We have two GE frame-6 GT at our plant complex which operate in parallel with GRID. Both the GT operate in droop mode. The running load on GTG-A and GTG-B is maintained near to 22MW each. The load on GRID remains around 4MW. GTG-B is kept on ISO-Standby mode i.e in case of GRID tripping, this machine will switch to ISO mode. Recently one incident happened at our complex. The grid frequency gone to 50.5Hz and GRID got isolated at this point by IPR relay at over frequency protection. Load shedding took place at the complex and a load of around 15MW was shed by Load Shedding System. Load Shedding System keeps monitoring the total generation and connected load and if any of the source get tripped, excess of load is shed. After the transient period and execution of Load Shedding, GTG-A was observed running at 21.5MW, GTG-B (now ISO-machine) was found running at 11.5MW. The IPR relay (MICOM P343) disturbance recorder show a load of around 27MW on GRID feeder just before tripping.

I could figure out that due to droop characteristics of governors of both the GT, they were supposed to throw load due to over-frequency at GRID. If, I take 27MW a reliable value, then we can say that both the GTG should be running at a total generation of (22+22+4-27) = 21MW i.e. during the transition phase, before opening of GRID breaker the generation from each GTG should be around 10.5MW each.

My query is how the GTG-A and GTG-B governors would have acted after tripping of GRID? How much time GTG-A will take to regain the thrown load (start generating) the same power as it was generating before the transient over frequency? From the results, it is looking like that the GT-A governor acted very fast and Load Shedding System recorded its generation near to 21.5MW before starting shedding sequence. Also, what will be the speed reference for GTG-B after it switch over to ISO mode and how fast the action of GTG-B machine will be? What will be the system frequency immediately after de-synchronization from GRID?

How this problem of over frequency can be tackled? Should we go for more conservative over frequency setting for Islanding relay or should we go for increasing the speed droops for governors?

 

neeuor,

This discussion presumes the two GE-design Frame 6 heavy duty gas turbine-generators use Speedtronic turbine control systems. It also presumes when you say the GRID load "...remains around 4 MW...." that means that the two GTGs, when synchronized to the grid through a tie breaker (a breaker that connects ("ties") the plant to the grid) and the 4 MW is measured as being exported to the grid as it's more than the plant load consumes. In other words, it's presumed the two GTGs, when operating at 22 MW output each and synchronized to the grid through a tie breaker are producing 4 MW MORE than the plant consumes. It's important that this be clarified for proper understanding.

Further, the discussion presumes neither of the GTGs is operating with Pre-Selected Load Control enabled and active. (This, too, is VERY important to clarify.)

Now, what should happen--presuming that GTG-B automatically transfers to Isoch when the tie breaker opens--is that the plant load, 44MW-4MW=40MW will be split as follows: GTG-A will continue to run at 22 MW (because that's the load that's proportional to the Droop speed control reference prior to--and after--the tie breaker opening), and GTG-B will drop to 18 MW because the total plant load (when separated from the grid) is only 40 MW. And, this will keep the plant frequency at 50 Hz. (There may be a little bobble in frequency as GTG-B transfers to Isoch, but it should quickly stabilize. When there is a little bobble in frequency, the load of GTG-A will also fluctuate slightly--the two units are SYNCHRONIZED together and must operate at the same frequency!)

Nothing else should be required when the two units are operating in parallel with each other and exporting power to the grid, and the tie breaker suddenly opens isolating the plant. If properly configured, the unit in Isoch standby (identified as GTG-B) should automatically transfer to Isoch speed control and automatically adjust its load to keep the frequency of the plant--and of GTG-A--at desired (50 Hz, or 60 Hz--whatever your site is supposed to be producing).

It's not clear what the plant load is when operating independent of the grid; it seems the load is more than one GTG can provide--which would make load-shedding necessary if only one GTG were operating when the plant was isolated (separated) from the grid. This value is important to know and understand, also.

If the two GTGs remained in Droop speed control (again--with Pre-selected Load Control disabled and inactive!) and the plant separated from the grid, if the two were running at 44 MW before the separation and exporting 4 MW to the grid then after the separation the two would remain running with a total generation of 40 MW (the plant load--but at a higher frequency. The Droop speed control references won't change when the tie breaker opens, but because the plant frequency will increase the load of the two machines will decrease to 40 MW--that's all the load there is, and if both machines are in Droop speed control (with Pre-Selected Load Control disabled and inactive) then the speed error will change because the load drops.

To correct the frequency, the operators would manually decrease the load turbine speed reference on one or both machines until the frequency returned to rated, and the load will remain at 40 MW (this presumes the load on the plant is stable and not changing very much during the disturbance).

Now, if one GTG were off-line and the other were producing 22 MW (with Pre-Selected Load Control disabled and inactive), and the plant load was 40 MW, the load across the tie breaker would be -18 MW--because that's the amount of power that would need to be supplied to the plant from the grid because the only GTG running was only providing 22 MW.

If the tie breaker tripped at this time (only one GTG running at 22 MW in Droop speed control (with Pre-Selected Load Control disabled and inactive)), then the Load Shedding system should immediately shed 18 MW of load to keep the plant frequency stable.

OR, if the lone GTG could or did transfer to Isoch speed control then the Load Shedding system should be programmed to know approximately what the Base Load rating of the GTG is and shed an amount of load such that the remaining plant load will not exceed the Base Load rating of the lone GTG. The GTG would automatically load to Base Load to maintain frequency as long as the load on the plant was reduced to an amount that was not more than the Base Load rating of the GTG.

If I have understood the loading situation, and if the standby Isoch unit automatically transfers to Isoch when the tie breaker opens, and if two GTGs are operating in parallel (synchronized) when the tie breaker opens, then the Isoch unit should automatically adjust its load be equal to the plant load minus whatever the other GTG was running at before the tie breaker opens--and that will keep plant frequency at rated. NO load shedding should be required--this presumes the plant load is less than the Base Load rating of GTG-B plus 22 MW (because that's the load the Droop speed control GTG will be at before and after the separation--presuming Pre-selected Load Control is disabled and inactive).

Changing the Droop setpoint of any machine--without getting approval from the grid regulator--is usually forbidden in most parts of the world. Grid regulators/operators count on the programmed value of Droop for machines when calculating grid stability, and changing the Droop setpoint without at least notifying the grid regulator/operators--and probably obtaining their approval before doing so--is usually contributing to grid instability problems, and may even be illegal. Any problem(s) at your plant--real or perceived--cannot be solved by changing the Droop setpoints of the machines. You'll only disturb many other things about the unit operation (like loading/unloading rates; etc.).

If I haven't understood the plant load--and I DON'T understand the 22+22+4-7 and 27 MW business at all (where did 7 MW come from??? and where did the "reliable" 27 MW come from??? are you saying that Base Load for the Frame 6s at your site is approximately 27 MW each???)--the principles above still apply. If plant load when separated from the grid is less than the capability of the two units operating in parallel (synchronized) can supply with both units operating at 22 MW, and one of them transfers to Isoch control automatically when separated from the grid then Load Shedding isn't required, because the Isoch unit will automatically adjust it's load to be equal to plant load minus 22 MW to maintain rated frequency.

 

Dear CSA,

I think the information provided by me was not sufficient to describe the issue correctly. I would like to add further information. The plant connected load is 48MW. 44MW is generated by GTG-A plus GTB-B and remaining 4MW is imported from GRID. Both GT-A and GT-B machines run in Droop mode. The droop settings for both the turbines are 4%. GT-A is running on Mark-V and GT-B is running on Mark-VIe.

The Incident: There was disturbance in the GRID frequency and GRID frequency was recorded as 50.5HZ (nominal frequency is 50Hz) just before Tripping of GRID TIE line feeder. Also the load on GRID Tie line was recorded as 27MW just before the instant of breaker opening. The timing accuracy is of the order of 1ms for the disturbance recorder for MICOM P343 relay which is being used as Island relay on Tie Line GRID feeder. The over frequency settings for TIE Line GRID breaker tripping are a combination of over frequency magnitude and rate of change. Setting are 50.3Hz AND df/dt>100mHz/s.

From your earlier posts and referring some text books, I could understand that GT when operating in droop mode will throw/reject the load if the frequency of GRID goes high as the difference (TNR-actual rotor speed i.e. GRID frequency) will reduce and droop governor will simply act to adjust fuel in proportional to this error signal. My query is regarding the action timing of the governor for rejecting the load due to change in frequency of GRID. Also I want your confirmation for my calculations for transient period.

Calculations: I have taken Base load of the machine as 36MW. Just before the disturbance, machines were running at 22MW load. For 22MW the TNR shall be around 102.4% (equals to 51.22Hz) for 4% droop setting. Now if the GRID frequency changes to 50.5Hz, the machine will throw around 9MW load and operate at 13MW. So if I am not wrong, a total of 9MW+9MW+4MW = 22MW (4MW was already running on it) will go to GRID TIE Line feeder if the GRID frequency goes to 50.5Hz. 50.5Hz is recorded value before the instant and its was increasing at a rate greater than 100mHz per second. So the exact value may still be higher. I therefore guess and think it was little bit higher and an Import of 27MW power was recorded on IPR P343 relay just before opening of the breaker.

Now if I assume, 27MW was being imported from the GRID before the TRIP, GT-A and GT-B might be generating 48MW-27MW = 21MW or 10.5MW each (total connected load is 48MW). My question is how much time GT governor will take to regain load from 10.5MW to 22MW as TNR is still set at 102.4% and rotor actual speed is normalized after isolation of GRID.

After GRID tripping, GT-B will be taking command for frequency control. I also ask you to please correct me if I am wrong in understanding that GT-B will immediately (after coming in ISO mode) change it's TNR to reduce the speed of rotors to reach rated frequency (50Hz).

 

neeuor,

What you're referring to as "throwing load" isn't the same as load shedding. And, you haven't said whether or not Pre-Selected Load Control was enabled and active on the two units at the time of the incident....

The time required for a Speedtronic turbine control system to return to rated frequency after a sudden change in load (such as when separated from a grid) depends on the magnitude of the load change and the tuning of the control system and/or its servo-valve outputs (the regulators that drive the servo-valve outputs). If the control system on a single GT automatically switches to Isoch speed control when the tie breaker opens AND the resulting load (the load on the unit after the tie breaker is opened) is less than Base Load, it should take less than a second or two for things to stabilize. And, the magnitude of any frequency excursion will depend on the amount of load change and the inertia of the machine (turbine and generator) as well as the tuning of the control system and/or its servo-valve outputs.

TNRI, Turbine Speed Reference-Isochronous, should become active as soon as the unit switches to Isochronous Speed Control. TNRI becomes TNR (the turbine speed reference) when Isochronous is active. So, yes; <b>if</b> TNRI is 100.0% when Isoch becomes active, the turbine control system will be trying to immediately reduce load to make turbine speed equal to 100.0%--which means 50.0 Hz. (I often find that TRNI is NOT at 100.0%, because operators or some external control system has mistakenly tried to adjust load on a machine running in Isoch using the RAISE- or LOWER SPEED/LOAD targets. This only changes TNRI--which changes the speed/frequency reference and causes the unit to run at a higher or lower frequency than rated. So, if TNRI is not 100.0%, then it must be returned to 100.0% and no "manual" changes made to TNRI after it's returned to 100.0%.)

I agree--if the load was changing at a rate faster than 100mHZ/sec it's probably a safe bet that the load across the tie breaker was higher than recorded. But, I haven't given this much thought and I have never encountered such a situation as this. Load can change VERY quickly.

When a unit is running in Droop mode and the grid frequency increases above rated the unit doesn't "throw" or "reject" load. What happens is that the speed error decreases which decreases the energy input to the prime mover which decreases the electrical load on the generator. It's this part of your thought process that I'm having a difficult time with.

So, let's say GTG-A's AND GTG-B's Droop speed control references were both 102.4% at the time the grid disturbance occurred. When the grid frequency increased to 50.5%, that was a change of 1% of actual speed, which means the speed error went from 2.4% to 1.4%, which means the load of each machine should have dropped by 9 MW (25% of rated) to 13 MW. (The four MW which was being exported doesn't count; it's all about speed error.) If the plant is 48 MW and the two machines were generating 26 MW combined, that means that approximately 21 MW had to be imported to supply the plant load. (Again, if the load was changing very rapidly, that could account for the difference in recorded reading--but it's safe to say that at least 21 MW was coming in from the grid to the plant when the actual speeds of the machines were at 50.5% when the TNRs were 102.4%.)

Now, if the tie breaker opened at that point--with the speeds of the two machines at 50.5% and the Droop speed control references of both machines at 102.4% the outputs of the two machines would be 26 MW with a plant load of 48 MW--for a deficiency of 22 MW. This would immediately start decreasing the speeds of both machines, and if both machines remained in Droop speed control with Droop speed control references of 102.4% the frequency of the plant would decrease below 50.0 Hz because the resultant load (44 MW) would be less than the plant load of 48 MW.

If GTG-B automatically switched to Isoch speed control when the tie breaker opened AND if TNRI was 100.0% GTG-B should immediately start trying to return the frequency of the plant to 50.0 Hz. As the frequency approaches 50.0 Hz the speed error of GTG-A will start increasing back to 102.4%--which will make the output of GTG-A increase to approximately 22 MW. The remaining plant load of 26 MW (48-22) will be taken up by GTG-B to make the frequency equal to 50.0% (100.0% speed). (There is usually a small deadband on Isoch control of +/-0.13%, so frequency variation can be expected between 99.83% and 100.17% for normal Isochronous control, and this rounds up to 100%.)

By my calculations, the TNR of both machines (presuming the load was split equally) for a plant load of 48 MW would have to be 102.67% to maintain 50.0 Hz while operating in Droop speed control mode, while a TNR of 102.4% for both machines would result in 44 MW. This means there is a deficiency of 4 MW in being able to maintain 50.0 Hz. (The two machines would be producing 48 MW, but at a lower frequency.) 102.67%-102.44%=0.23%. Since 4% droop equals 2.0 Hz on a 50 Hz system, if my maths are correct, this would translate into a frequency difference of ((0.23/4.0)*2)=0.115 Hz, or a plant frequency of (50.0 Hz-0.115 Hz)=49.885 Hz. When the operator increases the Droop speed control references of both machines to 102.67% that would increase the frequency back to 50.0 Hz while providing 48 MW.

The type of loads in the plant will also have an impact on the timing of return to rated frequency; if they are mostly resistive loads it should be pretty quick. If there are lots of motors driving loads (pumping water and driving conveyor belts with heavy loads and driving fans, etc.) that can have an impact on speed of return to rated frequency (there is some "sponginess" of some loads as the speeds of the motors driving the loads changes--it's all about inertia, which is true of grids as well).

The time it will take for the Droop machine to "recover" is totally dependent on how fast the frequency returns to normal (from 50.5% to 50.0%). And because the plant load will be more than the 26 MW the two machines are producing when the tie breaker opens, plant frequency is going to try to change very quickly, and will probably dip below 50.0 Hz while the Isoch machine is catching up. But, the Isoch machine--presuming TNRI is at 100.0% should recover pretty quickly. Then it's all about how fast the plant loads can return to normal (the inertia/sponginess) of the plant load. And, the inertia of the GTGs, and the response of the fuel systems. (Gas fuel will usually respond quicker than liquid fuel...) This is another consideration because changes in fuel supply caused by sudden changes in fuel control valve position can also have knock-on effects, though they should be minimal. The Mark V fuel control valve regulators respond at 128 Hz; the Mark VIe at 100 Hz, and the SRV regulator of most Mark VIe runs at 25 Hz (40 msec).

If there's not enough energy flowing into the prime mover of a synchronous generator (or two generators synchronized together acting as one generator) to supply the power required by the load connected to the generator (or generators), then the frequency starts to decrease. Some of the energy that is being used to maintain 50.0 H while driving the load gets converted to that required to supply the load. The load doesn't change--the number of motors and lights and PLCs and computers and computer monitors in the plant doesn't change just because the amount of energy flowing into the prime mover changes; the number of motors and lights and computers and computer monitors remains the same--the load remains the same. And that's what the generator (or generators) has to supply. If there's not enough energy to do that, then the frequency will start to decrease, which will increase the speed error which will increase the energy flowing into the prime mover(s)--but not by enough to return the frequency to 50.0 Hz. The operator has to do that for Droop machines--and in this example that means the speed reference has to be 102.67% for both machines in order to produce 48 MW at 50.0 Hz.

Conversely, if the amount of energy flowing into the prime mover(s) is greater than that required to supply the load then the frequency of the system will increase. The speed error will decrease, but not by enough to return the frequency to 50.0 Hz--the operator has to lower the speed reference(s) to reduce the energy flowing into the prime movers even further to return the frequency to normal--for machine(s) operating in Droop speed control.

When one of two machines supplying a load (independent of a larger grid) is operating in Isoch mode, then it will sense any change in load--which wold TEND to change speed--and automatically respond by adjusting the energy flow-rate into the prime mover to maintain rated frequency of the system (including the speed of the other generator's prime mover). The Droop machine doesn't care much about maintaining frequency--it "presumes" an Isoch machine, or an operator or an external control system, is maintaining frequency. If frequency varies, it tries to help support load which tries to help support frequency, but that's not its primary job. Its job is to vary energy flow-rate into its prime mover in proportion to the speed error. And because it's generator is synchronized to other generator(s), they all operate at the same speed which is proportional to frequency (which is related to load--because more load tends to make (speed) frequency decrease and less load tends to make speed (frequency) increase.

This is about as much as I have time for today. I hope this helps!

 

Dear CSA,

I sincerely thank you for your time. I think you have understood my issue but still I need to fine tune my inputs to reach the objective of this discussion.

Prior to Disturbance: GT-A was running at 22MW in DROOP mode and Preselect mode was not enabled. GT-B was running at 22MW in droop mode and preselect mode was not enabled. Plant was importing 4MW from GRID as total plant connected load is 48MW. GT-B was programmed to come to ISO mode if the GRID goes out of sync.

By load throw I mean rejection of load by generators due to reduced error between TNR and actual rotor speed(GRID frequency).

There is another separate controller working for Load Shedding Scheme (LSS). This controller has all the online data for plant's electrical network and it has logic for carrying out load shedding as per predefined steps based on load priorities. I will call this system as LSS from now onwards. The LSS keeps calculating the Total Generation (GTA + GTB ± GRID) and Total Load connected. If generation is less and connected load is more it will execute the corresponding calculated steps of load shedding. There is scheme for under frequency relay input based load shedding also but this scheme was found not activated for this time disturbance.

My objective for this discussion is to whether we should review Over-frequency settings of our Inter Connection Protection, IPR (MICOM P343) as we are many times facing this over frequency issue. The GRID is a standby source for us and our objective is that if any disturbance occurring on GRID, Our system should go to Islanding mode without any Load Shedding. This time part of plant load (15MW) was shed and it took some considerable efforts to restore the system back.

We can compromise up to 8MW of load rejection by Generators due to over frequency.

If I take 8MW as allowed value of load rejection by the generators (GT-A: 4MW and GT-B: 4MW) due to over frequency, over frequency setting for IPR relay can be calculated, the values comes as 50.22Hz. Should I go for 50.2Hz setting for Over-frequency Stage and increase the pick for df/dt from 100mHz/s to 200mHz/s? TRIP for GRID shall be actuated by AND logic of 50.2Hz and df/dt>200mHz/s. You may also note that I calculated the rate of change of frequency for this time by MICOM P343 (IPR relay) disturbance as 800mHz/s. Also, the maximum frequency variations due to normal load ON/OFF operations within our plant are expected within 50.1Hz.

Another corrective measure can be that as an operating practice we should run both the machines at slightly higher load as compared to this time i.e. we can increase our normal running TNR values so that amount of load rejection is reduced due to over frequency.

Please provide your views on my approach.

 

neeuor,

I am not a power system engineer; I'm a lowly gas turbine controls technician/engineer.

I can sympathize with your plight--it's unfortunately so common in your part of the world and your part of the world is recognized to have such market potential that I believe it's one of the reasons GE has put so much effort (time and money) into off-frequency operation for their newest class of machines, the HA class. (That and the fact that with the emphasis on renewables and gas turbine-based generation in general the potential for grid disturbances is growing in other parts of the world and there is a growing need for units which can operate predictably in off-frequency conditions to "keep the lights on" for everyone.)

So, myself, I would recommend increasing the amount of load being carried by the GTGs during normal GRID-connected operations, and I would decrease the df/dt time as you have suggested. 50.5 Hz is about at the upper limit of my comfort zone with over-frequency operation, and I was presuming that 50.5 Hz was the relay setting--but now you're saying it's approximately 50.22 Hz?

As far as the GTGs go, they are not operating close to Base Load, and the problem being discussed here is over-frequency--not under-frequency (which is my mind represents more of a problem for the GTGs than under-frequency), so I don't see too much of an issue with a slightly high over-frequency relay setting--but, again, not for any appreciable period of time.

It's my understanding that when a lot of the frequency disturbances experienced in your part of the world occur they have very short periods--meaning the frequency changes fairly quickly and the excursions can be anywhere from 0.1Hz to 2.1 Hz, or more--with the tendency for them to be around 0.5-0.6 Hz on average. I think the issue is more about the period--how fast the frequency is changing than so much about how much the frequency is changing, when trying to remain synchronized to the GRID.

I understand that when separating from the GRID on over-frequency the problem is that the load of the GTGs may be less than the plant load--and so the GTGs have to respond quickly to pick up load to maintain frequency so as not to trip on under-frequency. And, again, here the issue in my mind is df/dt, not so much the magnitude of under-frequency. The turbines--again operating at Part Load for the most part--should be capable of responding fairly quickly (presuming the GTG in Isoch standby automatically transfers to Isoch at the time the tie breaker opens separating the plant from GRID).

I can't suggest or calculate any relay protection values for you--again, I'm not a power system engineer, and even if I was I imagine there would need to be a lot more information needed to make a proper recommendation with values and calculations. But--based on the information provided--in my opinion, you are thinking along the proper lines.

Hope this helps; it's all I can offer.

For those reading this thread, it's becoming more and more recognized that operating GTGs with Pre-selected Load Control is probably the single biggest contributor to grid instability when a disturbance is encountered. Pre-Selected Load Control makes a machine do exactly the opposite of what it should do when a grid frequency disturbance occurs--and in some parts of North America both VAr/Pf control and Pre-Selected Load Control are being targeted for elimination as they both contribute to grid instability rather than contribute to grid stability. I'm NOT saying the plant in this thread is operating with Pre-Selected Load Control--but I am saying that from my experience many plants in this part of the world DO operate with Pre-Selected Load Control, and that just makes grid instability worse.

 

Dear Neuur,

I would like to convey some facts / theory to you.

1.try to understand when the grid frequency increases. When a new big machine is synced with grid or a feeder trips. you have not mentioned whether your GTG-A & B are sufficient to meet the total load of your plant without GRID.

If your GTGs are sufficient then, keep the grid connecting breaker a frequency protection so that in case of sudden frequency increase it should trip, as you have LSS you can configure less important 6/7MW load with grid breaker. i.e. (isolated load -GTGs step load capacity).

2. The micomP343 relay is showing 27MW during tripping, this value is a instantaneous value. Dont go for it. rather, use micom S1 studio and check the current variation and the current pattern of your grid breaker, from there you will know the actual frequency and current change.

3. The calculation your have posted in 10th march with frequency 50.22 hz etc, you have not considered the time for breaker tripping i.e. initiation of trip signal in micomP343 & breaker tripping will need around 50-80ms time, situation will go more adverse in this time span, so you need to actually generate the trip signal before this time. otherwise your idea is ok.