Or does anyone have something ready-made for Node-RED on the Victron?
Can’t you just lower the maximum charging voltage a little?
The only thing that matters is that the BMS doesn’t shut off, right?
Just lower the maximum charging voltage a bit …
You DON’T need NodeRed or anything like that
Yes!
I’m doing this right now because I want to get the charge down to 80%. That’s not a problem with the current amount of sunshine.
Charging isn’t optimal then, because it may not reach 80%.
It would be muuuch better if you could set the “state of charge” on the Victron!
That’s why Home Assistant.
What exactly are you getting at ???
There is no maximum SOC limit .. that can ONLY be controlled via the charging voltage…
But I have the feeling we’re talking about different things…
With an ESS, the battery is always charged first .. and then the surplus is fed into the grid… and if your household loads are sooo large that the battery doesn’t get fully charged in sunshine .. then the solar system is simply too small..
I know there is no SOC limit — unfortunately! That’s why I’m using the charging voltage instead — currently 53.9 V, with a default of 55 V.
That works reasonably well.
When the voltage is higher — 55 V — the battery reaches full charge more quickly and reliably.
Here’s a screenshot from my neighbour with 55 V (PV hasn’t been integrated into Victron yet). This shows what I mean.
I don’t understand why Victron isn’t able to add a maximum SOC.
At the moment, I don’t want to charge to 100%, but deliberately to 80% — and I want that done as soon as possible!
The plan is to implement “solar-yield-dependent” charging, so that the battery is fully charged to 100% by sunset.
Why would Victron do that … the systems are used for zero feed-in and to provide grid power … and that’s only possible this way.
And your “dependent” charging will get really complicated … because sometimes the forecast is wrong …
Sometimes the house needs more than planned … and the maximum SOC is effectively already there through the charging voltage. And the normal SOC is simply too inaccurate for some people …
The whole discussion has drifted somewhat off topic at the moment. The actual question was whether a Sungrow can be connected behind 3 MultiPlus units with Pytes.
It is clear that, in island mode with a fully charged battery, the Victron increases the frequency, forcing the Sungrow to reduce its output.
However, the frequency curves of the two systems do not match: the Sungrow disconnects at 51.5 Hz with 40% P₍ref₎/Hz (s = 5%). In other words, 40% of the maximum 15 kW is still being supplied when it is abruptly switched off.
The question is: can the Victron (which would be at 0 at 53 Hz) cope with this—and would anyone implement this under the conditions described?
I have attached Sungrow’s response so as not to make the text here unnecessarily long.
SUNGROW Can the frequency curve in the Sungrow be adjusted—and if so, how? Any experience with a Sungrow ‘behind a Victron MultiPlus’.pdf (87.7 KB)
It would surely be much easier simply to adjust the frequencies in the inverter or ESS Assistant. The Sungrow runs normally with the country code.
But as far as I know, this also works with the standard settings. I adjust the frequencies in the ESS Assistant, but I can’t judge whether that actually works better.