Frequency oscillations due to load control

D

Thread Starter

Dani

Hello everyone. We are trying to setup an automatic generation control system in a small electrical area (a small island). Due to the small amount of generators in the area, we need all of them contributing to secondary regulation. We are using a load control mode for the generators (gas turbines and diesel motors) that tries to fix a load, making the unit to go to the setpoint dictated by AGC. but when we put all the generators in load control mode, the frequency becomes very unstable. I think the problem is that this load control is interfering with the speed control canceling its decisions. Do you think i'm on the right way?

I think that we have to put enough deadband in this controllers to make them not to cancel speed governor.

Does anybody have experience in setpoint to raise-lower conversion for generation units?

Thanks in advance and sorry for my english.
 
I don't think your approach will give you the desired results. I will explain why:

1. You stated that you have a small island system to control. Most probably if you observe your frequency you will see that it is never exactly 50Hz (or 60Hz, don't know what is your nominal frequency), but oscillating about a mean value. Reason is that the load is continuously fluctuating, and your generators' control is trying to compensate. When you tried to ride over this control, by introducing your AGC, most probably you ended up worsening the situation.

2. Given you are an island, your primary target is to maintain the frequency as close to your nominal as is practically possible. In such setups you can never control the load (or for all that matters the global power output of your generators). Your generators must provide the power requested by the grid at any point in time, and this is continuously varying, so your generators must simply follow.

3. It is not clear whether in your setup you are changing the setpoints of the individual generators' governing systems, or whether you are controlling directly the fuel input to the generators. In the first part you stated that you are modifying setpoints, whereas in the last sentence you asked for conversion from setpoint modification to direct raise/lower control. I would not disable the generators' individual control, as you still need to cater for emergency situations, like sudden load rejection, sudden load pick up, etc. Your setup may not handle these.

4. Do your individual generators use droop control, or otherwise for speed governing? Do all the prime movers have the same droop settings? Do you run all generators in the same control mode, or do you have different settings, i.e. some generators in droop, and others in constant load mode?

My proposal is that instead of load control you design a master frequency control system. Use a PI controller to measure the system frequency and compare it against the required frequency setpoint. The output of this controller will determine what should be the global output of all the generators. If all your generators have the same droop settings, then this signal can be sent as a speed setpoint for all your generators. Mind you this controller should have a slower response than your generators control systems, as otherwise oscillations in frequency will arise.
 
P

ProcessValue

Machines running independently "must not be put in load control mode". load control mode is used only when there is a gird supply available or a large machine is in the system which is put in droop mode.

The Problem : cause of frequency hunting

case 1 : Load control in a independent machine.
Let us suppose that the independent machine is running at 50 MW. so this means that the independent section load is 50MW. now the machine is put in load control mode and the setpoint is kept at 50. if the load does not change nothing happens , but let us say that the section load went up by 4 MW , and the load is now 54 MW. the generator will instantaneously take the load. but you have given the governor the load setpoint as 50MW. The governor in order to reduce the load will reduce the fuel input into the turbine/prime mover. but as the section load is more or less independent of the machine frequency it remains the same. The machine will go on reducing the speed till it trips on under-frequency. the reverse happens in case of a load throw off. the actual load is only say 45MW but the setpoint is 50 , the governor so as to increase the load , increases the fuel input , which in turn goes on increasing the frequency till the machine trips on overspeed. this is the reason , for all turbine control systems a tie logic is introduced which checks if the machine is connected to the grid ,before it can be put in load control mode.

case 2 : all machines in a micro-grid put in load control mode.
this is similar to the above case but with more complexity. let us suppose that there are three machines supplying to a section load of 60 MW with three generators having the loads as 10MW , 20 MW , 30 MW . now let us suppose that the machines are now put in load control mode with the same setpoint as supplied by the respective machines. now let there be a load increase in the section by 5 MW. how this 5 MW is going to be shared by the three generators depends on

a. the response chara of the individual generators

b. the network topology or the distribution scheme in the island.

so from the above it can be seen that , it cannot be predicted and only a simulation study with ETAP or PSCAD will tell you what will happen in this condition. let us suppose that the machine 1 took 1 MW the machine 2 two took 4 MW and the machine 3 took none/minimal load in the transient condition. so with a stable frequency in the beginning , you will now be having three machine running at three different speeds momentarily , with the speed reduction governed by the droop characteristic of the generator. now the machine 1 with a increase in the load will reduce the fuel input so it will unload , a similar condition will happen in machine 2. the load shed has to be taken , it will fall on machine three. what you will have in this scenario is massive load and frequency hunting which will ultimately lead to the collapse of the island.

from the above i hope you understand the futility of having the machine put n load control when running independently.

The scenario : what you are trying to do .

though i am taking a guess here , i think you are using the AGS to optimize the micro-gird operation. the optimization algorithm is calculating the optimum setpoint for the generators to operate so that your operating costs will be the minimum / improve the efficiently of the system as a whole/ optimal load flow if you have a diverse system. you are trying to feed the optimized setpoints to the generators from this. am i right ?? . on the whole are you trying to optimize your generation in some way ??

The solution : How to do it in a more practical way

a. with the load control : the only way by which the setpoint control you are trying will work is when all the generators are connected to the grid. you do not have to export to the grid , the grid can be kept in the floating condition so that your generators will crater to the section load only. in this case , if there is a increase in the section load , the grid will take care of it. now the ags sensing that there is a import from the grid must calculate the new setpoints for the generators for the new load , thus bringing the import from the grid back to nil. the reverse is for a load reduction. this is the most common method i have seen. you can include a grid islanding scheme for protecting the island from grid level faults.

b. with droop control alone : this is suitable if you don't want to connect to the grid. all the machines are put in droop mode in this. here the speed raise/lower is used to control the load in the connected machine. ie the droop reference in changed in the machines to get a desired machine output. this is a slow controller and does not have quick response. this method takes more time for the optimum to reach and does not have good transient stability. you will have to disable the AGS during transient conditions to avoid hunting.

good luck with the project and write back with what you have done :)
 
Dani,

You should run one of your unit in isochronous mode and the rest in droop speed control mode to be able to run your island in a stable and controlled manner.

If you are running all your machines in droop speed control mode while operating as a small island, your machines will increase their generation as the load on your island increases, but the frequency will fall. And your machines will decrease their generation as the load on your island decreases, but the frequency will rise.

If you are running all your machines in isochronous mode while operating as a small island, as the load on the island increases, the frequency will start to fall, then all of your machines will try to increase their generation at the same time to bring frequency to rated, which will result in unstable frequency (too much fluctuation in frequency). And if the load on the island decreases, the frequency will start to rise, then all of your machines will try to decrease their generation at the same time to bring frequency to rated, which will result in unstable frequency also (too much fluctuation in frequency).

If you put one of your generator governor in Isochronous control mode and all the other in Droop control mode, when load changes, which will tend to change frequency, the governor of the isochronous machine will automatically change load to keep the frequency constant and the governors of other machines which are operating in droop mode, will change their output to compensate the change in load. Hope this helps.
-M M Ahsan
 
M M Ahsan,

one has to be careful when a mix of isochronous and droop controlled generators is employed.

In such a situation, the isoch controlled generator will take up all the system load variations, and will not allow the frequency to vary enough for the droop controlled machines to take up or reject any appreciable load, as the change in frequency would be small. In such a situation the system operator will have to manually adjust the setpoints of the droop generators to balance loads between the generators.
 
jojo,
In a small island you should set the largest machine to Isoch and all other units to droop. The frequency responce to Isoch is almost instant. If you are using AGC to control there is too much lag in the communications/calculations of the SCADA system. Remember the frequency in a small grid is much more unstable due to the frequency response characteristic of the island.

You have to watch the unit loading to see when you are to high or low on the Isoch unit and adjust the set point on the droop unit accordingly.
 
Kevin,

This is the part that people never seem to understand when it comes to islanded operation: <b>It takes constant vigilance of the load on the Isoch machine, with adjustments to the droop machine(s) to keep good frequency control.</b>

Everyone seems to think it's just a "set it and forget it" operation, and it's not. Not. NOT. <b>NOT.</b>

There are all these schemes for Isoch Load Sharing and Load Control and they all have too much lag in them, especially if the load can change "drastically" and "frequently" (and both of those terms are relative, to many things). Large motors being started across the line, or large motors being stopped very suddenly, or blocks of load being added or subtracted very suddenly.

I have really only seen one good automatic island control scheme in operation, and I suspect the only reason it was successful was because there weren't lots of large motors starting and stopping and the load was relatively predictable during most days and there wasn't a lot of intermittent transmission problems (block trips, etc.). But, the operators were also well trained and knowledgeable so that when there were upsets in the system they knew exactly how to respond to ensure the Isoch unit wasn't "under-loaded" or "over-loaded" so that it was capable of keeping the grid frequency fairly stable. The operators were also able to spot when the automatic load control system wasn't responding well to load changes and take manual control and keep the lights on!

But, most sites and operators don't get good training, and these automatic control schemes are never well thought out, implemented, or tested. And so there are lots of issues and misconceptions and blame where blame isn't due or appropriate.

Thanks for helping with this. Sometimes it feels like a very lonely struggle. Everyone just thinks that when you put one unit in Isoch mode, everything will be just fine without any manual intervention or observation.

And, then they try to set a load setpoint on the Isoch unit and don't understand why the grid frequency is not constant!

 
Dear All

I want give example for you to explain by suppose the load.

we have three generator and capacity 20 Mw . two turbine /generator loaded with 12 Mw. Total load 36 Mw. what happen if we select two generator with droop mode and one generator with Ischo. mode,
Which turbine /generator will pick up the load if the load increase 2or 3 Mw?

Can i change control from ishco to droop on line?

or can i change control from droop to Ischo on line? or when i can used these mode?

How we can sharing the load between generator in the diff. control system?

If generator but in isochronous mode or control can I control from HMI mark VI or need external control?

Thank you , I hope it is clear my question
 
I would still recommend what I stated earlier, i.e. that a PI controller monitors the frequency and sends setpoints to all the generators in a system appropriately.

As CSA commented, one should let the droop control of each generator to handle all the power demand surges in such island grids. The PI controller will then correct the generators' setpoints to bring back the frequency to nominal after a sudden change. Slow changes will be corrected by the PI controller.

My recommendation is that Isoch control is used only in stand alone generators. One must in addition keep in mind that undue stresses will be imposed on GT components when operating in isoch control and the GT is exposed to sudden and relatively large power surges. In its attempt to maintain the frequency steady, isoch control will increase/decrease firing temperatures fast.
 
If you have three generators each with a capacity of 20 MW, and a total system load of 36 MW, and you are operating two of the generators in Droop control mode each with 12 MW of load, then the third generator will also have 12 MW of load. If the third generator is in Isoch mode, if the load changes by any amount at all then the third generator, the one being operated in Isoch mode, will respond to the load change.

The load change will tend to cause the frequency to decrease (presuming the load increases) but the Isoch unit will very quickly adjust it's fuel to increase the power output of the unit in order to keep the frequency at the setpoint.

The other two units operated in Droop control mode <b>should not</b> be in Preselect Load Control mode. They should just be manually loaded to the desired setpoint (12 MW in your example) and left there in Part Load Control.

So, let's say the load increased from 36 to 39 MW and the load on the Isoch unit increased from 12 to 15 MW. Now let's say the load increased further to 42 MW. The load on the Isoch unit will increase to 18 MW, which is getting close to the 20 MW capacity of the unit.

Now, let's say the load increases another three MW to 45 MW, the Isoch unit can only increase it's output from 18 MW to 20 MW which means that the frequency of the system will start to decrease because the Isoch unit can't produce the third MW required to maintain frequency.

As the frequency decreases below setpoint, the two Droop units will increase their power output to make up the 1 MW deficit to produce the power required by the system--but the frequency will be less than desired. Again, this presumes the Droop units are <b>NOT</b> being operated in Pre-Selected Load Control mode at Part Load.

Now, all the operator needs to do is to load one or both of the Droop units manually to some load above the 12.5 MW (this presumes the two Droop units have the same Droop setpoint so they will share the 1 MW load deficit, each supplying an additional 0.5 MW). Let's say the operator loads one of the Droop units to 15 MW.

As the Droop unit is loaded, the frequency will start to increase. And the load on the other Droop machine will start to decrease as well, back to 12 MW where it was operating before once the frequency reaches the setpoint. The load of the Droop machine being loaded by the operator will be at 13 MW, the other Droop machine will be at 12 MW and the Isoch machine will be at 20 MW, for a total of 45 MW, and the frequency will be back at the setpoint.

Once the frequency reaches the setpoint as the load of the Droop unit is increased further the frequency of the system will start to increase. But, the Isoch machine will reduce its power output to maintain the frequency. With one Droop unit at 12 MW, the other Droop unit at 15 MW, and a total system load of 45 MW the load of the Isoch unit will decrease to 18 MW. If the operator loads the other droop unit to 15 MW, the load of the Isoch unit will decrease by another 3 MW, to 15 MW. Now, the two droop units will be at 15 MW, and the Isoch unit will be at 15 MW, and the system frequency will still be at the desired setpoint.

When a turbine is in Isoch control mode, an operator doesn't control the load of the Isoch machine directly. You can change the load of the Isoch machine by changing the load of the Droop machine(s) as in the example above, but the only thing the operator can directly control on an Isoch machine is the frequency setpoint. The load will be a function of the system and the load on the other Droop machine(s).

And, if you try to operate Droop machines in "island" mode with an Isoch machine and you put those machines in Pre-Selected Load Control you will most likely have frequency stability problems if the load on the Isoch machine is at or near its capacity/rating.

For an island to operate properly, the load on the Isoch machine must always be monitored such that it is never at nor near the maximum capacity of the unit or at or near the minimum capacity of the unit. And, the way to adjust the load on the Isoch machine is by adjusting the load on the Droop machines!

If the load on the Isoch machine is at or near the maximum capacity of the unit and the load on the island system increases, then the frequency will likely decrease.

If the load on the Isoch machine is at of near the minimum capacity of the unit and the load on the island system decreases, then the frequency will likely increase.

So, it's important to keep the load on the Isoch machine somewhere around a point at which the maximum expected load changes will not cause the output of the unit to exceed its capacity/rating, nor cause the output to be less than zero MW.

And, again: The way to adjust the load on the Isoch machine is by adjusting the load on the Droop machine(s).

Even if there is an external load control system trying to maintain frequency, it should <b>NOT</b> be adjusting the output of the Isoch machine. The Isoch machine's governor (turbine control system) will automatically adjust its output to control the frequency as long as the power required is not more than the capacity/rating of the machine, or less minimum (zero). Any external system should be monitoring the load on the Isoch machine, and changing the load on the Droop machine(s) to keep the load on the Isoch machine in a range that it can respond to any expected load changes.

As for whether or not you can change from Droop to Isoch when the unit is on-line, that depends on how the turbine control panel software was written and configured.

That's how to use an external load control system; not to control the Isoch unit, but to control the Droop machines to allow the Isoch unit to respond to load changes.
 
I want to address one issue that was not completely addressed in my previous response, that of when the various modes can be used and when you can switch between them.

We don't know if your power island can be operated in parallel with a larger grid or if it's continuously isolated. If your facility is operated in parallel with a larger grid you should <b>never</b> select one or more units to be in Isoch mode. I will even go so far as to say if your facility can be operated in parallel with a larger grid that there should be interlocks to <b>prevent</b> putting any unit in Isoch mode.

If your facility is operating as an island, separated from a larger grid, any <b>one</b> unit can likely be set to be in Isoch mode, but you should <b>never</b> put two units in Isoch mode at the same time. (You will never forget the experience; don't even try it without a good torch in your pocket.) [NOTE: The exception is if there is some kind of Isoch Load Sharing mode/scheme enabled, and while they are sold and installed in some plants, they rarely work as advertised.]

If your facility is operating as an island and one unit is in Isoch mode and you want to "swap" Isoch (frequency) control to another unit, you should wait until the load is fairly stable and put the unit currently operating in Isoch mode into Droop mode, and then put another unit in Isoch mode. If the load is stable, the "swap" should be relatively smooth.

Remember, you cannot "pre-select" a load on the Isoch unit. And, you should not operate any unit in island mode in Pre-Selected Load Control mode if the expected load changes are sudden and/or large, unless the Isoch unit is being operated such that it can handle the expected load swings without reaching maximum output or zero output.

If you click on SPD/LOAD RAISE or -LOWER of a unit operating in Isoch mode, all you are doing is changing the speed (frequency) setpoint, <b>NOT</b> the load setpoint. The governor (turbine control) will automatically adjust the load of the Isoch machine to compensate for any change in load which would tend to change frequency. That is, up to the maximum capacity of the Isoch unit, or down to zero output of the Isoch unit.

Lastly, to reinforce the point: The only way to change the load of the Isoch unit is to change the load of a Droop unit. If you increase the load of a Droop machine, the load of the Isoch machine will decrease. Conversely, if you decrease the load of a Droop machine, you will increase the load of the Isoch machine. But you can't control the load of the Isoch machine from the turbine control system of the Isoch machine. It's already been told (when it was put in Isoch mode) to automatically adjust it's load to try to maintain the system frequency and that's exactly what it will do. As the system load changes, which would tend to change the system frequency, the governor of the Isoch machine will change it's load accordingly (up to the maximum rating of the machine, or down to zero output of the machine, at which point the frequency will decrease or increase, respectively).

The issue of speed control and how Droop and Isochronous machines work together to control speed (and frequency) had been covered many times on control.com. An analogy to a tandem bicycle with two riders who have to work together to maintain a constant speed (frequency) has been used to describe what can happen if load changes or one or both riders are not working together to maintain the speed (frequency).

An electrical system is no different than the tandem bike analogy. Use the 'Search' feature of control.com to learn more about speed (frequency) control.
 
N

not_quite_nyquist

I know this is an old post, but the moderator rejected the new post that I tried to start... Thought you might be able to help me out...

I've been told numerous times that multiple generators cannot be run in iso mode without some sort of load sharing scheme... and being that some of these statements were proclaimed by people that have been in the industry longer than I have been alive, I didn't question. Recently, I placed two small laboratory synchronous generators (200W units) in isochronous control, synchronized them, and expected to see something fun and violent. But, I was thoroughly disappointed when nothing happened... they just ran together perfectly fine. I performed some step load testing, and after 20+ minutes still nothing.

While I am not doubting numerous people have had bad experiences when multiple units are placed in iso mode without load sharing, I'm really curious to know exactly what happens, and why... not just blanket statements like "they will hunt and possibly go unstable." If anyone has an explanation related to control theory principle, I would appreciate it greatly. And if you know exactly how iso load sharing is supposed to work, that would be nice, too.
 
B
If you consider an "isochronous" control as one which tries to maintain a constant speed using an integrating term in the speed feedback loop (as opposed to a droop control which is purely proportional) ...

Consider two machines, A with a speed set point of 50.000 Hz and B which has a set point of 50.001 Hz.

If A is initially running and loaded to say 80 % with the resulting in equilibrium, and B is then synchronised **, B will initially pick up a small amount of load. It will stabilise at 50.000 Hz, and will see a speed error of -0.001 Hz. As a result, it will increase output to try and bring the speed up. Because there will now be an excess of power in to the system, the overall speed will increase (at a rate which depends on the total connected inertia of the system).

As speed is increased, A will now see a positive speed error and back off its mechanical power source. Eventually, A will be at 0 % and B at 80 %. The dynamics of how the transfer of load from one to another is complex and will depend on things like the inherent time responses of the prime movers, control system tuning settings, and so on.

Now imagine that problem repeated several thousand times - the interactions become a lot worse and far more complex.

** Before anyone quibbles that two machines cannot be synchronised if the speeds are different, I would point out that if the synchroscope pointer is rotating there must be a small difference in the two speeds.

In a real situation, the controllers will respond to errors between the set-point and measures speed signal. This may include actual differences between the levels of the signal, or just errors in measurement or signal transfer. If you are making measurements and carrying out control calculations in say a single digital computer, the measurement error terms may disappear and not come into play - but in any other system a set-point of 50.000 Hz on one machine may appear as 50.001 on another.

Cheers,
Bruce.
 
N

not_quite_nyquist

>(as opposed to a droop control which is purely proportional)

Minor quibble: I would disagree that droop control is purely proportional. Droop still controls the the prime mover to a specific speed reference (so it needs to have integral control action), however that reference changes (based on the droop setpoint and droop factor) as the machine is loaded and unloaded... as opposed to the isochronous controller which is set to 50/60 and left alone.

Regardless, if I understand you correctly, you are saying that in some controllers analog signals are used to set the speed setpoint. Errors/noise in that analog signal as seen by the controller will cause each unit to be driven to slightly different speeds, which causes load to transfer between the different machines, and eventually leads to one machine hitting its rails while the others may be relatively unloaded? That makes a lot of sense to me.

So, that would lead me to believe that iso load sharing is accomplished by looking at the output of the group of generators and biasing each unit's setpoint so that each unit is sharing according to predefined rules... i.e. if a unit is running faster than the other and taking more load, its setpoint is biased lower so it starts to unload, while the slower machine will be biased higher and begin to pick up load... and they magically meet in the middle.
 
B
A droop control is purely proportional in that, if the speed of the machine falls, the power developed will rise. For a conventional power turbine, the normal variation is about 4 % so that if a full loaded machine is running at 3000 rpm and it is gradually unloaded the speed will rise to 3120 rpm. This may be done in a number of ways - the approach of using an internal integrating element as part of a cascade configuration is only one.

In ANY feedback controller, a measurement is made of the controlled variable and compared with a set-point. Any measurement process will have errors of some sort - whether analogue or digital. An integrating term in the controller will act on whatever error is present (however small) and act to drive the manipulated variable up or down.

There are a number of different strategies for load sharing. One is to put most of the machines on to power control so they maintain a fixed output, with a single set used to maintain frequency. The power set-points on the fixed-load machines can be altered to hold the power output of the variable set at around 50 %. But, in general, trying to operate a bunch of different producers of any sort as independent systems with "matched" set-points is very difficult - the same problem arises when running several boilers connected to a single header.

Bruce
 
N

Namatimangan08

Questions:

1. What is your selected AGC regulation time step? 4 seconds?

2. What is your droop percentage set point, the maximum number of units on bars and the biggest per unit capacity of a prime mover on bars?

3. If you have a set point which is called frequency control set point, what is its value in MW/0.1 Hz?
 
E

egacarpentieroc

hello,

what do happens with a generator operating on power control, if the frequency of grid changes?
 
egacarpentieroc,

The answer depends on several variables.

Is the prime mover at or very near rated load, or at or near zero load?

How severe are the frequency variations?

How is the prime mover governor (control system) configured?

IN GENERAL, if a governor is in Load Control and Droop speed control is NOT enabled and active then the governor will try to maintain the load setpoint during frequency deviations. BUT, if the load control scheme is driving the Droop speed control reference which is looking at actual prime mover speed then Droop speed control will try to change the load per the frequency deviations BUT load control will counter that and try to maintain the load control setpoint--which is the EXACT OPPOSITE of what the prime mover should be doing to try to help stabilize the frequency.

It's a common misconception for people to think the unit at THEIR site should maintain steady and stable load during frequency excursions--but that's not the case. If grid frequency decreases it's generally because the present generation is less than the present load--so in order to maintain the load some of the energy required to maintain rated speed is used to maintain the present load--which means the frequency of the system will decrease. The opposite occurs when the present generation is greater than the present load--there's too much energy for the present load so the excess energy causes the system frequency to increase.

If all the units don't respond properly, then the grid can be severely affected. AND, if the grid frequency decreases (because the present load exceeds the present generation) AND some or many units are already producing rated power--those units can't produce any more power. So if the units operating at part load (less than rated power) don't pick up the load because their load control scheme is fighting Droop speed control then the grid can be severely affected.

Droop speed control provides assistance for grid frequency deviations--but only when it's free to be able to respond appropriately and when a unit (prime mover and generator) are not at or near rated load during a grid frequency decrease. If Droop speed control is being over-ridden by load control then the unit is actually contributing to grid instability instead of supporting gird stability.

Hope this helps!!!
 
Top