ESS and Charger behavior (Backup generator for Power outtage)

[kevin88252]

Hello there
Since i have spent a lot of time on research and could not get a answer i am hoping for some clarification here.

I have a setup with 6 multiplus5000/48 and 28kwh storage on Pylontech Batteries. 2 450/200 MPPTs attached to 7 PV Strings.
The multis are configured as ESS and everything works as expected and is great so far.

But there is one thing i do not understand:
In case of a longer Power outtage i want to use a Generator to charge the batteries in case the PV power is not sufficient. I am Talking about serious generators, one is a 20 kVA (Yanmar engine with a MeccAlte Generator) the other i tried is even bigger, 100 kVA (GCB engine with Leroy Somer generator) so high quality stuff with lots of power.

The problem i am facing is, if the Multis (or Quattros, tested both) are configured as ESS they do weird stuff when charging the batteries; the 20 kVA starts to oscilate with Voltage and Frequency and the Circuit breaker is triped within a few Seconds) the 100kVA also oscilates with Voltage and less Frequency but the Current goes up like hell and the Controller of the Generator opens the GCB and shows a Error message (sometimes voltage, sometimes frequency).

Thinking outside the box, both gensets can do grid parallel i was thinking “hey lets give it a try and connect the genset to AC out” (again, the gensets can do grid parallel meaning they synchronise their voltage and frequency with the inverter, close the circuit breaker and gently start to push energy)

without ESS this works fine, even if i don’t think the inverters are made for this setup (but if they can handle a PV on AC out, they can also handle a genset don’t they?) the genset synchronises and starts to push power in the system, the power goes backwards in the batteries (just make sure the BaseLoad of the genset is less than what the batteries and inverters can handle ;-))

but if i enable ESS i have exactly the same behavior. They can start to generate huge reactive power. The frequency of the genset starts to oscilate (genset shakes) until the controller disconnects immediately because it detects a “unstable grid”.

So why is the charger behaving differently if it is setup as ESS than other? And what does it do exactly?
what would be the right setup if i want to charge my batteries with a Genset during a power outtage? Is there any other solution to feedback to grid the exess PV power without ESS configured?

Btw. the issue is only there if there is no grid, if i connect the Genset no matter to AC out or AC in, and do a Grid parallel application everything works (but there is no reason to let a genset running when grid is available )

thanks in advance for your reply and help

[ricardocello (Rick)]

Probably LOM issues.

https://www.victronenergy.com/live/multiplus_faq
See Section 1.5

https://www.victronenergy.com/live/ve.bus:grid-codes-and-loss-of-mains-detection
See Q3

A muliplus only has 1 AC input, which means reconfiguring it to use a generator.
This is very inconvenient and potentially unsafe.
Quattros have two AC inputs, and LOM is configurable independently for both.

Have you considered a big AC to DC charger to convert the generator output directly to DC to charge the batteries and run the loads through the inverters?
Yes, it is a double conversion but totally solves your AC synchronization issues.

[kevin88252]

Thank you for your reply

If i understand LOM correctly, if it would be a LOM issue, the inverter would disconnect?! but it does not, it just does weird stuff and 3x8000va inverters will overload a 100kva generator…

yes i have already considered your suggestion about using a big battery charger but to be honest i don’t like the idea since i already have a device that is technically capable to do exactly this…
Also a AC DC charger that can handle the full power of 16kw might be a expensive thing

and with the setup having the genset on AC out i can use it as backup in case the inverter fails as another benefit…

[ricardocello (Rick)]

Assuming your Multipluses are in three-phase configuration, 2x per phase.
Assuming the big generators/alternators are also three-phase.

There is a setting in ESS for multiphase regulation to the grid setpoint.
I would try the per phase setting to see if ESS is making the power is oscillate between phases. You should also disable Feed-In.

You have to follow the 1:1 rule if you put the generator on AC out, the multis have to be able to grid form. But with 30 kVA of multis, it doesn’t sound like that is your issue (at least for the 20 kVA genny).

Let’s see what others say.

[Christian82 (Christian)]

Hello Kevin,

As Rick already stated - it smells very much like a LOM issue. I use a system of three Quattro II 5000 in a three phase arrangement with a generator to recharge the batteries if needed. 20kVA Induction generator with three-phase sine-wave IGBT-Inverter after it. (Yes, lot´s of conversion but i´m kind of a power electronics enthusiast and it negates a lot of other problems regarding stable frequency, phase balance, reative power, inrush, fast reaction to overcurrents and so on.)

But I’m digressing, although the inverter thing will become important later…

I run the generator on AC in 1 of the Quattros and it works like a charm. Grid is on AC in 2 with LOM. On AC1 LOM is disabled. At first i tried with LOM enabled, but noticed oscillating currents which i later found out were caused by LOM by constantly trying to push and pull the Quattros out of sync to the generator feed to detect if it´s still there. With the inverter setup they had absolutely no chance to do so, because the frequency is absolutely fixed. Although it worked, it annoyed me and i disabled LOM for that input.

But with a “classic” generator in an unfortunate scenario this poses the risk to escalate.

For example: LOM tries to raise freqency → unloads gen → engine reacts → LOM tries to decrease → gen draws more current → engine reacts. And if the timing is wrong, we can get a runaway feedback loop. If this escalates “enough” than the gen can detect an unstable grid or something trips - either Multi/Quattro overload or breaker. According to LOM description this can be promoted by a high impedance of the generator line. Another escalation scenario could be that an autosync-capable generator tries to sync on the Multis which are trying to unsync and so on. But in that case LOM should trip because that´s the point of it.

Because of the two independently configurable AC inputs i chose quattros.

Long story short - can you try with ESS enabled and LOM (temporarely for the test) disabled?

[kevin88252]

Thank you for your explanations.

That makes sense for me now…

I will give it a try without LOM and see how it works, also wondering now if it makes a change on the AC out charging…

Do you think the inverter also does something when grid is disconnected? As it would make sense that the frequency is not a solid 50hz
(The generator in parallel mode tries to keep constant power, meaning if the frequency goes up, the power drops (almost immediately) then it increases engine power trying to achieve the power, next frequency goes back and all the power (mechanical momentum) goes to the inverter (and ends up with a triped circuit breaker)

Would totally make sense for me so far…
But why should it do something like this when it is “grid forming”?

[Christian82 (Christian)]

If i understand you correct, the parralel mode of the generator causes the generator to try to sync on the Multis. Would make sense for me - as it means it has to sync on another source. If so, this could be the issue in its self!

Because then both the Victron inverters AND the generator are trying to sync on each other and because LOM is constantly trying to (simply said) “desync” the could well end in the described runaway scenario… Even without LOM the “trying-to-sync-on-eachother-action” could be the cause theoretically.

You could also test running the generator in standalone mode, so that it is not actively trying so sync on an another source and connect the generator on AC in. With grid disconnected and AC in configured as generator to prevent the PV from trying to feed back.

In this config we have only one component trying to sync on the other. If my assumption is right, this could solve your issue - even with LOM enabled.

Another thing - when you are trying to feed into AC out (like a AC-Coupled PV) then it could cause additional issues. And they are not LOM related, because LOM ist only for INPUT power sources on AC in and without a source, you are “grid-forming”. Exactly as you stated. But with ESS if there is no source and the AC PV feed is becoming to much for the Multis or they “like less”, then they are rising the frequency and through this the AC PV cycles down / shuts down. But a parallel genset does probably not - except it is equipped with this feature. Have that in mind. I would try to charge the batteries by configuring AC in to “Generator”, set generator to standalone and try with and without LOM.

Greetings Christian

[kevin88252]

I think a little clarification is needed about generator parallel mode:
if the genset is connected to AC in, it does not sync to anything, it works like a normal genset, running at 1500rpm no matter what happens. so no Sync issues should happen here.

if it is connected to the AC Out (or grid) it synchronizes first, then close its circuit breaker and raises the throttle little by little until it reaches the requested power (setting in the genset controller) it then stays at this constant power.

The multis may not be configured with raising its frequency because that does not work by physics with a genset (the multis can alter the frequency extremely quick, the genset on the other side has its mechanical limits there) i have a external controller telling the genset how much power it may push.

so in one application (AC In) the genset is grid forming and the multis will sync on the gen.
In the other application (AC Out) the Multis are grid forming and the genset will sync on the multis.

The LOM setting makes totally sense if the genset is connected to AC in. But for me not if it is connected to AC out… (or do the multis also have a kind of anti Islanding for AC out?)

anyway, i’ll give it a try without LOM and see how it performs then…

[kevin88252]

I did some tests this evening with 3x 8000 Quattros and a 20 kVA genset.

The genset is a standalone with a mechanical speed governor; no load 51.7 Hz, full load 49.5 Hz. So i only did tests on AC In.
I’ve setted the target power to 16kW (100% load)
I took the German VDE grid code and tried several LOM settings:

• Above selected gridcode plus LOM B:
  Quattros Sync, close the breaker load genset with about 2 kW and disconnects, process repeats every minute
• Above selected gridcode, no LOM detection:
  Quattros sync, close the breaker, load genset and the load starts oscillating between 2 and 6 kW quickly (bounces up and down with the load about every second or even faster)
• Generator (Grid code “other”, no LOM detection):
  Quattros sync, close the breaker, load genset and the load starts oscillating between 8 and 15 kW quickly (bounces up and down with the load about every ten seconds)

when i disable ESS and use gridcode None, everything works as expected, the quattros sync, and load the generator to 16kW, the current on the generator stays within ± 0.5A

What i haven’t tested is gridcode VDE and no ESS, will do that on Monday…

anyway i must oversee something or am i really the only one who wants to use a generator as backup on a ESS?

is there any way to feed back DC PV power to the Grid without ESS?

Checkout these videos: the first is no LOM the other is Generator, no LOM.
on the first video you can see that the genset never gets loaded properly. The display does not update quick enough but if you hear carefully at the engine noise you can hear that the load is not even close to constant.

[Christian82 (Christian)]

Okay - that´s intresting. And crazy.

I´ve set my AC1 Input to the german VDE grid-code with LOM disabled and it works fine so far. I´ve limited the charge current to 50A in VEConfig. Worked without limit as well - it´s only there at the moment because my voltage dropped slightly below 220V at full power. (Different story - homemade capacitor excited induction generator, working on a automated capacitive reactance compensation with static switches by now that will solve that )

If i disconnect the grid on AC2 an connect AC1 to the generator, then it takes around 30-60 seconds for the Quattors to sync and close relays. Then they slowly start to ramp up the power - it takes another good minute to reach the desired power draw / charge current. I have three analog amp-meters on the generator - if you look closely, you can see small oscillations repeating approx. every 1-1,5 seconds. But very small - 0,5-1 amp i would say. Engine sounds smooth and on the GX display there arent any fluctuations worth of note.

I have a 64kWh battery an have done hour-long charging runs without problems.

My generator, due to the aforementioned inverter, puts out a perfect sine-wave. Your quattors seem to have some kind of problems regarding power quality. But i´m confused - you have a large and professional generator that should have absolutely no problems delivering clean-enough power.

An idea for further narrowing down the possible cause:
“Simulate” a perfect generator by hooking up the grid on AC1 configured as generator input on the GX and with VDE grid-code without LOM (be careful). Possibly with reduced charge power depending on your supply. If possible attach a clamp meter on one of the phases to look for fluctuations. If this runs and charges as expected, then it has to be a power qualiy issue. I did that for a short test before i connected my real generator the first time to be sure that my config worked in principle and if later there are issues that they then have to come from the Generator.

If Above worked, next thing i would try is hooking up an oscilloscope (with suitable differential probes) to the generator power and have a look on it with and without the quattros attachted and trying to charge. It has to be something, distortion, harmonics that build up or whatever.

Here my generator/inverter waveform (L1/N) when charging with 8kW. (Ignore the 144V - probe setting was wrong).

[kevin88252]

Thank you very much for your input.

After a lot of research i have found out that most people use inverter generators (and if i got your reply right, you are also using Kind of a inverter generator?)

Of course inverters behave completely different to AVR, they always have stable 50 hz no matter what the engine RPM does…

I will do some tests tomorrow using a high speed fluke power meter with recorder…

Also i want to find out if it is because of ESS or because of LOM / grid code…

[Christian82 (Christian)]

Hello Kevin,

yes, i´m using an inverter generator - with the following scheme:
self-excited induction generator → chokes → 3ph thyristor rectifier → dc-link → 3ph igbt-inverter - lc-filter - dyn5 isolation transformer.

I think it is ESS related. If i guess right (but I could also be completely wrong not knowing the exact inner workings of that ESS gird-parallel thing), without ESS the Quattros are behaving more passive - they use what they get on the input, step it down, rectify and charge. With ESS i presume they are using a grid-parallel active method, where they create there own sine-wave synchronous to the input which they modify in amplitude and phase to control the direction of flow of active and reactive power. If my assumptions are right, then the power quality requirements would be much higher in ESS mode because every mismatch in waveform shape would result in equalization currents between the quattros and the genset.

If it comes from a bad waveform than i have suggestion for a test that might sound completely crazy but could actually work. I had once a case that the servers etc. in our company had to be supplied for serveral hours by generator due to transformer maintenance and the UPSs didnt take the power because the waveform was too bad. Load was approx. 5kW and generator was 16kVA, synchronous, AVR. No chance. And it had to - whatever necessary.

Now the crazy mac-gyver solution part - i took a 5kW (or 7,5kW?) 3ph induction motor, compensated it with three capacitors in delta to around cos-phi 0,98-0,99 and connected it parallel to the genset. Free-running, whithout load. This contraption forced and ironed out the distorded waveform into shape “good enought” for the UPSs to now accept the generator power. And it ran reliably the whole day

I’m looking forward to your test results and whish you luck.

[kevin88252]

Yes it makes sense that the inverters behave completely different in ESS mode…

Just what i do not understand is why don’t they change their behavior if they “know” that the input is a genset?

And why do they also behave differently if the genset is connected to AC Out?

Btw your issue with the UPS not accepting the genset sounds like your ups created a big capacitive reactive power, generators can’t handle that…
Or the “small” genset i assume has a mechanical speed governor and gave a to high frequency at no load…

This at least should not be the issue with the generators i use as they are electronically controlled and run at a constant speed of 1500rpm…

Again my question, is there any way to feedback DC PV to the grid without ESS (could add a PLC telling the Cerbo over modbus what it should do and use it like a ESS)

[kevin88252]

Allright, i have some news (at least for the use on AC In)

Grid Code None and ESS configured works without any issue, as soon as i configure a Grid code it does not work, the load on the genset is very instable and sometimes 100% asymmetric!

i will continue doing a test on AC out without grid code and ESS enabled soon to see if it behaves the same way…

i guess nobody will answer the question “why is it not possible to use a generator as backup if i have a Grid code enabled?”… (i mean what is the purpose of having 2 AC inputs then?)

Is there a way to change the grid code remotely? so i can set it to none during a outtage…

[Christian82 (Christian)]

Ah, interesting.

Perhaps dumb question - did you alread try this arrangement?

Generator without LOM, UPS deactivated and wide input range? I know - on some systems (like in mine) not all options are available.

Perhaps another dumb question - have you already tried with ESS, grid code and the quattros set to “charger only” mode? I haven´t on my system - but then they should go into passthrough/“dumb-charge” mode. That could be remotely enabled and disabled. It would have to be switched in the following order to prevent power loss on the output:
first “inverter only” → then changeover to generator → then charger only to allow charging → then, when finished back to inverter only → changeover to grid → then “on”.

Regarding feed in AC out - it´s different because the quattros are the grid-forming part and there is no LOM and no need to try to sync on something else. But AFAIK, AC PV control via frequency control is then active, if AC-PV on output is set as enabled as a must-have safety option.

Regarding your question - i dont know about a method for simulationg ESS behavior like DC PV feed in without ESS by an external PLC or so. Neither i know about a method to switch grid codes remotely.

I think the question “why it is not possible to use a generator if i have grid code enabled?” is not easy to answer - becuase it is. In my case it works and victron is even promoting it. The difficult part is, why your system is so picky about it.

Btw. - the UPS-thing was power-quality related and had nothing to do with capacitive reactance from he UPSs. We checked that out by using a power quality analyzer that told us voltage and freq. were in the ball park region to work and power factor was ok, but the waveform was crap and was then partially pulled into shape by the misused freewheeling motor.

[kevin88252]

I tried a lot, turning off the UPS function is not really a option for me, anyway according to my research this is only helpful if the inverter fails to sync on the AC input.

Putting the inverter in charger only mode will cut off the AC out…

The problem seems to be with the combination grid code enabled and AVR genset. I have not found anyone in the deep web who did this… Most people use either no grid code or inverter generator…

I will try to get a official answer from my supplier.

And i will absolutely try if it works with no grid code and a synchronised genset on an out. I would not be impressed if the charger does not care if the current comes from AC out or in and behaves identical

[Christian82 (Christian)]

Hi. I think the assumption in Charger Only mode is not correct. See here:
5. Operation

The document is for a MP but should also accord to a quattro.

Charger Only mode does not break the connection to AC out 1 - beause there is no relay/breaker to break:

It goes into dumb passthrough.

And - disabled UPS function does also lower the requirements for power quality:

Although it prolongs the time needed for switch over if input fails. But it is often helpful in a situation where your system is connecting and disconnecting sporadically or in a loop which you stated. Would give it a shot, if you havent already.

Reasearched myself a bit and also found no reliable real world experience with sync AVR gensets + gird code + ESS. Whish you luck to find a solution and get it working.

[kevin88252]

Ok that makes sense… But anyway i don’t think that this is really a solution, no matter if it works or not…

I just had another idea, do you know if AC in feedback still works with a external grid meter but without grid code? (If i have loads directly on the grid and want to feed them from the multi/Quattro)

[kevin88252]

Nevermind, i should RTFM before asking questions

[BjoernK (Björn Kellermann)]

The Quattro has two AC_in to connect to grid and generator withour reconfiguration on grid failure.

On recent MP2 with a relais for AC_out_2, you can configure the behavior of AC_out_2.
If grid fails, I keep AC_out_2 connected until SOC falls below 20% and connect again on SOC >30%. This is done in VE-Configure using the relais assistant.