How does grid tied solar “know” when the grid is down?

[DougSpindler]

Trying to learn/understand how grid tied solar knows when the grid is down.  Looking for a simple technical explanation. 
I was told a contactor’s coil it attached to the grid.  When no power is senesced on the grid, the contactor contacts open isolating the solar panels from the grid.  But if the panels are feeding electricity to the grid how does the contactor’s coil know if the electrons are coming from the grid or solar panels. 

Thanks

[trobbins]

One method is to stop the PV inverter for a short duration, and sense what the grid does during that short time.  That works well for a stiff grid but has caused some grief for SMA users on weak grids or where other grid connected inverters don't like such  'glitchs'.

[NiHaoMike]

A negative impedance is inherently unstable, pair that with voltage/frequency limits and it simply won't stay in bounds for more than a split second without a voltage source keeping it in check.

[Faringdon]

...though all the many 100's of solar inverters pumping power into the grid would keep the grid up anyway?

[trobbins]

...though all the many 100's of solar inverters pumping power into the grid would keep the grid up anyway?

Perhaps best not to hijack the thread.

[Faringdon]

Thanks, I thought it was relevant to the original question.

[DougSpindler]

So does anyone really know the answer?
Good point about other solar panels pumping electricity on to the grid.  How do they know the “grid iis down”? 

[Faringdon]

After thinking back, I think that if the 100's of connected solar inverters are keeping the grid up...then somehow they fail to do so within a certain tolerance...and so the inverter kind of estimates that the grid has gone down, since the mains period is not 1/60 ms +/- x.

[bdunham7]

So does anyone really know the answer?
Good point about other solar panels pumping electricity on to the grid.  How do they know the “grid iis down”?

Yes, the answer is well known and has been more or less stated in this thread.  The inverter can sense the voltage on the output and since the stereotypical grid-tied inverter is programmed to just put out the maximum available current, if you remove the grid the voltage will quickly go out of bounds--high for an open circuit (say just your house is disconnected) and low for an overload (say where the substation disconnects and now your inverter is trying to power the neighborhood).  IOW, the whole premise of a grid-tied inverter is that it has to be connected to a low-impedance (stiff) AC voltage source with the voltage within the specified limits, otherwise it trips off for at least 5 minutes.

In theory, if you were to carefully balance load and supply, you could trick the inverters into remaining operational even though they aren't connected to an actual grid.  Schneider has/had a product that essentially did something like this, using a battery, inverter and charger combo to simulate a grid in an isolated system.  As long as it is able to keep the voltages in range, the inverter(s) should keep working.  Some newer inverters and microinverters can regulate their output and operate in a specific island mode, provide reactive power and other neat tricks.  Absent any of these solutions, a standard grid-tied inverter will trip off immediately without the grid.

[bdunham7]

...though all the many 100's of solar inverters pumping power into the grid would keep the grid up anyway?

The inverters are essentially a current source.  You'd have to exactly balance load and supply somehow. 

[Faringdon]

Thanks, and i think each inverter kind of randomly tries to change the frequency a bit, and so without the real grid there, it still all goes off frequency, even though there are many inverters...at least  i think thats it....

[DougSpindler]

Thank you all - Great explanation.

[tszaboo]

One method is to stop the PV inverter for a short duration, and sense what the grid does during that short time.  That works well for a stiff grid but has caused some grief for SMA users on weak grids or where other grid connected inverters don't like such  'glitchs'.

You don't even have to do that. It's enough to try to make a bit of imaginary current and measure if the real current lines up or not.

[trobbins]

One reference for those who are into detail: https://iea-pvps.org/wp-content/uploads/2020/01/rep5_09.pdf

SMA have various docs discussing their techniques for anti-islanding detection, but generally are fuzzy on the exact mechanism.

For small domestic PV inverters in a domestic environment it all works well, but caveat emptor for large commercial/industrial when such a PV system is based on domestic or small commercial inverters.

[AlbertL]

One method is to stop the PV inverter for a short duration, and sense what the grid does during that short time.  That works well for a stiff grid but has caused some grief for SMA users on weak grids or where other grid connected inverters don't like such  'glitchs'.

You don't even have to do that. It's enough to try to make a bit of imaginary current and measure if the real current lines up or not.

Yes, I'm pretty sure that's how it's done.

[David Hess]

The grid has a very low impedance which is what regulates the grid voltage.  The grid-tie inverter has a high output impedance and acts as a current source with an output current which is proportional to the grid voltage.

If you had a low impedance AC voltage source then you could fool the grid-tie inverter into remaining on and supplying power.

[Jeroen3]

There are multiple ways:

1. the inverter switch opens: the current will stop/voltage will rise.
2. the grid fails: a sudden jump in current. (you're feeding the entire street)
3. the grid islands: a sudden jump in frequency or power factor. This is also called ROCOF and Vector Jump.

It's very difficult to fool an grid-tie inverter. You'd need a 4-quadrant inverter for that. As far as I am aware, only Victron sells those that support islanding with batteries an grid-tied solar installations.

Yes, but what if I have exactly the amount of load?
You need to match the exact load, including the exact power factor of 1. Grid tied inverter operate in power factor mode.
Any reactive current on your island will not be supplied by the grid tied inverter.

One method is to stop the PV inverter for a short duration, and sense what the grid does during that short time.  That works well for a stiff grid but has caused some grief for SMA users on weak grids or where other grid connected inverters don't like such  'glitchs'.

You don't even have to do that. It's enough to try to make a bit of imaginary current and measure if the real current lines up or not.

Yes, I'm pretty sure that's how it's done.

The standard 50549 does describe a resonant detection method but it does not go technical. I suppose you could do that. You cannot interrupt power because you'd be injecting harmonics.

[tridac]

Just compare total load power with power coming from pv array, with the difference being that coming to or from the grid ?. I would think most grid tie inverters provide such info to the customer on a real time basis...

[niels747]

If the grid island is pretty big, and has for example big factories with large synchronous, and induction motors running, these could stabilize the frequency a bit. The motors would also be able to supply the reactive power that the solar inverters can not. If supply and demand would coincidentally be exactly in balance (on each phase even) there might be a chance. The chances of this are near 0 of course, but just hypothetically speaking, I don't see a reason why it couldn't work for a little while, until the frequency drifts out of spec and the inverters shut down.
Correct me if i'm wrong.

[bdunham7]

I don't see a reason why it couldn't work for a little while

It will work for a little while.   

[jbb]

I think SMA do the Sunny Island battery <=> inverter series

[trobbins]

Many battery inverter manufacturers provide grid-forming mode of operation type inverters.  But they can't be normally connected to a domestic grid due to safety and anti-islanding.  Normally a PV inverter has to be certified for use on a grid, with country specific parameters for how and when it must disconnect.