I have one question regarding the operation of the gas turbines in islanding mode. our gas turbine is not having the mode of the isochronus selection. It is always connected to the grid and also supplying power to the proces plant. However if the grid breaker fails, then the machine will go to the islanding mode and maintains the house load.
I understood the diff.between the modes but I wonder if the machine is in islanding operation, what is the speed control mode: isochronous or droop?
Presuming you have a GE Speedtronic turbine control panel, there have been new frequency control schemes developed which actually keep the unit running in Droop Speed control mode while maintaining frequency--in other words, without Isochronous Speed Control.
It would be necessary to review the sequencing used in your turbine(s) to know exactly how frequency control is accomplished. It may even be done with an external source from another control system (DCS or ???).
Isochronous Speed Control requires trained operators who understand the limits of operating a power island with a unit in Isochronous mode--especially with other units in Droop Speed Control mode.
Thank you for your reply. Ours is a GE 7 FA with Mark-VI. Somebody told me that we will have to increase our megavars when operating it in islanding mode. But I don't think it is needed.
When operating the machine in islanding mode once it went to temp. control and later it got tripped on under frequency when there was some demand. So if we want to operate the machine in islanding mode, what are the limitations we have to follow? Can you plz clarify?
If you are in island mode (or better still if your Gas turbine is supplying loads alone) your connected loads should NEVER exceed the capability of the gas turbine. This means that the total connected load should be such that the gas turbine control never reaches temperature control. You have to keep in mind that temperature control mode is a load limiter to the gas turbine, i.e. to prevent the gas turbine from being driven in overload condition. The same applies to the reactive power load. This should not exceed the capability of the generator/AVR.
Apart from that, you have to be sure what your control system is programmed to do. As I know it, the gas turbine is normally controlled in droop mode, and isochronous mode can only be selected manually. Apart from that, the control system will maintain the same setpoints for voltage and frequency that were present before the system was disconnected from the grid. Droop mode will NOT maintain a fixed frequency and voltage but will vary with the load you have (with reference to the setpoint). So if your loads are frequency critical, then you have to switch over to isochronous mode. Isochronous mode should NOT be used when the gas turbine is connected to the grid.
This is a very complicted subject, and very difficult to explain in writing.
Without the ability to look at your .m6b file to determine what kind of speed control is being used when your unit is separated from the grid and is operating in "island mode" (a very unspecific and widely misunderstood term), it's impossible to say with any degreee of certainty how your unit operates when separated from the grid and supplying an electrical load. We also have no idea of the load on your unit (real and reactive) which is present on the unit when it's separated from the grid.
There are so many possibilities as to what can happen.... For one possible scenario (only related to real power--MW), let's presume the unit was operating at Part Load Droop Speed Control Mode (NOT on Preselected Load Control) at approximately 100 MW and was suddenly separated from the grid and did NOT go into any kind of frequency control mode (Isochronous or some variation of "droop frequency control" which is now being marketed and referred to as 'power island mode control'). Let's further presume the load present on the unit after being separated was approximately 120 MW, near rated output for a Frame 7FA. The speed would decrease (because there is NO frequency control with straight- or Constant Settable Droop speed control) and as the speed decreases the air flow through the machine decreases (because the axial compressor is slowing down) and CPD (Compressor Pressure - Discharge) also decreases. At the same time, the exhaust temperature is rising (because the air flow is decreasing)--so the unit would likely go on CPD-biased Exhaust Temperature Control at some point in this scenario. Since the speed/frequency was probably around 98-99% (just a scientific, wild-@$$ed guess), the underfrequency relays on the generator or the electrical system would most likely trip the unit after some time delay. Sound like what happened at your site? (The problem would be greatly "amplified" if the unit was being operated on Preselected Load Control at 100 MW when the unit was separated from the grid.)
This author has been at MANY sites, especially in the Middle East, where the power plant "operators" and supervision believe that prime movers can maintain frequency when separated from the grid with a small amount of load WITHOUT switching to Isochronous Speed Control mode, or having some form of frequency control while still operating on Droop Speed Control mode. So they insist on a test where the unit is operated at Part Load while connected to the grid, then they suddenly isolate a "block" of load AND the gas turbine from the rest of the grid. The unit maintains the LOAD, but the FREQUENCY drops and the generator breaker is opened (tripped) by the underfrequency relay after a brief time delay--and they declare the testing to be a failure! (Let's not forget it takes several minutes to restore the electricity to the "block" of load which was just blacked-out during the "test".... There's some happy utility Customers!)
In every case, the turbine control system was never designed to (automatically) sense separation from the grid and automatically initiate some kind of frequency control--either Isochronous Speed Control or some form of Droop Frequency Speed Control mode, and they declare the testing and the results to be a failure and unacceptable, respectively. All it would take is a single contact from the utility tie-line breaker to put the unit into Isochronous Speed Control Mode--and, voila! Problem solved.
There is a fundamental difference between maintaining load, and maintaining load AND maintaining frequency. Systems and testing have to be designed to achieve desired results--the system has to be capable of operation in the intended mode, and the testing has to be capable of demonstrating desired operation.
When generators and their prime movers are operating in parallel with other generators and their prime movers on an electrical grid, the frequency of the grid--and all the generators connected to the grid--is being controlled by some means through a "central" control scheme. The individual generators connected to the grid are NOT controlling their own speed, it's being controlled by the "grid." So, when generators are disconnected from the grid, something in the prime mover's control system has to be enabled to allow the speed of the prime mover--and the frequency of the generator--to be at rated and be stable.
There seems to be some fundamental misunderstanding of speed control--just because speed is controlled through acceleration to FSNL (Full Speed-No Load) and synchronization--does NOT mean it's controlled once the generator is connected to a grid in parallel with other generators. The frequency of all generators connected to a grid is the same--one 150 MW unit cannot operate at a different frequency than 2000 MW of other electrical generators, nor can it have an appreciable affect on the frequency (unless the total capacity of all the generators is equal to the load--in other words, all the prime movers/generators are being operated at rated output and that output is equal to the load; in that case, any small increase in generation will increase frequency, and any small decrease in generation will decrease frequency).
When a GE-design heavy duty gas turbine-generator is operated on Droop Speed Control, the actual speed of the turbine is LESS than the turbine speed reference. If the turbine speed reference is 102% and the actual speed is 100%, the difference between actual and reference is used to control the amount of fuel being admitted to the turbine--and, hence, the power output of the generator. BUT, the fuel is NOT being controlled to make the actual turbine speed equal to the reference--because when the generator is connected to the grid the speed is fixed by the frequency of the grid. Droop Speed Control REQUIRES this differential to provide stable power output when connected to a grid with other generators and their prime movers--THIS is what is meant by "sharing the load."
However, when a generator is operated independently of a grid to supply power to some load (and we're speaking about AC electrical power generation), the speed of the prime mover and generator have to be very tightly controlled, and fuel is added or removed in order to maintain SPEED as LOAD increases or decreases. This is typically accomplished using Isochronous Speed Control.
It is believed that this is another problem with understanding power generation--that of "loading" or "unloading" units. The load is the load--it's the aggregate sum of all the motors and lights and transformers and air conditioners and heaters and refrigerators and elevators and pumps and--and the aggregate total of the output of the generators (driven by the prime movers) CANNOT exceed the load on the system--OR the frequency will increase. If the aggreate load exceeds the aggregate generation capability, the frequency decreases.
To operate an AC electrical generator independently of an electrical grid while supplying power (real and reactive) to some load, there are TWO main considerations: frequency and load. The amount of the load, including in-rushes experienced during starting of large AC motors, must not exceed the rated power output of the prime mover. And, in order to maintain rated frequency the prime mover must be switched to some kind of automatic frequency (speed) control mode--or the operators had better be darned quick at loading the unit while still in Droop Speed Control mode.
A unit CAN be operated in Droop Speed Control Mode while operating independently of an electrical grid and supplying power to a load--but the operator has to increase and decrease the fuel to the unit to maintain the frequency. In effect, the operators become the "frequency control" in the absence of some automatic frequency control being enabled in the prime mover's control system.
As for the VAr output, if a single generator is the only device "supplying" VArs to a load, the output has to be adjusted to match the reactive load present on the system when it's being operated separately from the electrical grid. Usually, that means reducing excitation, but it differs with the type- and nature of the reactive load the generator is supplying. In power plants, one should not believe everything one is told.... Nearly every site and situation is different, and while there may be the odd case where two sites and loads are nearly identical, the odds against that are very high.
If on a system with 4% Droop and turbine running in Preselect Mode and trying to supply 50% load, it's likely that a difference of 2% will be created between TNH and TNR.
If the load of 50%, for some reason (some condition in the field like incorrect valve opening etc), is not "achieved" despite the 2% difference between TNH and TNR, then will the Preselect load control increase the TNR (and hence the difference between TNH and TNR) still further in an attempt to match the load with Preselect Setting?
You are correct; if the droop has been calculated correctly and the unit is basically in a new and clean condition and ambient conditions are near rated and the constituents of the fuel being burned nearly matches the fuel constituents supplied to the packager during the requisition phase, a unit with 4% droop being asked to supply approximately 50% of rated load would have a TNR of approximately 102%. But there's a lot of "ifs" in the qualifications above. There's also compressor cleanliness, inlet filter condition, hot gas path component condition, etc., etc., etc.
ANY time a unit is being operated with Preselected Load enabled, the Speedtronic will adjust TNR as necessary to make the actual load equal to the Preselected Load setpoint. If a unit is rated at 120 MW at 59 deg F and 60% relative humidity, and the ambient is 100 deg F and the relative humidity is 86.7%, then 50% of rated load will NOT be 60 MW, and if a Preselected Load Setpoint of 60 MW is entered into the Speedtronic, it's more likely that the TNR required to produce 60 MW will be greater than 102%--how much greater is a function of all the "ifs" and the ambient conditions.
Valve position or valve position feedback has nothing to do with TNR or TNH when the unit is operating at Part Load with Preselected Load enabled. When the reference is a load setpoint, and the feedback is the value from the load transducer, the fuel control valve will be driven to whatever position is required to make the load feedback equal to the load reference--and, in the case of Preselected Load, TNR is driven up or down to achieve the equality of load reference and feedback.
Mark v guy,
you mean to say that as the ambient changes, which will changes the air density and resulting in final mass of combustion and produced torque and affects in final load?
How the ambient exactly results in different base loads?
Can u please clarify?
Moderator's Note: Mark V Guy no longer posts on this forum. Can someone else answer this?
When producing power at rated frequency, the axial compressor (of a single-shaft heavy duty gas turbine) is running at a constant speed.
When producing power while operating at Base Load (CPR- or CPD-biased exhaust temperature control) the IGVs are at maximum operating angle.
A combustion turbine is a mass flow machine--get more mass flow (air and fuel) through the machine and the torque produced by the turbine increases. If the mass flow through the machine decreases the torque produced by the turbine decreases.
So, because the axial compressor is rotating at a constant speed AND the IGVs are "fully open" (maximum operating angle), as air temperature--and density--changes, the mass flow of air (and to a very small degree, the mass flow of fuel) changes. And, that affects the torque produced by the machine--which affects the amperes produced by the generator.
Again, this happens when the unit is operating at Base Load, on CPR- or CPD-biased exhaust temperature control--when the turbine control system is putting as much fuel into the machine as it can to maintain the exhaust temperature reference, and when the IGVs are at maximum operating angle.
You can see this through the course of a single day when the unit is operating at Base Load. When the ambient temperature is at minimum, the power being produced by the turbine will be high. As the day warms and the ambient temperature increases, the air density will decrease, the mass flow through the machine will decrease and the torque produced by the turbine will decrease--which means the power being produced by the generator will also decrease. As the day goes on, and the sun goes down, and the ambient temperature decreases the power being produced by the unit will increase. If you plot the power being produced by a turbine (DWATT) when operating at Base Load over the course of a day, and also plot ambient temperature (or axial compressor inlet temperature, CTIM) you will see the effect of ambient temperature on power output.
MANY other things also affect the power produced by a heavy duty gas turbine--axial compressor cleanliness, IGV LVDT calibration, inlet air filter cleanliness, axial compressor mechanical condition (clearances), turbine nozzle condition, turbine bucket condition, ambient pressure, exhaust duct back pressure. BUT, ambient temperature has the same effect on power output regardless of these other conditions.
Hope this helps!
While you're plotting DWATT and CTIM, also plot CPD, TTXM and TTRX.
You can set the plot to record at the slowest possible rate (I forget what that is for Trend Recorder, or Trender) because you don't need high resolution (which will consume a lot of hard drive space). You can let the data gather while the unit is running (just minimize the Trend Recorder/Trender window)--it will NOT cause the unit to trip under ANY circumstances.
Please be sure to let us know what you learn!
TNH - Turbine Speed, High-pressure shaft (%)
TNR - Turbine Speed Reference, High-pressure shaft (%)
When the unit is operating with Preselect Load enabled, the Speedtronic will adjust TNR to whatever level it needs to be at to make the actual load equal to the Preselect Load setpoint--up to exhaust temperature control.
TNH stands for Turbine instantaneous speed and TNR stands for Turbine Speed Reference.
When your generator is connected to the grid, the instantaneous speed of the machine is determined by the frequency of the grid... the machine cannot spin at a speed slower or faster than the other machines connected to the same grid. This means that TNH is effectively constant and cannot modulate (unless the grid frequency changes). Now if the machine is at 5MW and its MW setpoint is changed to 15MW, then additional fuel needs to be sent into the machine to meet the load demand. The signal that "tells" the machine to supply additional fuel to the machine is TNR.... The Preselect Mode Control will raise the value of TNR to create a proportionate difference (depending upon the droop setting of the machine) between TNH and TNR. This difference is a crucial information to the machine which interprets this difference into a fuel valve open command (FSRN)....
P.S. I owe this understanding of the process to control.com and its posts. When I started even I had no clue to this concept. Read some earlier posts dating back to 2006 and you'll find very elaborate discussions on the topic...
We have four generators supplying the process load of 50-55 megawatts. The excess power is exported to the infinite grid. So mostly two generators are run on temp control mode (one gives 18.2 megawatts and other gives 20 megawatts) and other two are at preselect load setting (15 megawatts and 15 megawatts) depending of the instantaneous demand of the process.
During monsoon, the grid turbulence causes the breakers to trip frequently. In an event when process demand is 50 megawatts, the total export is 20 megawatts and the machine goes offline (disconnects from the grid), when the grid frequency was 49 Hz, will all the generators shed equal amount of load? What will be the final loading on each generator? What will be the resultant frequency when not connected to the grid? Will the frequency at the moment of disconnection from the grid and the frequency after the disturbance has settled down be the same? In such an event, which parameter as an operator should I first control? Is there a trip also associated with the rate of change of frequency?
All generators are on 4% droop mode.
Maintaining frequency is like riding a bicycle at a constant speed. If the road is flat, the amount of torque required to maintain speed (frequency) is constant. If the road starts to climb, it requires more torque to maintain constant speed; if the road starts to drop, it requires less torque to maintain constant speed.
Now, imagine four bicycles all coupled together, and trying to maintain a constant speed on a flat road while pulling a trailer with some crates on it. Imagine two of the riders are pedaling as hard as they can (the two Base Load units), and two of the riders are only pedaling at approximately 75% of capability and the bicycle is traveling at about 98% of the speed its supposed to be traveling at (49 Hz out of 50 Hz).
All of a sudden, 30% of the crates fall off the trailer, and all the bicyclists are still pedaling at the same "rate" supplying the same amount of torque to the pedals. Worse, the two pedaling at 75% of output maintain their output at 75% regardless of the speed. The bicycle "train" speeds up, and the two pedaling at maximum output can now pedal faster at the same torque!
If something doesn't tell at least one of the riders to reduce their effort (torque) the "train" will just reach a new speed equilibrium--but at a higher speed than desired (greater than 100%).
If one of the riders was varying his/her output to maintain the "train" at a constant speed, he/she would be acting as the Isochronous speed control person. His/her torque production would be increased or decreased as needed to keep the train's speed constant. That includes going up and down over hills (which can be considered "load") or adding or subtracting crates (which is also "load").
Here's this author's "prediction" for your site...
NONE of the units are presumed to be in Isochronous Control mode. Grid @ 49 Hz; two units at Base Load (on exhaust temperature control) producing approximately 38.2 MW; two units operating at Preselect Load with setpoints of 15 MW each for a total of 68.2 MW (approximately 70 MW) from all four units. Process is approximately 50 MW, with an export of approximately 20 MW, and suddenly the four units are separated from the grid.
With approximately 30% excess fuel flowing for the process load of approximately 50 MW (20/70 ~ 28-30%) and NONE of the units in Isochronous Control Mode, the frequency of all four units is going to INCREASE. It is presumed the rated output of each of the four units is approximately 20 MW, so the guess is that the frequency will increase to greater than 50 Hz, but the exact value can't be accurately determined. The load will be approximately 50 MW, approximately the same as before the separation, but the additional fuel will cause the frequency to increase since there is no load to "consume" the extra fuel--so the extra torque causes the speed (frequency) to increase.
The real problem is the two units running with Preselected Load enabled. The Preselected Load units will do whatever they need to do to try to maintain their power output at 15 MW each--adjusting TNR to try to maintain 15 MW each! As the frequency increases, the tendency for the Preselect Load units would be to decrease the fuel (because the error between actual speed and speed reference would be decreasing). Decreasing fuel would tend to decrease power output--but, Preselected Load control will increase TNR to try to maintain the Preselected Load Setpoint--and since the maximum load is 50 MW, the extra fuel just becomes more speed--frequency!
The other two units are already at Base Load, and being commanded to keep fuel flow as high as possible to make the exhaust temperature equal to the exhaust temperature reference. The problem with is that as frequency increases, CPD increases and power output increases--so the two Base Load units are also adding to the problem of increasing frequency!
In this scenario, there is nothing controlling speed (frequency). Droop Speed Control is straight proportional control; if the error increases between the reference and the actual, the fuel will increase. And the two Base Load units are just trying to produce as much power as possible without overfiring (exceeding the exhaust temperature reference).
Without quick action, the over-frequency relays (presuming they are present) might start tripping units off line.
The first thing to do is "cancel" the load control selection of ALL the units--the Base Load Units and the Preselect Load units. This is as easy as issuing a LOWER SPD/LD command from any source (a target on a screen, the bat-handle switch on an operator station or Generator Control Panel).
On one of the Preselect Load units, start manually lowering load (this is usually the fastest unloading rate) using the target on a display or the bat-handle switch. As load is decreasing, watch the frequency--and stop lowering load once the frequency is at about 50 Hz. It is suspected that the load will drop down to about 5 MW, but that's just a SWAG (Scientific Wild-@$$ed Guess).
Then, if you can, enable Isochronous Load Control of the unit with the lowest load (the one with approximately 5MW). Then, decrease the load of the other unit which was on Preselected Load until the load on the Isoch unit is about 50% of rated (rated is presumed to be approximately 20MW), or approximately 10. Then reselect Base Load for each of the two units which were previously on Base Load.
Actually, in thinking about the problem, it might just be best to open the breaker of one of the Preslect Load units as soon the separation occurs
and then cancel the load selections of the remaining three units and bring the speed of the remaining Part Load unit down to control frequency. Then, select the Part Load unit to be the Isochronous unit.
In the absence of the ability to select any unit as the Isochronous unit, the operators have to adjust the load of one of the units to maintain frequency--one of the partially loaded units.
Now, we can't know if your site has overfrequency relays, or the setting(s) of those relays. Some relays have a function that if the magnitude of the parameter being monitored changes quickly, they will trip in a shorter period of time (it is believed that a term for that is "inverse time" function). It's also not known if your site has underfrequency relays, and if they have a similar function to trip quickly if the magnitude changes quickly.
If the two units which were NOT operating at Base Load were NOT in Preselected Load control when the separation occurred, while the frequency would increase the problem would probably not be as severe as if they were in Preselected Load when the separation occurred.
The IDEAL situation would be to be able to switch one unit into Isochronous Control mode to start controlling frequency as soon as the utility tie breaker opened. The turbine control panel would do so by reducing its load as necessary to maintain frequency.
A prime mover is nothing more than a torque-producing device. The generator converts the torque into a medium that can be transmitted long distances over very thin wires, and converted back into torque through an electric motor.
When speed is being externally controlled--such as when generators are connected in parallel with other generators on a electrical grid--the torque produced by a prime mover is converted into amps in the generator.
If speed is not being controlled by some external means--such as when a power producing facility and its load is separated from an electrical grid--increasing fuel will not increase LOAD, the doesn't change because the fuel changes. Increasing fuel doesn't start more motors or turn on more lights.
If the load remains relatively constant, changing fuel will result in changing the speed of the prime mover (and the frequency of the alternator). If fuel remains constant and load increases (motors start, or lights are turned on, etc.), then frequency would decrease. The operator can increase fuel manually to maintain speed/frequency as load increases, or, if the unit has Isochronous speed control mode it will be done automatically.
Write back to let us know if this helps!
Many thanks for your nicely explained reply. All our operators in control room took print outs of your post and distributed in all shifts. We need such easy stuff for our understanding rather than all the complex talk about "turbo-machinery dynamics and behaviours in parallel", etc., which we usually hear from our more informed colleagues. Thanks.
We forgot to mention in the earlier post that there is some automatically generated command from a Load shedding PLC (Honeywell's 620-35 stand-alone PLC system), which sends a pulse and brings the generators to float mode. So we don't have to run to the panels to unlock the preselect modes of the generators.
In view of this information, we wanted to know how will the generators settle after disconnection from the grid. Their frequency at the time of disconnection and post disconnection? Will all the generators shed equal loads and finally settle to maintain the process load alone? How much would be the final loading on each generator if the operator doesn't intervene?
Our aim is to calculate a safe export value so that in an event of sudden disconnection we should still have our generators connected and running. Hence it would be important to know how much the frequency increases when a known export is suddenly disrupted. That may help us reverse calculate and keep the turbines humming during monsoon turbulences. Maybe we can make an Excel spreadsheet for various safe-export values vs. grid frequencies. Our process load is relatively constant.
Describing Droop- and Isochronous Speed Control mode in writing is very difficult, so If you have some understanding from the above--that's great!
If the load-shedding PLC drives the units down to the loads at which the frequency is at rated, then the operator has nothing to do. The units will supply the process load at the rated frequency as soon as he loading/unloading rate allows the total fuel to be reduced to the amount required to maintain the process load at rated frequency.
If you want a "bumpless" separation from the grid, the load needs to be reduced to the amount that is equal to the process load. If the load-shedding PLC drives the unit loads down to the point at which the total of the four units is equal to the process load (so that rated frequency is being maintained), then the frequency will jump to a value greater than rated immediately after the separation, then will be reduced to rated at the unloading rate activated by the input if the units all remain in Droop Speed Control mode.
If the loading/unloading rates of all the units are equal, it's likely the two units on Preselect Load will shed equal amounts of load. But, because of the way FSRN (Fuel Stroke Reference - Speed Control) is maintained at a preset differential above FSRT (Fuel Stroke Reference - Temperature Control) it will tank at least a minute or three for the Base Load units to reduce TNR (Turbine Speed Reference) to the point that load begins to decrease.
So, it's likely that the Preselect Load units will shed more load than the Base Load units.
If all the units were on "float mode" (Part Load Droop Speed Control mode) when the separation occurred, and the loading/unloading rates of all four units were the same then they would all shed load at the same rate (equally).
There are many consulting firms, in addition to GE, which can analyze the conditions and predict response to transients such as separation from the grid. It's just not possible to do it in a forum such as this; there are too many variables and too much information required to be able to do such a study via this forum.
If you can understand how your load-shedding PLC works, and the loading/unloading rates of the turbines, you can begin to understand how the units will respond to a grid separation event. If you need exact figures and timing, you need to commission a study for finite details and predictions.
A 2007 post helping me understand a problem 10years later!!! Bravo!!!
My issue is that the grid forces us to maintain a frequency so other generators can come in. To maintain frequ navy when we are given additional load, we are not permitted to increase speed, instead more natural gas is consumed to maintain frequency-the impact is that if the dedicated gas plant does not reach to supply more gas, through operator intervention, then the demand is transferred to the gas well which causes the wells to shut off via the subsurface safety valve to protect the reservoir.....
In the light of what you have posted, I think there are several solutions i I can pick. I am also thinking of dedicating a well to fill this demand for additional gas when it comes.
Many thanks again
We are trying to use a turbine in isoch speed mode - droop voltage, parallel with other generator which are in droop speed mode. When we synchronized the turbine we got about 2500 KVAR so the voltage got up and the breaker opened. What´s happening? We think if we can adjust quickly KVAR we could keep turbine on load.
I would like to know more about the statement, you can run the machine on droop still maintaining the frequency. Could you please elaborate how it is accomplished?
We have 6 7FAs with islanding provision. So when the machines get islanded will they still run on droop control and at the same time maintain frequency?
Are you operating all of the units as a "power island", independent of any grid or other units?
Without being able to see how frequency is to be controlled at your site (the "islanding provision") it's impossible for us to say for sure how the units will be controlled and in what mode they are to be operated in.
In general, the plant designer prepares a document that describes such operating conditions and the parameters and configuration of all the equipment in the plant. Have you consulted the documentation provided by the plant designers for details of the islanded operation?
Many times something generally referred to as a PMS, or Power Management System, is used when a plant is operating in island mode to (attempt) control of the island frequency and the loading of various machines to try to distribute the load per some scheme and to shed load, if necessary, to be able to maintain frequency in the event one or more units is unavailable and the island load exceeds the available generation. (PMSs can, and quite frequently do, have additional functions as well; they are very site-specific. A "PMS" function can also be integrated into an overall plant control system, typically called a DCS, though quite often it is implemented in a stand-alone controller.)
PMS functions may also include reactive load sharing, in which case the PMS would send signals to the various units to adjust voltage and VArs and Power Factor as designed. Is it possible the "PMS" at your site was not configured correctly for the operation you were attempting and was sending signals to cause what you saw and are reporting?
You can see we just don't have enough information to be able to be of much help. Please consult the documentation provided by the plant designers to understand how "islanding" was intended to be operated. In some applications, all of the units in an island are operated in Droop Speed Control mode and signals are sent to each turbine control system from a central control system (the "PMS") for control of load to maintain frequency and for control of voltage and reactive power sharing amongst the units. In such a case, it's not likely that any unit should be operated in Isoch Speed Control mode, but we can't possibly know how your site is configured; we can only speculate and that wouldn't help you very much.
Hopefully we have given you some useful information to help in your quest to understand how your power island can and should be operated.