New Solar PV System - Connection to Consumer Unit

[SimonH2]

Yes indeed, once upon a time it would have been done manually - and I recall doing that at uni in the machines lab with the three lamp method (two bright, one dark).
These days there probably isn't anyone around to do it manually - all remote control and monitoring. Units to automatically control the throttle (and field) until the genny is in sync and close the breaker at the right time have been around for decades.

 

[bernardgreen]

Believe it not the three lamp method is still used as a fall back ( fail safe indicator ) method on some marine generators.

 

[JohnW2]

There's definitely some gap in my understanding that's escaping others. Maybe it's a bit basic! .... So to pick an extremely naive example, imagine pair of DC alkaline batteries at 1.5v. ... They are each powering a small load and have a cross connection to parallel them. ... The batteries include anti islanding protection, so if the cross connection were lost, the anti islanding would shut down the battery to prevent it service the small load. ... The issue is I don't see what would change at the broken link to alert the Anti islanding that it would disconnect.

That is, of course, more-or-less the very conceptual very point I have been making and a similar "I don't see what would change..." that I have been talking about - hence my question about how it is done.

As I've said, I could understand how it could be done if there were a small impedance (resistance in your analogy) in the cross connection. Sufficiently sensitive circuitry could then sense what was happening to one voltage source, despite the connection (now not zero impedance) to the other source.

Kind Regards, John

 

[JohnW2]

Why ? The inverter is solidly connected to the mains, and hence to the local distribution network, and hence to the grid. While there may be some voltage changes due to currents in the network - but there is no way the frequency at the power terminals of the inverter will be different to that of the grid.

Well, if that 'solid' connection were lost, the frequency observed at the output terminals of the inverter would (in the absence of other 'things going on') not immediately change if the grid supply suddenly disappeared - since that frequency would be the frequency of its own output (which was locked to the grid frequency up until the fault arose).

I can understand that if the inverter were connected to an appreciable amount of the grid network (following a fault which disconnected that part of the network from the grid supply), that it obviously could not service the full load on the part of the network to which it was connected - hence would either 'stop' or reduce dramatically in output voltage if the grid supply failed, even if there were no explicit control/protective mechanisms.

However, what if a network fault resulted in disconnection of just a handful of houses from the grid (several/many of which had PV installations), whilst leaving those houses' installations joined together? If that happened at a low-demand of the day (and when there was sun!), I would not have thought that it would be impossible that the PV sources could have a reasonable stab at continuing to service the prevailing loads - and, at least in the short-term, each of those inverters would be seeing the same frequency 'on the grid' as it was seeing before the local part of the network was disconnected from the grid supply.

Kind Regards, John

 

[SimonH2]

Indeed, the type of protection (detection) needed will vary by the characteristics of the generator.
As you point out, the inverter is likely to have fairly stable frequency - but it could easily be designed to try and lead the grid at all times and thus have a change in phase/frequency on LoM (I don't know, I've not been involved in designing them). Also, as I pointed out, unlike some generators which are controlled by AVR (automatic voltage regulator), a PV inverter will be designed to push as much power as it can into the grid at all times (maximise production from solar insolation) - thus unless the load does closely match output, the voltage will change dramatically and cause the unit to trip on voltage too high/low.

 

[JohnW2]

John, Bit of an explanation here: https://pure.qub.ac.uk/portal/files/868297/03-2006-JA-PS-1-020[1].pdf I have to admit I can only take a away a fleeting overview of how it actually works from that, the maths is a bit beyond me, I never understood calculus of anything but the most basic functions ...

Many thanks. At a first glance, the maths is not a problem for me, but I need to spend some time reading the paper properly.

However, also at first glance, it looks as if this may well answer my question, and somewhat support what I've been saying. This bit of text and 'phasor diagram' (and the corresponding Fig 2, in the 'islanding' state) ....

... indicates that the islanding state is essentially detected on the basis of a change in the phase angle "a", which is a function of the difference between Ef and Vt. The existence of that difference to exist relies on the presence of Xd, which is therefore seemingly the 'something' that I have been postulating has to exist between the inverter supply and connection to the grid - since, in the absence of Xd, Ef and Vt would be the same (hence zero difference between them), under any/all circumstances.

However, as above, that is all based on a very quick glance at the paper. I'll have a proper read and will comment further if necessary.

Kind Regards, John

 

[JohnD]

Believe it not the three lamp method is still used as a fall back ( fail safe indicator ) method on some marine generators.

I have seen the 3-lamps used on demonstration and teaching generators (some rather old) retained in the electricity industry.

plenty of painted iron and polished brass.

For all I know they still have them in production stations.

 

[ericmark]

I see how a solar panel detects disconnection from the grid, however what I question is the speed at which it works, be it voltage, frequency or both it will in fullness of time drift outside of limits and disconnect, it has a time before it can reconnect, so with a street full of solar panels one by one they will disconnect and stay off line until the mains has be reconnected for a set amount of time.

However what is in question is how fast it will disconnect? Because the voltage, frequency or both need to drift out of limits, we are left with the question how long will it take for the supply to drift? Will it be fast enough to automatically disconnect in the time allowed? We have 40 milliseconds for it to disconnect if we have it connected in a way where it can supply power to the load side of the RCD supplying multi-circuits.

I can't see how we can guarantee it will drift within that time, It may drift within the time 9 out of 10 times, but it has to do it 10 out of 10 times. Since the RCD will take time to trip you have not even got 40 ms if the RCD takes 25 ms then only 15 ms for it to disconnect, and the RCD could take 39 ms of course. So simply can't see how you can have the same RCD supplying the MCB to the grid tie inverter as any other circuit.

Only option is a double pole RCBO and although some makes of consumer unit will permit fitting of single modular width double pole RCBO are the only option with many consumer units. Then comes the whole idea of a consumer unit, it is a type tested distribution unit, and unless the manufacturer states you can connect a second supply to output terminals then you can't do it, as it would invalidate the type testing.

So if Wilex says you need a double modular RBCO then that is the only way with their box.

The old Polycarbonate version costs over £300 not found a metal version.

 

[bernardgreen]

it has a time before it can reconnect, so with a street full of solar panels one by one they will disconnect and stay off line until the mains has be reconnected for a set amount of time.

Large areas of solar PV panels do create problems for the electricity supply networks if they have to isolate a section for repair work.

It has happened that the fuses to a section have been pulled and instead of the section becoming dead it has remained energised from solar PV panels. Houses with solar panels had to be visited and the invertors shut down manually. ( dissing the incoming at the meter in an external box is one option )

How does Mr Brown's solar PV inverter tell the difference between 50 Hz 230 volts from the DNO substation and 50 Hz 230 volts from Mr Snith's solar PV invertor ?

 

[Lectrician]

I see how a solar panel detects disconnection from the grid, however what I question is the speed at which it works, be it voltage, frequency or both it will in fullness of time drift outside of limits and disconnect, it has a time before it can reconnect, so with a street full of solar panels one by one they will disconnect and stay off line until the mains has be reconnected for a set amount of time.

However what is in question is how fast it will disconnect? Because the voltage, frequency or both need to drift out of limits, we are left with the question how long will it take for the supply to drift? Will it be fast enough to automatically disconnect in the time allowed? We have 40 milliseconds for it to disconnect if we have it connected in a way where it can supply power to the load side of the RCD supplying multi-circuits.

I can't see how we can guarantee it will drift within that time, It may drift within the time 9 out of 10 times, but it has to do it 10 out of 10 times. Since the RCD will take time to trip you have not even got 40 ms if the RCD takes 25 ms then only 15 ms for it to disconnect, and the RCD could take 39 ms of course. So simply can't see how you can have the same RCD supplying the MCB to the grid tie inverter as any other circuit.

Only option is a double pole RCBO and although some makes of consumer unit will permit fitting of single modular width double pole RCBO are the only option with many consumer units. Then comes the whole idea of a consumer unit, it is a type tested distribution unit, and unless the manufacturer states you can connect a second supply to output terminals then you can't do it, as it would invalidate the type testing.

So if Wilex says you need a double modular RBCO then that is the only way with their box.

The old Polycarbonate version costs over £300 not found a metal version.

You are not meant to supply them from a shared RCD, you never have! Dodgy installers do.

 

[ericmark]

Large areas of solar PV panels do create problems for the electricity supply networks if they have to isolate a section for repair work.

It has happened that the fuses to a section have been pulled and instead of the section becoming dead it has remained energised from solar PV panels. Houses with solar panels had to be visited and the invertors shut down manually. ( dissing the incoming at the meter in an external box is one option )

How does Mr Brown's solar PV inverter tell the difference between 50 Hz 230 volts from the DNO substation and 50 Hz 230 volts from Mr Snith's solar PV invertor ?

Theory is without the grid the voltage or frequency will drift out of spec. This is why they need a timer before reactivating so once disconnected it will stay disconnected while all the others in the street disconnect. There have been problems where the maximum voltage 253 is exceeded and the panels close down, because of the time delay this only has to happen a few times in the day to mean panels spend most of the day closed down, if your panel is just 0.5 volt above your neighbours then their panels stop first, so some installers would set volts a little high.

So with even just 5 panels 253, 253.1, 253.2, 253.3 and 253.4 the last one would produce most of the day, first one was always locking out. To start with the output was not monitored, it was simply assumed they would produce power when the sun was out, as smart meters were rolled out, early installations would get updated. As to if this actually happened is another question? Also who tests the safety disconnection?

It is all well and good saying you need a certificate to work on solar panels, but anything mechanical or electronic will at some point fail, now I have a C&G2391 however I would not have a clue on how to test a solar panel disconnection time, so the standard EICR does not really cover solar panels, all the electrician can do if manual switch off and see they close down, where really he needs a test rig which can sink the power produced and slowly ramp up and down voltage and frequency to show the panels are working within spec.

The load bank for generators is expensive enough, without being able to control voltage and frequency, and I have not seen anything which says how often the panels need testing.

And it will only take one rouge panel to hold in for other panels to exceed the time out and start reconnecting. However as long as the houses have smart meters a signal can be sent switching off all the meters from that transformer, maybe this is why they are so keen on smart meters?

 

[JohnW2]

Theory is without the grid the voltage or frequency will drift out of spec.

It will. However, that would/could take an appreciable time, during which period there would be a potential hazard, at least to anyone working on the affected network - which is why I have been speculating about ways in which it could be guaranteed that a loss of grid supply would be detected almost immediately.

As I've said, if a significant part of the network were affected, then PV installations connected to it almost certainly could not cope with the loads, so their voltage would drop dramatically and they might just 'stop' - affording plenty of ways in which the situation could be 'sensed' (and acted upon). However, as I've also said, if just a few houses, several of which had PV installations, were suffering from disconnection from the grid supply (but remaining connected to each other), then things could be different.

Kind Regards, John

 

[SimonH2]

It will. However, that would/could take an appreciable time, during which period there would be a potential hazard, at least to anyone working on the affected network - which is why I have been speculating about ways in which it could be guaranteed that a loss of grid supply would be detected almost immediately.

Simple answer - it cannot be guaranteed ! There's a tradeoff between reliable/fast detection of LoM and avoidance of false detections causing generator shutdown.
Note that in the paper linked to previously, one of the techniques mentioned was for the "grid" to broadcast very high resolution timing information on supply phase, and for the protection detection to compare it's local phase to that (using GPS to maintain clock sync) - and hence detect if local phase timing is drifting away from the "grid". Such a technique should provide a very reliable method without false positives, even when there are disturbances on the grid.

As already suggested, it's highly unlikely that an island supplied by PV installations could remain balanced to the load for long - even if load and supply were perfectly balanced at the point of islanding, changes in load or sunshine would unbalance that eventually and then everything would shut down. And embedded generation is one of the reasons that SOP for the electrical industry is "disconnect, isolate, earth, and only then allow working on the line/equipment" - by solidly bonding the phases together, and connecting to earth, it removes any induced voltages that the line might pick up; and ensures that any generators trying to start up will meet a short circuit.
On that latter point, someone I was at school with got very seriously burned (to the extent that the medical team were surprised he survived) by a standby generator with a delayed start while he was working on incomer equipment at a factory. I don't know the details, but AFAICT the supply was isolated, someone tested for dead, and then he was instructed to do whatever it was that they were doing - then the genny started up and he got fried.

As to the RCD - again simple answer is that the inverter is not connected to the load side of an RCD supplying other circuits, therefore the inverter is incapable of supplying loads after the RCD protecting those loads has tripped. No-one would ever connect the inverter to the load side of a shared RCD would they

 

[JohnW2]

Simple answer - it cannot be guaranteed ! There's a tradeoff between reliable/fast detection of LoM and avoidance of false detections causing generator shutdown.

Given that little, if anything, is 'perfect (i.e. 'guaranteed'), I have to agree. However, as you partially go on to suggest, I would have thought that some of the techniques mentioned in that recently cited paper would probably usually detect a loss of grid supply within a very short period of time (quite possibly only a handful of cycles) - and, I would have thought, plenty fast enough to remove the hazard before anyone started working on the network. Even if they were working on it (initially 'live') at the moment of power loss, there would presumably be a period of 'testing for dead' (and maybe, as you say, 'shorting things') before thy actually touched anything potentially live.

Kind Regards, John

 

[SimonH2]

I rather think that electrical industry procedures will assume that a system is live until it's been proved dead and earthed - on the assumption that anyone can make up a widowmaker and backfeed the network, or an embedded generator could have faulty protection, or ... I doubt that anyone in their right mind would just switch off and then assume that the circuit is dead and safe to touch !
EDIT: Though it's not just a case of people working on the circuit. There's a fault, a section of network is islanded - if it manages to stay live due to embedded generation, then without synchronisation there could be "a bit of a bang" when the auto-recloser tries to switch the supposedly dead section back on after a few seconds