Has islanding of solar generation inverters on a utility line ever happened? 2

[dwlawrence]

When the solar PV generation industry talks to utilities about connecting projects the conversation sometimes seems to be about whether the internal protection of the inverters is adequate to detect faults or loss of supply on the utility grid even though the inverters are certified with advanced protection functions and anti-islanding algorithms. The concern is anti-islanding and that where multiple inverters are installed on a line they will fool each other or that a rotating generator will maintain what looks like the grid. Does anyone know if such an event has ever happened in the real world and if so for what duration? Apparently it's theoretically possible but I have heard it said it's tough to even simulate in a lab under controlled conditions to certify the inverter for AI functions.

 

[MikeHalloran]

Did the utility with whom you were arguing happen to mention a particular standard certification test procedure?

If one does not exist, it would be in the PV generation industry's interest to write one to which they have a fair chance of actually testing and certifying, and to get a utility trade association to buy off on it.

So, which dog is yours?


Mike Halloran
Pembroke Pines, FL, USA

 

[dwlawrence]

There are a number of standards but IEEE 1547 is the benchmark. The point the utility makes is that the inverters are tested on their own not with other inverters or generation even though 1547 implies the anti islanding test simulates other inverters on the feeder. But whether it does or not is the unknown. That's why I ask the question about whether islanding has ever happened. The utility fear is a remote possibility. How real is the risk is what it boils down to? Even if there is not one known incident I suppose the comeback will be it is still possible.

 

[davidbeach]

Remember you're dealing with people whose job it is to be concerned about and protect against rare but significant events. The UL standard specifically tests one inverter against one simulated utility connection; no other sources in the potential island. Slow collapse of the island can be just as bad as a stable island. If that slowly collapsing island gets closed back into when the utility source recloses, any rotating machines in that island (motors) will be closed into out of phase. It is entirely appropriate that the onus of proving no harm falls on the entity wishing to interconnect. No other customer should be subjected to power quality issues just so somebody else can have a cheaper interconnection.

 

[dwlawrence]

So let's just follow that thread. Let's assume the issue is the quality of the interconnection protection and the cost. Can the protection in the inverters be upgraded to provide that functionality? They are made to standards but do the standards adequately reflect what utilities need?

 

[stevenal]

http://prod.sandia.gov/techlib/access-control.cgi/2002/023591.pdf

A few years old, so other methods may exist. I find it interesting that an algorithm forcing a positive frequency bias can be canceled by another forcing a negative one. Definitely room for further standardization.

 

[davidbeach]

There have been modeling studies that show that a mixed fleet inverters will tend to slow the individual anti-islanding while, at least with some manufacturers, the more inverters the faster the anti-islanding will respond.

 

[dwlawrence]

David: Are those modeling studies internet accessible?

 

[davidbeach]

Some have been published, try IEEE xplorer; but the more interesting that I've seen have been conducted under all sorts of NDAs for specific interconnection requests.

 

[dwlawrence]

Stevenal: Just went through the Sandia report. Now I read it I suspect that's the source of the utility concerns. As you say it is now 10 years old and much has changed in that time. I will try checking the references from DB but of course the ones that would be the most interesting would be those NDA ones. If I infer correctly from DB's comments there would still be issues with multiple inverters.

 

[tinfoil]

Look at it from the Utility POV:

The system was designed and run for 100 years with a centralized generation - radial distribution paradigm.

The historical goal of the utility is to deliver to all of its customers a 115.00 VAC 60Hz perfect sine wave on a 24/7/365 basis (or 220.00 VAC 50Hz, OR... but you get the idea).
When something breaks, we need to SAFELY deal with it, and get everyone back to the ideal ASAP.

Anything that makes this harder is bad.

Mixing in Distributed Generation through a plethora of solid-state inverters that add short-circuit energy, less-than-perfect sine waves and are worse-than-useless when the main source goes away qualifies as 'makes this harder'.

Adding a non-zero risk to restoration workers (no matter how small) makes this harder.

Having 15% of a feeder's generation vanish with resultant system voltage swing BECAUSE A CLOUD ROLLED OVER makes this harder.

Please do not be surprised when utility folk require robust precautions to interconnect!

 

[tinfoil]

BTW, I forgot to mention that the risk to restoration workers is in the 'risk of fatality or critical injury' category.

I never, never want to have to talk to the family of a dead worker to explain that 'the studies revealed that the risk was quite low.'
Do you?

 

[dwlawrence]

Full disclosure. I am ex utility and am asking the questions as devil's advocate for others that think the utility is being over conservative. The original question was whether there were real world examples to make the utility positions understandable.

 

[rcw retired EE]

I had a wood fired steam generator island the city of Susanville, CA despite the 27/59, 81 O/U frequency and other anti-islanding relaying. Load at Susanville was close to the plant output when the only utility tie to Pacific Gas & Electric tripped.

The plant nor the utility relaying sensed islanding for a few minutes. (One operator claimed 20 minutes). The plant was in commissioning at about 20% load and operating on steam pressure control. The controls were not designed for islanding. It was only the second or third time the 30 MW generator had been synchronzied. When the feed pump VFD started hunting and making funny noises, the start up engineer aborted the run and asked the operator to trip the generator breaker. They were surprised when the plant and neighborhood went black. He was accused of opening the 60 kV breaker by mistake but its red light was still on.

Most of us would have bet that could never happen.

That was a couple decades ago. Since then improvements in rate of change of frequency relaying, multiple frequency settings plus utility-plant breaker communications make this scenario even more unlikely.

But with a solid state inverter locked onto 60 Hz.... I would not bet that islanding could not occur.

 

[LionelHutz]

I recall there being a large recall on a manufacturers solar inverters, likely close to a decade ago now, due to the anti-islanding not always working. Possibly Xantrex units? And possibly one of the reasons Xantrex ended up being sold to Schneider?