Islanded Generators - Control of Real Power and Frequency

[EddyWirbelstrom]

An islanded powerhouse has 3 off steam-turbine generators of 25 MVA and 1 only gas-turbine 50 MVA generator. All generators are connected to a common 22 kV bus via generator transformers.
All of the steam leaving the steam turbines feeds into a refining process. Similarly, the GT’s waste heat is converted to steam in a heat recovery steam generator which also feeds into the refining process.
We are unsure of the existing governor control modes for these machines. We do know that an automatic frequency controller is used, but only on the steam-turbine generators, to correct for frequency droop due to load changes. The controller restores frequency to 50 Hz but we are advised that it has a time delay of approximately one minute. The gas-turbine is not controlled by this system.
What governor control modes must have been selected to result in the following characteristics exhibited by the attached traces:
• The steam-turbine generators supply the load variations,
• The gas-turbine generator supplies a constant MW base load.
One anomaly to this pattern exists. At approximately half way through the recording period, in response to a 3 MW plant load increase, the GT responds by taking up most of the load increase. However, after a minute or two the three steam generators increase their MW outputs to take up all of the increased plant load and the GT returns to the same base load MW output it was delivering prior to the plant load increase.
The questions are:
1. Why is it only in one isolated event that the GT responds to a change in plant MW load?
2. If one of the Turboalternators experiences an unexpected electrical trip will the GT continue to output constant MW at a lower speed or will its governor, on sensing the lower speed, automatically switch to droop control and increase MW output in accordance with its droop setting? Or do other possible GT governor responses exist?
In the attached file the plot colors represent :
Dark Blue Plant load in MW
Blue Frequency in Hz
Pink Sum of the 3 steam turbine generator outputs in MW
Yellow Gas-turbine generator output in MW

 

[waross]

Without seeing the installation we must make some assumtions, dangerous as that may be. Please take the following as just my opinion of the most probable control scheme.
Governor modes. The frequency controller is acting in isochronous mode. The first response to a load change is droop action. The controller then detects and corrects the frequency error by adjusting the set point. This happens with an increase or decrease in the load. With 3% droop, the frequency error is 3% x 1% for every 1% change in load.
A 10% block load will result in a frequency error of 0.3%,
or a frequency of 49.85 Hz. During the following minute the frequency set point will be increased to 50.15%.
At steady state 100% load the frequency set point will be 51.5 Hz. The actual frequency will be 50 Hz.
The controller is sending the same control signal to all the steam turbines and they act together as one 75 MVA turbine.
The gas turbine is likely running with less droop than the steam turbine controller.
For example, with all machines running in droop, if the gas turbine runs 3% droop and the steam turbines run 6% droop the gas turbine will be 100% loaded when the steam turbines are 50% loaded. This would appear to limit the maximum load to 50 MVA +(75 MVA x 0.5) = 87.5 MVA What actually happens is the output of the gas turbine should be limited to 100% by the maximum fuel supply and the frequency drops enough further that the droop settings of the steam turbines will pick up the rest of the load.
Remember that the first response of an isochronous system is droop. All machines will respond to block loading as if they were in droop mode. Then the ischronous feature will adjust the set point of the steam turbines and the GT will drop back to base load.
In regards to the spike loading of the gas turbine, that is normal. If the percentage load increase on the GT is disproportionate then consider:
The GT may be running less droop than the STs.
More likely, the GT control system has a faster response time than the control system on the STs.
In the event of a steam turbine trip, the GT is probably already in droop mode and will continue to run in droop.
Droop = proportional plus offset.
Isochronous = droop plus reset, or proportional plus reset (and possibly pre act) or proportional plus integral (and possibly derivative).

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

 

[EddyWirbelstrom]

Thanks Bill for your excellent description of speed droop and load sharing of islanded generators.

To confirm my understanding of your example where the GT supplies 100% of its rated MW at the desired 50Hz, while the ST supplies 50% of its rated MW at 50 Hz due to a 3% and 6% speed droop respectively I have attached a speed droop diagram of the GT and ST.
The diagram assumes :
• GT has a fixed no-load-frequency-set-point of 51.5Hz and a speed droop setting of 3%
• GT has a fuel limit setting which limits GT output to 100% of GT MW rating. ( corresponding to 50Hz )
• ST has a variable no-load-frequency-set-point which is controlled by the frequency controller, and that the ST has a speed droop of 6%.

Can you please confirm my conclusions :

1. The GT supplies constant base load of 100% GT rating only while the sum of plant load is equal or greater than the sum of 100% MW rating of GT plus 50% MW rating of ST. This is because the GT max fuel setting corresponds to 100% of GT MW rating. Extra load is taken by the ST.

2. If a step reduction in plant load to below 100% GT + 50% ST occurs, the frequency of all generators will increase. If before the frequency controller has had time to reduce the ST no-load-freq-set-point, a step load increase occurs, the GT will supply most of the increase until it reaches 100% of GT rating, at which time the ST will take the extra load.

3. The frequency controller will eventually respond and increase the no-load-freq-set-point of the ST and increase the frequency of all generators to 50Hz.

Would this explain the MW spike in the GT plot ?

 

[waross]

Nice drawing.
A couple of caveats;
The droop figures that I used were for example only. 3% is common for the diesels that I am familiar with. I don't know what percentage is common for those big sets.
Two things may cause the spike (I thought I saw some negative spikes also, as a result of dropping a block load.)
The first is different droop settings on the governor.
The second, and perhaps more likely in your instance is different response times of the controls.
If you are able to expand your plots they should be easy to differentiate.
How long does the spike last? If recovery takes a minute the issue is probably differing droop settings.
If the recovery is more rapid, the issue is most likely slow control response on the STs. I suspect that this may be your issue.
Re response times, I was at a sawmill where it took both diesel generators to start the big hog, on a soft start. One diesel had an electronic governor and one had a hydraulic governor. In the morning we had to wait for the oil in the hydraulic governor to warm up before we could start the hog. When the oil was cold the response was too slow and the electronic governor set would take most of the load and the breaker would trip.

And a heads up. If ScottyUK drops in, he is the guy for GT and ST information.

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

 

[ScottyUK]

Bill! You are making me blush!

You guys seem to be doing ok on your own. What is the plant configuration? It sounds a little odd for a CCGT or CHP with one GT and three ST's so I guess it's a conventional thermal plant with a GT added on. I ask because in a CCGT or CHP plant the GT output and the steam-raising capacity of the HRSG are intimately linked, so it isn't quite as simple as base loading the GT and allowing the ST to pick up the remainder because if the ST were at part load there'd be a hell of a lot of steam to get rid of. If the ST's have their own boiler plant which is not dependent on GT and HRSG then it opens up other control options.

That spike in the trend is difficult to interpret because the time resolution is so coarse - could be anything from control loop tuning to a fault on the system to a big motor starting to another generator synchronising or tripping off the bar. If you're islanded you have more chance of finding out what happened at that moment from the plant Historian or the alarm handler. If you are connected to the grid then those events often end up in the 'sh;t happens' category because there's no way of positively identifying the cause.

Post a few details of the plant configuration when you get time.


----------------------------------

If we learn from our mistakes I'm getting a great education!

 

[EddyWirbelstrom]

The plant originally only had 3 steam-turbine generators which have there own boilers. Exhaust steam from the steam-turbines is used for the process plant.
The gas-turbine generator was added at a later date. The gas turbine has a heat recovery steam generator ( HRSG ) which also is used for the process plant. The HRSG is independent from the ST boilers.
I have attached plant recordings taken at 15 second intervals over a two day period. <Chart03> shows the GT responding to a 3 MW increase in plant load while <Chart02> shows in increase in plant load being supplied by only the STs. A transducer delay must be the cause of the offset between plant MW load and ST MW.

 

[ScottyUK]

I find charts 2 and 3 interesting because the GT has virtually no response to a falling frequency in chart 2, yet on Chart 3 and with a more gently falling frequency the GT loads up significantly for a short period until the ST group picks up the load.

I'm a little uncertain what mode the GT is operating in on Chart 3. The engine must be good for about 40MW or so based on a 50MVA set - an LM6000 perhaps? - so it should not be on the thermal limit on any of the data sets unless something like IGVs are artificially increasing the back end temperature. A MW control loop should be fast enough to avoid any MW major excursion with shifting frequency, and appears to do just that in Chart 2,which begs the question 'What is going on in Chart 3'? You'll need to trawl through the event and alarm histories to find out I suspect, but at first glance it doesn't look like plain MW control or the temperature limiter in control.

If the GT is running at a constant MW setpoint as it appears to be in Charts 2, 4, & 5 then it effectively does not have any droop, leaving the frequency to float with the ST dropp characteristics and the supervisory trim controller. If maintaining frequency stability is important then you are not using the GT to best advantage because you are not using it's rapid response capability to restore frequency to setpoint, although there may be good commerical reasons for not doing so.

I will say I'm no expert on islanded plants, so I am thinking aloud to some extent here.


----------------------------------

If we learn from our mistakes I'm getting a great education!