Islanded Generating Station 6

[Mbrooke]

How does an Islanded 600-2000MW generating station behave? What does typical critical clearing time look like? What is the voltage and frequency divergence during and after a fault for various fault scenarios? How is active and reactive power dispatched/controlled? How does generation behave during trip and reclose events? And how do you "block load"?

 

[FreddyNurk]

Whilst not directly related, I'd be interested to know the largest islanded systems in existence.

That said, for a single generation source network, I'd expect that the load would dictate active / reactive power, and the extent of dispatch would be the amount of spinning reserve to be held. I would have thought that 2GW would be large enough that at least one other source would be present on the network, but this is a size I'm not familiar with.

EDMS Australia

 

[ScottyUK]

Is this question in relation to the behaviour of a grid-connected station which is suddenly islanded from the system, or a station operating normally in islanded mode?

 

[Mbrooke]

Both. And what is done to the latter so it can run continuously in an islanded mode.

 

[Mbrooke]

@FreddyNurk: I disagree, it is very relevant and on topic. I've been wondering all day.

 

[FreddyNurk]

Almost all of my experience has been with islanded systems, although the largest one I'd seen was about 5MW, hence the consideration.

In order to operate an islanded system, the source needs to be able to:
[ol 1]
[li]Supply power between the minimum and maximum required by the connected load[/li]
[li]Maintain voltage and frequency within nominal limits, including any specific limits for step load and load rejection[/li]
[li]Be capable of being blackstarted[/li]
[/ol]

Its beyond my personal experience, but the hydro system in Russia that had a catastrophic failure had specific power bands that weren't meant to be operated in, this can cause a problem if the required output is within the required load. I'm sure that there are means of configuring the generating plant so that this isn't a significant issue, it'd be much easier to avoid in a grid connected situation.

Connection of the load in stages is generally required, whether this is managed by specific feeders, or the network itself is sectionalised I'm not sure. If this isn't done then the transient frequency and voltage deviation is likely to be unacceptable.

ScottyUK has significant experience with larger machines and can likely comment on what load rejection looks like at that scale, along with further insight as to what happens when sections of the network are islanded.
The one grid connected installation I was aware of that used to end up islanded every so often (37MW Gas Turbine) had a hell of a time reconnecting to the network, as whilst it was all connected the voltage was fine, but its range of adjustment to resynchronise after islanding left a bit to be desired. If I recall correctly they used to have to use the OLTC on the feeder transformer to adjust the network voltage to bring it back again.


EDMS Australia

 

[Mbrooke]

If you can, what are these systems like? How many feeders? How many paralleled gensets?

 

[FreddyNurk]

Just about all of them were up to 5 generators onto a common LV bus, up to four feeders, mostly feeding MV networks via step up transformers but some LV feeders.
All networks were radial feed from the power station, a few may have had the ability to remotely parallel feeders but it wasn't used very often.
Sizes ranged from multiple 70kW sets up to about 1.5MW sets.

All of them had some sort of scheduling system installed to match spinning reserve to load, as well as a means to bring on extra capacity for closing feeders. None of them ever closed onto multiple feeders at once.
Power sharing was always done by parallel isoch via various controllers, VAr sharing was a mix of cross current compensation and automated VAr control, depending on the site. There were no network reclosers present on any of the networks, reclosing was done at the power station only. The systems would (generally) bring on extra capacity and then attempt a reclose, assuming that the network fault didn't send the station black first.

Block loading, if you mean baseload or watt-setpoint output, was never used. Its possible to do with some control equipment, but it was never implemented, rather the load was generally shared across the combination of the running sets. In order to block load with an islanded system, only some of the running sets can be blockloaded, the rest have to follow the load and make up the margin.

At this size frequency and voltage deviations are significant during step load and load rejection events, often in the range of 10%. Note that the behaviour is the same as for much larger sets, but the time and magnitude scales proportionally. As an example, the subtransient fault current might disappear within 5 cycles at this size, but might well be 5 seconds for a much larger system. The frequency deviation is obviously less as the size increases compared to the fault.

I don't know what would happen for a total load rejection event for say, a 250MW turbine, but I'd certainly be interested if anyone else has anecdotes.

EDMS Australia

 

[Mbrooke]

Basically diesel gensets run in parellel as in a backup system? I've heard of this being down for small islands.

 

[FreddyNurk]

Pretty much. Sizes of diesel can vary, larger ones can run medium rather than high speed, and there's a few differences between backup equipment and prime generation, but yes.
It wasn't unheard of for some feeders to have fault current not significantly above load current due to line impedances either.

EDMS Australia

 

[Mbrooke]

That is also my concern for larger stations- 138kv and 345kv fault currents would be much smaller.

 

[crshears]

Electrical islands occasionally occur in four known geographic areas of our system; the surviving connected generation in three of them varies from ~20 MW to ~200 MW, with the connected load sometimes more, sometimes less than that; the fourth island is almost always grossly over-generated, and about half of the time collapses due to instability.

The governors on the larger hydraulic units connected to our system typically have both permanent and temporary speed droop applied to promote their stability in island situations, and in islands that are over-generated the governors generally tend to run these units back more or less by the same percentage amount each; if any of the governor adjustments are out of whack, unit hunting may occur and collapse the island on instability. Once governor action has ceased and the island has stabilized at a higher frequency, the generation operators within the island will, with IESO concurrence, undertake such loading adjustments as required to optimize unit loadings and water usage.

For under-generated islands, under-frequency load shedding schemes will operate in such a way as to hopefully arrest the frequency decline at some stable, lower frequency, hopefully not so aggressively however that the frequency will rise above normal, causing the generator governors to hunt and/or go unstable, tripping off generating units, causing further frequency declines and more UFLS operations until the wheels fall off...and if the island remains stable, the generation operators within the island will, again with IESO concurrence, undertake such loading adjustments as required to optimize unit loadings and water usage.


CR

"As iron sharpens iron, so one person sharpens another." [Proverbs 27:17, NIV]

 

[Mbrooke]

What are the black start procedures? I'm really curious how you would control a radial load at the generating station. It would be amazing to see adjusting all the excitation result in a .5% increase of real load.


Also on my mind is fault ride through. Tripping a line on a temp fault and then reclosing it... can stream governors even do this?

 

[crshears]

FreddyNurk asked:

I'd be interested to know the largest islanded systems in existence.

Hmmm...when does an island become a continent?

I'm not trying to be funny; some islanded systems become so large their operation, behaviour and characteristics are little different than that of interconnections.

Mbrooke asked:

What are the black start procedures? I'm really curious how you would control a radial load at the generating station. It would be amazing to see adjusting all the excitation result in a .5% increase of real load.

In my province, black start procedures are not carved in stone; we are instead provided with a series of black start guidelines [ which can be found in the Ontario Power System Restoration Plan, a public version of which is available at

http://www.ieso.ca/-/media/files/ie...stem-operations/so-ontpowersysrestoreplan.pdf

] that we employ in choosing the best way to rebuild a system or portions thereof, since true system collapses, whether partial or almost complete, very rarely happen the same way twice.

Addition via Edit: The Electrical Reliability Council of Texas has a system fundamentals manual available at

http://mis.ercot.com/misapp/GetRepo...entals Training Manual&showHTMLView=&mimicKey

where you click on the little pdf link to download the document; the System Restoration section begins on pg. 548, and states much the same thing as the OPSRP.

"Control a radial load at the generating station" - huh? If the scenario is one of a smallish electrical island with one control facility for generation, transmission and distribution, it is entirely possible to control a radial load from the generating station using a SCADA system, but at the generating station? Even peak shaving via voltage reduction is performed at < 50 kV, generally by providing a means of biasing the mid-point of the voltage regulation control band downward by three or five percent; the operation of the HV system remains unaffected. Indeed, when 3 or 5% voltage reduction is applied in my province due to system conditions, 3 or 5% V/R is indeed applied to the transformer secondary control bands, which lowers the 44 kV voltage to suit...but our residence remains unaffected, or at least is not affected for very long, since the under load tap changers at the 44 kV fed distribution station that supplies the local customers operates autonomously, and any reduction in 44 voltage that lowers the 8 kV distribution voltage is soon corrected by the ULTCs.

It is for this reason that lowering the voltage on all the generator terminals would be pointless.

Hope this helps.

CR

"As iron sharpens iron, so one person sharpens another." [Proverbs 27:17, NIV]

 

[Mbrooke]

Picture a single generating station feeding island of electrical load.

 

[FreddyNurk]

I'll admit that the definition I'd been using for 'island' is what MBrooke asked, a single generating source feeding load.
This may not be strictly correct in terms of what defines an island.
An example: in Australia, NT has a network that runs from Darwin to Katherine, with multiple sources of generation feeding it, and a 132kV backbone. From the point of view of Australia's National Electricity Network, its an island. From the point of view of connected generation sources and loads, its a network or grid.
My question was based around the definition I'd been using, given that multiple geographically separate sources behave differently to a single generating source on a network.

What is even more fascinating to me, is how islanded sections of network, each with multiple sources of generation, manage to reconnect.

EDMS Australia

 

[DM61850]

Mbrooke,

I think you would be interested in getting a hold of a utility's blackstart plan. When a system is brought back up from being black, it starts up its blackstart units. Most generating units need internal or external power to start up from being completely dead or black. These typically at least from my experience aren't very big. They are sometimes generating sites that have oil burners, hydro plants, or recently a site with battery backup. The site with battery backup had a 12 MW bank, I don't know the MW-H, but it was enough to provide to provide two tries to start a combustion gas turbine. Once the turbine is up to about 30% speed, it can power itself. Once a blackstart unit is brought online, the first action is to start a larger unit and that unit starts other units. The dispatcher than calls the plant and they increment the generation and load in lock step. Putting load on normal system isn't an issue because there is a lot of inertia and lot of generators that can adjust their generation to keep the frequency fairly steady.

You might want to look into NERC reliability standards PRC-006, PRC-010, and PRC-024. They dictate the requirements for Under Frequency Load Shedding and Over and Under Frequency Tripping of Generations. The requirements for short times for generators is very lax to give the generator a chance to bring the frequency back to 60 HZ during islanded conditions. Quebec allows tripping to begin immediately for swings exceeding 66 HZ and the for the other interconnects it is 62 HZ. I think that on older systems, they lived with brief swings like this until the exciters and governors brought the system voltages and frequencies back.

https://www.nerc.com/_layouts/15/Pr...ency Load Shedding&jurisdiction=United States

https://www.nerc.com/_layouts/15/Pr...er=PRC-010-2&title=Undervoltage Load Shedding

https://www.nerc.com/_layouts/15/Pr...ive Relay Settings&jurisdiction=United States

 

[Mbrooke]

Thanks DM61850.


If curious, here is what I have in mind. 3000MW generating station; 2,250-2,500 MW peak loading.

Half load circuits- ie transmission lines rated 3000amps thermal, 1,500 amp actual peak loading. About 150-300MVA per circuit. Normally open ties to the neighboring system.

I know, the drawing far from perfect lol- but to give readers and idea.

https://files.engineering.com/getfi...a21-97975f9ffa58&file=autodraw_1_8_2020-1.jpg

 

[Mbrooke]

FWIW, here is an example of a generic station type I have in mind . I should have posted this earlier. 10 units- about 250MWs a piece.

 

[waross]

When the plant is grid connected the sets will be base loaded and the swing set may be elsewhere.
When the plant becomes islanded, it will revert to 5% droop control.
If the island is exporting power, the frequency will rise and stabilize on the 5% droop curve.
If the island is importing power the frequency will drop on the 5% droop curve.
If the island load is greater than the plant capacity steps should be taken to shed some load.
A real world example.
Vancouver island, British Columbia, Canada.
The island has hydro generation, co-generation and an undersea DC feed from the mainland.
Although now defunct, the island had at one time a world class open pit mine and mill.
The primary mills were a total of 36,000 HP.
This was one of the loads that was shed should the island become islanded and the frequency drop.
The protection scheme included a utility sealed under-frequency relay set at 55 Hz that tripped off the mill in the event of a 5 Hz drop in grid frequency.
There was a second mill owned under-frequency relay that allowed the mill to voluntarily reduce their loading so as to attempt to avoid a general trip.
I was associated with the mill construction.
I can't add anything about the overall utility procedures.
The view from a customer that was subject to load shedding on an island that could become islanded may be interesting.

Bill
--------------------
"Why not the best?"
Jimmy Carter