Isochronous load sharing lines for generators in parallel (Island system) 1

[NickParker]

I read that for this scheme to be implemented, the speed controllers of all the turbines have to communicate with each other; but I don't see anything regarding AVR/Excitation on the same units.

How about AVR controls on the generators for this scheme?
- Do the AVRs of all the generators need to communicate?
- (or) Voltage droop mode implemented on all the generators

 

[Tony Tonon]

Not that I am aware of.
Normally this logic to control the paralleling of the gens is controlled by the ATSs

 

[waross]

Old school.
A quadrature or cross compensation scheme was used for VAR sharing.
Var sharing depends on voltage settings.
A reactive current in line C will be at right angles to the voltage from line A to line B.
The odd phase CT current was dropped across a resistor and the resulting voltage used to bias the sense voltage.
These were inter connected so that the VAR production was balanced between sets.
An AVR set up for quadrature compensation has an extra terminal and an internal resistor.
In a pinch you can convert an AVR by adding an external resistor.
I don't remember the exact details of the circuit but it is quite simple.

--------------------
Ohm's law
Not just a good idea;
It's the LAW!

 

[xnuke]

The old school way waross mentioned still works, and doesn't require communication between AVRs because the CT feedback voltage created by the resistor provides negative feedback that eliminates the circulating current that arises. See this paper by Basler: Link. This only stabilizes the system against the AVRs going in opposite directions and creating circulating currents that lead to generator breaker trips; you should only have one machine in voltage and frequency control mode (isochronous) and all others in droop.

New school.
Yes, the AVRs must communicate. Each generator gets a generator controller that handles paralleling and load sharing. The generator controllers have outputs to both governor and AVR and communicate over a network to control both active and reactive load sharing for the parallel machines. One parallel machine's controller is set as the master controller and regulates voltage and frequency, but unlike the old school method where only the isochronous machine swings to regulate these, all paralleled machines receive incremental commands to adjust power and reactive power output in order to regulate voltage and frequency.

xnuke
"Live and act within the limit of your knowledge and keep expanding it to the limit of your life." Ayn Rand, Atlas Shrugged.
Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[waross]

With the swing set in isochronous, the load dispatch center would monitor the load on the grid and dispatch the droop machines by telephone.
The local operators would be instructed to run their machines at 90% load, 10% load or go offline.
Our small plant with only one generation facility ran all machines in droop.
That worked well.
The operators checked and recorded all parameters every 15 minutes, and tweak the frequency if needed.
They would put machines on line or off line as the load varied.
I discussed isochronous control with the plant owner and we rejected the idea and stayed in droop control.
Isochronous would have placed added responsibilities on the operators and we were loath to do so.
Droop was more forgiving of operator mistakes than one set in isochronous.
With one set in isochronous, the droop sets run as base loaded sets.
You may still run all machines in droop mode with VAR compensation.
It's simple and dos not require a load controller.
Not, grid machines run in 5% droop, islanded machines run at 3% droop.
As the years rolled by, none of our customers ever noticed let alone complained that the frequency was not always exactly 60 Hz.
One or two found that there old electric clocks did not keep perfect time and just replaced the clocks with new clocks.
The synchronous motor driven clocks are pretty much all gone now.
The choice depends to some extent on an overall appraisal.
Type of load and expected load swings.
Type of prime mover.
Operator controlled or fully automatic.
Grid tied, run at 5% droop. The grid operator will run the isochronous machine.
Islanded sets, under the KISS principle, all machines at 3% droop. BUT there are a lot of exceptions.

Isochronous with diesels:
With one set in isochronous, that machine would run 24/7 and accumulate hours to overhaul much faster than the other machines.
Changing another set to isochronous would entail governor adjustments that we were not willing to trust to our operators.
A load control center was out of the question due to cost.
Retailing diesel produced power, the profit margin was non-existent.
The original investors had a sweetheart rate for their homes, but not businesses.
The two major investors both ran supply ships and imported and sold diesel fuel to the plant.
They charged the full going rate for fuel but neglected to remit the import taxes.
Other than that, there was no profit.

--------------------
Ohm's law
Not just a good idea;
It's the LAW!

 

[Drivesrock]

Also note that if there are any main bus interconnectors/bus-ties (and also ring connection) that separate (reconfigure) the online generators, the status of these must be connected to the controllers or droop CT / cross compensation circuit so that only the parallel generator sets kW and VAr share. If this is the case, then the AVR CCC physical circuits can get a little complicated and there are more failure points so leaving the AVRs just in droop mode individually and adjusting the voltage manually if required, is much simpler. Good modern AVRs like from Basler have internal logic programming and additional I/O so that this can be done inside the AVR and is very good.

 

[NickParker]

Droop was more forgiving of operator mistakes than one set in isochronous.

Can you explain this?

I thought one machine in Isochronous and others in Droop was more forgiving than All machines in droop, because
a) All machines in Droop mode - every time the island load changes the frequency will change and the operators will have to change load on one or more of the droop machines to respond; more work for operators
b) Whereas one machine in ISO and others in droop, - the operators have to work only when ISO machine being overloaded (capacity limit) or being motored.

 

[crshears]

Now you're getting into my wheelhouse . . .

Droop was more forgiving of operator mistakes than one set in isochronous.


I thought one machine in Isochronous and others in Droop was more forgiving than All machines in droop, because
a) All machines in Droop mode - every time the island load changes the frequency will change and the operators will have to change load on one or more of the droop machines to respond; more work for operators
b) Whereas one machine in ISO and others in droop, - the operators have to work only when ISO machine being overloaded (capacity limit) or being motored.

To which I say: it depends . . . and I seem to have lost track of the right way to attribute quotes . . .

Moving on:

Bill already spelled out a few things in his most excellent post; I'll try to add to this from operating experience.

"All machines in droop" or "one machine in isochronous and others in droop": which is preferable and which makes for more or less work for the operators is in large measure determined by [a] the size of the grid under discussion, its generation mix, [[c] the geographic dispersion of the generation relative to the load, [d] the capacity and configuration of the grid itself and the voltage levels at which its various components operate, etc., etc., in other words, properly answering this question involves actual power [flow] system analysis, especially in the larger systems. This is further complicated by supply balance considerations used to determine to what value both the real and reactive loadings should be adjusted at each generating facility for optimum overall system efficiency / lowest overall energy cost.

With all machines in droop, it generally takes a fairly substantial change in load for the frequency to deviate from its nominal value, so there may not be as much operator work as one might at first think. Once the change becomes necessary however a determination will need to be made as to how to address the imbalance: will governor setpoints on all machines need to be adjusted in equal percentage amounts so as to restore frequency to nominal? Add or remove generation? Adjust sales to or purchase from neighbouring utilities, if applicable?

With one machine in isochronous and others in droop, the process can be, but is not necessarily, simplified; the balancing authority / system operator monitors the loading of the isochronous unit [or plant] on a more or less continuous basis, iteratively determining which generation resource will be the next to be loaded/unloaded/added/removed. In larger systems or interconnections, one individual is generally designated the "generation chaser" for that shift.

I await further responses before continuing.

CR

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

 

[waross]

We discussed changing our small generating station to one set in ISO mode.
Our operators were at the limit of their competency synchronizing sets.
We rotated machines to even out the wear.
The same machine would not always be the ISO machine.
This would have meant changing governor modes each time the ISO machine was rotated out of service.
That was more responsibility for the operators than we were comfortable with.
I'll spare you the list of operator errors as it was.

By the way, large grids with one swing set (the ISO machine) typically tun at 5% droop.
Islanded droop is typically 3%.
That means that for a load change of from no-load to full-load on all machines in service, the change in frequency will be 3%.
The customers never notice.
One giveaway would have been synchronous motor based electric clocks, but those clocks are obsolete and very few are left in service.
I understand that large systems use cycle counting and at times deliberately run a little over or under frequency to maintain an exact average of 60 Hz over an hour or day, so as to keep any old clocks in time.*
That again was more than we could expect of our operators.
Technology changes and computer based load control panels will easily do much more than we ever did manually.
"The times, they are a changing."
*Note: In the early days, a lot of power was generated by reciprocating steam engines controlled by fly-ball governors.
The faster the engine went, the further out were the fly-balls.
After a period of heavy loading caused the loss of cycles the plant would be running slightly over speed to regain the lost cycles.
Running at maximum over speed was described as running "Flat out", or in operator slang, "Balls to the walls".
Both terms implied running at maximum speed.
Balls to the walls referred to the fly-balls on an old style governor.

--------------------
Ohm's law
Not just a good idea;
It's the LAW!

 

[waross]

By the way; In a large station, the swing set or ISO set will be at only one station.
All the other stations will be running in droop anyway.
Our operators used to check and balance every 15 minutes.
With a mix of 8 cylinder diesels and 12 cylinder diesels, our head operator reported the load as "How many cylinders were online".

--------------------
Ohm's law
Not just a good idea;
It's the LAW!

 

[crshears]

In Ontario, Canada, I was told one entire hydraulic generating station was the swing plant, and for the province, if not the whole Eastern Interconnection, the speed droop is 4%.

CR

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

 

[waross]

Thanks CR.
While I don't always know the droop of a standby set, I have never encountered a standby set with other than 3% droop.
I have heard numerous times, 5% droop reported by our friends to the south for grid tied sets.
But come to think of it, the 5% is often quoted for co-generation and distributed generation.
Hypothetically, with a grid at 4% droop and distributed generation at 5% droop, the distributed generation would be less affected by grid disturbances.

--------------------
Ohm's law
Not just a good idea;
It's the LAW!

 

[crshears]

Very true, Bill; and in Ontario, at least, zero reliance is generally placed on distributed or co-generation, regardless of size, to provide any active grid stability whatever. And as I understand it the Technical Interface Rules for these are now written in such a way that islanding smaller plants [ <10 MW ] with local load is actually prohibited.

As a matter of fact, within numerous small Ontario legacy hydraulic generating stations that were originally built with full-on governors capable of supporting islands of load, when it came time to rebuild the governor, the owners have more often than not determined that they can't be bothered with the expense or this now-pointless exercise, and have instead replaced them with gate positioners coupled with relatively primitive protection schemes.

Indeed the up-front capital expenditure of auto-start equipment for these has often been determined to not be worth the trouble, and these plants now require on-site human attendance to restart them, with forebay water level control being accomplished via set-and-forget stoplog configurations.

Addition via edit: oh yeah! My aviation industry brother-in-law told me "balls to the wall" refers to aircraft engine throttles, often having spherical knobs at the end, being pushed all the way forward until they are in contact with the instrument panel, such that maximum engine power is developed during take-off . . .

CR

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

 

[charz]

Depending on the amount of load to be added/removed (variations)?
If the variations are low: one in ISO and other machines in droop are preferable, since the ISO machine can handle variations in load without getting overloaded or motored.
If the variations are high: All machines in droop are better, because the ISO machine alone cannot handle the load variation and there is a chance that the ISO machine will be motored/overloaded.

Is my understanding right?

 

[crshears]

Your logic is correct, provided the system has no tie lines with neighbouring entities; in such instances, having one large generating station in isochronous mode, and set up so that all units ramp up and down together at equal output, will serve to absorb swings quite well.

Where tie lines are present, automatic generation control [ AGC ] is commonly used instead. AGC systems generally use frequency deviation to perform proportional control, along with factoring in tie line flows, modulating overall electricity production so the vectorial sum of all tie line flows remains on schedule, analogous to integral control.

I recall learning of times where multiple units in multiple generating stations were placed on AGC together; the droop of the units remained unchanged, with the raise or lower commands applied to all governors of all plants in AGC simultaneously. Slope calibrators were used to tweak the thermal units' response characteristics so their behaviour remained smooth and excessive generation rate ramping did not occur.

 

[waross]

Disclaimer:
My experience is with older, hydraulic and mechanical governors and systems.
We were limited in what we could do due to the limitations of our equipment.
What could not be done, or was avoided due to reliability, stability or possible failure modes, is now easily done with load control panels and computer controls.
The old school systems reacted to load changes in droop.
When a large step load was either connected or disconnected, the change in system frequency was related to the ration of the load change to the system on-line capacity, times the % droop.
So even a step load change of 10% of the system capacity would result in a frequency change of 10% x 4% (in Ontario) x 60 Hz = 0.24 Hz change. (Ontario is mentioned to correlate with the 4% droop of the Ontario grid.)
The ISO machine would detect this and adjust it's prime mover input to correct back to 60 Hz, or over-correct for a short time to correct the hourly average cycle count back to 60 Hz.

What happens if the ability of the ISO machine to correct system load changes is exceeded?
This may happen if communication with load control is lost or if there is a large scale progressive blackout.
As the load drops the output of the ISO machine drops.
Once the ISO machine is at zero output it loses control. The system frequency continues to drop in droop, until, at zero load on that part of the system, the frequency will be at 60 Hz minus 4% = 57.6 Hz.
In a progressive blackout, some parts of the system may lose the load.
In that instance the frequency will rise 4% to 62.4 Hz.
Some areas may become islanded away from the ISO machine and will revert to droop control.
Over the last ten or twenty years, a lot of facilities, generation and substations and switchyards have been tied together by communication systems and computer control introduced so as to avoid the large scale progressive or rolling blackouts that we experienced in the past.

Does anyone else feel like an analogue dinosaur in a digital world?

My description is the way it was.
The digital kids have been working with communications and computers for about twenty years now to mitigate, avoid or eliminate most of the constrictions that were common for decades. (Or generations)

--------------------
Ohm's law
Not just a good idea;
It's the LAW!

 

[waross]

Hi CR.
Fly-balls.
Throttle balls.
They both work.
The point is that "Balls to the Wall" (aircraft.)
or
"Balls to the Walls" (power plant)
Are not off colour anatomical references.

--------------------
Ohm's law
Not just a good idea;
It's the LAW!

 

[crshears]

waross wrote:
What happens if the ability of the ISO machine to correct system load changes is exceeded?
This may happen if communication with load control is lost or if there is a large scale progressive blackout.
As the load drops the output of the ISO machine drops.
Once the ISO machine is at zero output it loses control. The system frequency continues to drop in droop, until, at zero load on that part of the system, the frequency will be at 60 Hz minus 4% = 57.6 Hz.

Hey Bill, I'm confused . . .

Are we talking progressive loss of generation or progressive loss of load?

On loss of load, system frequency will rise.

On loss of generation, system frequency will drop.

Please clarify.

CR

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

 

[waross]

Both and either.
With a progressive major blackout, a section will become overloaded and a sub may trip offline. That will overload other sections which will have the same effect as loss of generation. The frequency will drop.
A section may become overloaded and open an interconnect. That may be loss of load for that section and that section may go over frequency.
It depends.
Either way, the ISO set may lose control but when it does, the frequency excursion will be limited to 4%.
There is a third case where an interconnect opens and a section is badly overloaded past the ability of generation to carry the load.
If and when that happens, I expect that shortly more breakers will open leading to either outages, over frequency or under frequency.
The point is, when the ISO loses control, droop takes over and does it quite well.

What happens if the ability of the ISO machine to correct system load changes is exceeded?

exceeded may be the wrong word here.
Loss of ability to control may be either zero load on the ISO or 100% load on the ISO.

--------------------
Ohm's law
Not just a good idea;
It's the LAW!

 

[crshears]

Hi again Bill,

Maybe different terminology is clouding my perception . . .

Where limited system contingencies occur, electrical islands may be created, which may be either over-generated [leading to over-frequency] or under-generated [leading to under-frequency], usually with droop greatly mitigating the deviation. If the island is sufficiently under-generated, Under Frequency Load Shedding schemes should trip off enough selected load to allow the frequency of the island to stabilize somewhere within an acceptable bandwidth.

With a progressive major blackout, a section will become overloaded and a sub may trip offline. That will overload other sections which will have the same effect as loss of generation. The frequency will drop.

This is where I'm not following you . . .

Are you referencing situations where there are [normally] multiple paths between generation and load, but some sections have tripped out overloading the remaining lines? In such scenarios, I could foresee lines being tripped out of service for exceeding thermal limits, or possibly stability limits; but I'm not getting what you mean when you speak of overloading other sections "which will have the same effect as loss of generation" and "The frequency will drop."

If sufficient generation remains on line to supply the load, I would not foresee any frequency drop, even if the lines supplying the load are incipiently cherry red from overheating but have not yet tripped out; in such scenarios nothing but a system separation should cause a frequency change anywhere . . . at least as I understand it.

CR

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