Speedtronic MK5 - Frequency Deviations


Thread Starter


Dear all,

With more and more renewable non-synchronous energy coming onto the grid on the island of Ireland, each of the generators have been asked to carry out a RoCoF (Rate of Change of Frequency) study - to determine how each machine will react to fast grid frequency deviations (as much as 2Hz/second in Northern Ireland), to enable a higher number of renewables to operate.

I have a query around the operation of a 50Hz GE machine with MK5 controls which normally operates in pre selected mode with droop control enabled. I don't believe it will perform as expected by the grid operator.

I assume that the machine will not perform with typical 4% droop characteristics as the pre selected load controller will 'fight' against the droop controller. This assumption seems to be covered and confirmed in other posts I've read but I would like some clarification.

Will operating the machine in base or peak load encounter the same issues?
If so, can these controllers be disabled when the machine reaches the desired value, so that it operates in free governor action?

I would like to keep the new operating procedures (if any) as close as possible to the existing.

You are correct; Pre-Selected Load Control will fight Droop Speed Control to maintain the Pre-Selected Load Control setpoint during a frequency disturbance when the unit is at Part Load (below Base Load or Peak Load). Pre-Selected Load Control can be used to change load, but when the unit reaches the desired setpoint the operator should click on RAISE- or LOWER SPEED/LOAD to cancel/disable Pre-Selected Load Control. Contrary to (false) popular belief, when Pre-Selected Load Control is canceled/disabled the load WILL NOT change wildlly (unless there is a grid frequency disturbance, in which case the governor (the Mark V) will respond as it should). Pre-Selected Load Control should only be used to change loads when operating at Part Load and then should be canceled/disabled in order to provide "free governor response."

When the unit is operating at Base Load or Peak Load (or Peak Reserve if so equipped) in exhaust temperature control with the IGVs fully open the unit will not respond as the grid regulators desire, in fact, it will respond exactly opposite to "free governor action." GE, and other knowledgeable service providers, can implement code which will--for a specified period of time (that is, not unlimited) allow for over- or under-firing when operating on exhaust temperature control during a grid frequency disturbance. This was first done in response to electricity board requirements in the UK, and is well-tested and works well. But, as normally configured "from the factory" the Mark V does not have this capability and the Control Sequence Program must be modified to include it. (Having said that, it is possible that units in Northern Ireland may have been ordered with the option, but it was not normally provided with most units back in the Mark V production days.)

When a unit is operating in exhaust temperature control with the IGVs fully open it is trying to maintain an exhaust temperature setpoint--regardless of speed/frequency. When the grid frequency decreases the speed of the axial compressor will decrease which will reduce the air flowing through the unit which will tend to cause the exhaust temperature to rise, so the Mark V will reduce the fuel to prevent exhaust over-temperature and so the load will go down. Which is exactly the opposite of what the unit would do on Part Load in Droop Speed Control with Pre-Selected Load Control disabled. If the grid frequency increases while on exhaust temperature control and with the IGVs fully open the speed of the axial compressor will increase which will increase the air flow through the unit which will tend to decrease the exhaust temperature which will cause the Mark V to increase the fuel flow to maintain the exhaust temperature setpoint, which is the opposite of what should happen. So, it's necessary to modify the CSP to allow the governor to sense speed/frequency changes while on exhaust temperature control with the IGVs fully open to respond appropriately, but only for brief periods of time (I think it's typically just for a couple of minutes maximum). And the CSP modifications usually include an accumulator for the amount of fired time above exhaust temperature control during frequency disturbances--because this must be included in maintenance outage planning, of course (gotta sell those parts and services!).

Hope this helps!

the gingerbeer

I would however add that the implementation of Pre Selected Load control should be understood as there is more than one variety and there is more than one variety of Droop.

Whether or not pre-selected load is engaged the turbine control will normally react to frequency fluctuations provided it is not on temperature control. This is the primary response and will be seen in the first 5 seconds. Usually Constant Settable Droop is set to 4% and has a first order lag of 3 to 5 seconds. Also the speed control algorithm which sits between droop control and fuel control also has a 3-5 second first order lag in the fuel control feedback.

Normally pre selected load if it is engaged will monitor a setpoint and will issue digital raise and lower commands to the speed reference setpoint. This will have the effect of adjusting the load and so the response following the droop respose will be moderated linearly to recover the load back to the pre selected setpoint if engaged. This might be called the primary reserve response. this may blend into the secondary response depending on the Grid Code definitions agreed with the generator. If pre selected load control is not engaged then the unit will continue to follow by responding to grid speed changes in droop. Droop can also be implemented as "Linear" or "Non Linear". In linear droop it has a characteristic, normally 4% across the load and frequency range where by it is expected that for a 4% change in frequency the load will change by 100%. Non linear droop may have one droop characteristic near synchronous speed and different characteristic away from synchronous speed. This is intended to either sensitise or desensitise frequency response in accordance with Grid Company requirements.

Please also note that unit load can be altered by the operator giving speed reference raise and lower commands. It is not necessary to use pre-selected load to adjust unit loading or unloading.

Once the actual requirements are known it may well be possible to implement in the CSP, some sequencing which will support the required operation. This would normally be agreed in the first instance between the Generator and the Grid Company as an annexe to the Connection Agreement and Grid Code. The code would be implemented and then the unit tested by introducing test code to the CSP which would simulate step and ramp frequency excursions in accordance with Grid Company requirements and agreements. Such testing may require to avoid DLN transitions and may be constrained by emissions regulations which should also be considered.
Thanks for the useful response CSA, it's greatly appreciated.
It's strange that the 'normal' operating modes for these units doesn't provide the frequency response that the grid operator would surely want.

Most of the main gen sets - coal/gas etc on this island will have frequency response or transient loops included but this isn't the case with these OCGTs.

I don't know if I particularly like the idea of the machine over firing?

I would like the machine to respond within it's existing operating window, with the procedure mostly the same, but with a 'frequency control loop' or similar programmed. I'm hoping this is a request which has been made before!
"Normal" is usually dictated by the operator, and sometimes I have found that the operator has expectations that are not always necessarily in the control. This particularly happens where there is an owner who contracts out operation to others and can change the operator from time to time.

That aside - provided the unit is left in auto - that is with no pre selected load setpoint engaged, and is allowed to follow frequency under conventional droop then the nominal load can be adjusted using speed reference setpoint raise and lower commands. That would leave the frequency support component of droop available for short term frequency support. Another loop could be written in sequencing, whereby speed setpoint raise/lower signals could be generated following comparison of the current frequency with a frequency setpoint and perhaps include a deadband too. The rate of change of the setpoint and thus the loading or deloading rate could be made adjustable to provide something appropriate to the Grid Company by way of a secondary response.
My advise is that whatever you propose this should be agreed with the Grid Company and very well detailed to the OEM or the company who provides sequence support of the MKIV, V VI or VIe Controllers.

Conceptually it seems reasonably simple. these kind of arrangements have been made before. If your turbines have waste heat boilers, then the Grid Company may require a more involved system in which the gas turbine performs the droop function in terms of the whole plant, which basically means having a droop with a setting less than 4%, but as the boiler output changes over time after a step change in frequency then the GT output is modified to maintain a steady load, with frequency control over that as a secondary response.

If you have older controllers these may not be supported by the OEM, however there are a number of providers who can probably modify the sequencing as you wish. On many sites the longer term frequency adjustment is handled by the DCS for the whole station.

The overfiring thing is used on many gas turbine power stations in UK and is well understood by the OEM in terms of frequency excursions and prospectively the number of times it statistically can be activated in any year. From memory the demand rate was estimated at something like 3 event per year.

Pre-Selected Load Control was never meant to be used as a "full-time" operating mode; it's just that operators believe (for no known or logical reason) that if the unit load isn't being automatically controlled that it will drift willy-nilly, which simply isn't true.

Before Pre-Selected Load Control (which should only be used to change load--and once the desired load is arrived at then Pre-Selected Load Control should be canceled/disabled) there was only the bat-handle RAISE- and LOWER SPEED/LOAD switches for operators to control load (unless there was some kind of external load control which could be discrete signals or an analog signal); this was known as REMOTE LOAD CONTROL. Sometimes AGC (Automatic Governor Control). But, no matter what it was called, it still would fight Droop (straight Droop or Constant Settable Droop) when there was a frequency disturbance, unless some special modifications had been implemented.

Using Pre-Selected Load Control just gives operators--and their supervisors--a false sense of security about the machine "automatically" controlling load. They have NEVER tried just using the RAISE- and LOWER SPEED/LOAD buttons/switches to change load and then observe what happens when they reach the desired load and just watch. They are literally TERRIFIED to do so. For no good reason. Because the don't understand Droop speed control they don't know how it works.

Now, in some parts of the world (mainly industrialized portions) the grid has always been fairly steady so using Pre-Selected Load Control "full time" was never really a problem. But, as you noted, with the advent of of renewables on a very large scale that are subject to natural events (clouds; wind speed variations and lack of wind; etc.) the grid has become more susceptible to grid frequency fluctuations in these industrialized nations/grids. And, so, the thermal generators must be proven to operate as they should in these cases, which weren't so prevalent in industrialized countries/grids in the past. And, in it's original implementation Pre-Selected Load Control will fight Droop speed control's response to grid frequency fluctuations--which grid operators in industrialized areas DON'T like or want. So, GE has developed "Primary Frequency Response" which is a way that the Mark* senses a grid frequency excursion and temporarily "over-rides" Pre-Selected Load Control to allow Droop speed control (straight Droop or Constant Settable Droop) to properly respond to the frequency upset--while still giving operators and their supervisors the warm fuzzy they get when they see Pre-Selected Load Control is active. (They still believe the unit output should remain constant during a grid frequency disturbance and don't understand why it doesn't (shouldn't)) but they know that Primary Frequency Response is mandated by the grid operator and will live with that, while complaining about power fluctuations....)

Second- and third-world areas/grids have always had grid frequency issues, some more than others. And, the use of Pre-Selected Load Control has been proven to be a contributing factor to grid frequency disturbances and problems--BUT, the operators, their supervisors and plant owners simply don't understand how an AC power generation and transmission system works and believe their unit's power output should remain constant in the face of grid frequency disturbances and get upset when it doesn't. (That's why some grids/nations are promoting "free governor control"--which is just another word for Droop speed control without any load control bias.)

Overfiring for a short period of time isn't going to hurt a gas turbine. And, the level of overfiring is also limited (it's not open-ended; it's just a few (10 or 15 deg F) degrees increase in the firing temperature, and again only for a limited period of time.

All you have to do is to convince, cajole, persuade, threaten the operators to just use RAISE- and LOWER SPEED/LOAD to change loads. Just start a trend of power (DW or DWATT) and TNR (Turbine Speed Reference) and on some morning or afternoon when things are quiet in the plant, have everyone sit down and watch what happens when the unit is NOT in Pre-Selected Load Control and the operators use RAISE- and LOWER SPEED/LOAD to control the unit. (Of course, if the unit is to be operated at Base Load, the operator should select BASE LOAD and let the Mark* do its thing without any manual intervention.) Everyone will be surprised and amazed at how well the unit actually maintains load without Pre-Selected Load Control enabled and active. Actually, I've been to sites where Pre-Selected Load Control wasn't properly tuned and the unit "bounced" more than 1-2 MW above and below the Pre-Selected Load Control setpoint--and that was ACCEPTABLE to the plant operators, their supervisors and management!!! But, when Pre-Selected Load Control was disabled at the desired load setpoint the unit load didn't really oscillate very much around the setpoint at all. Yes; over a couple of hours it was necessary for the operator to make a small adjustment once or twice (and that was arguable, actually) but for the purposes of the test (and as long as the grid frequency was stable (TNH was stable) the load was stable <i>without Pre-Selected Load Control.</i> Amazing, I know. But, you would be even more amazed at how many plants <b>won't even try operating without Pre-Selected Load Control enabled/active!</b>

Finally, frequency response is part of Droop speed control. It's not necessary to add it, it just is. The energy flow-rate into a synchronous generator prime mover (steam turbine; gas turbine; reciprocating engine) operating below rated load is a function of the difference between the prime mover speed reference (TNR in the case of a GE-design heavy duty gas turbine) and the prime mover's actual speed (TNH in the case of a GE-design heavy duty gas turbine). TNH is a function of grid frequency (difficult concept in itself for most people--but it's true) and when the grid frequency is stable the error between the speed reference and the actual speed will be stable (as long as the speed reference is stable). When the speed reference changes while the actual speed is stable then the load (power output of the synchronous generator) will change. <b>AND,</b> if the speed reference is stable BUT the actual speed changes then then error between the two will change which will cause the power output of the synchronous generator to change--which is frequency response at it's most simplest.

Droop speed control is used by almost every synchronous generator prime mover governor because it is the operating mode that allows multiple prime movers and generators to be synchronized together and stably provide power as one large generator and prime mover (at the same frequency). So, it would be very surprising to learn the OCGTs didn't have Droop speed control (straight Droop or Constant Settable Droop) if they connect to (synchronize to) the island grid.

You can let your operators use Pre-Selected Load Control to change load (most are afraid they will get repetitive stress injury if they have to click on the mouse button too long (for more than ten or twenty clicks)) and then when the desired load setpoint is reached then can just click once on RAISE- or LOWER SPEED/LOAD to cancel Pre-Selected Load Control. That's a simple change to the operating procedures if they can't be convinced to just click on RAISE- or LOWER SPEED/LOAD--but I warn you, it's not the procedure that's the most difficult to change. It's the mindset of the operators--they just will resist and fuss and it will take weeks for them to become accustomed to this heretical operating method. But, after a while, it will become second-nature--for most. Some will never accept that a unit can--or should--be operated this way.

Or, you would pay GE more than USD20,000.00 to implement Primary Frequency Response....

Hope this helps!

I know this is an old chestnut. On a site I am visiting now they have the same operating mindset of keeping PRE-SELECT LOAD enabled.

Just as a thought process. If the proper operation is giving a raise or lower pulse after pre-select load set-point has been reached could you comment on why you think this has never been automated in the control sequence with a second level choice to keep the pre-selected load enabled. Or have you heard of this this control already in place


Most gas turbine operators seem to have a very narrow idea of what their job entails. For Part Load operation they believe all they have to do is enter the desired Pre-Selected Load Control setpoint and enable Pre-Selected Load Control and their job is done.

I can't say why Pre-Selected Load Control was ever added to the logic, except that someone (someone who didn't have a lot of actual gas turbine operating experience) thought, "Hey! This would be cool!" And, that was that. It was born. Was it properly vetted and made to be actually serviceable? Not really. Does it work for every application or site? Absolutely not. Is it ever properly tuned during or after commissioning? Usually not.

It wouldn't take much to write some logic to automatically cancel Pre-Selected Load Control once the desired setpoint was achieved. But I don't understand the bit about having it as a secondary or back-up. Now you want the control system to check to see what happens if the actual load gets too far adrift of the Pre-Selected Load Control setpoint? And, what if the site/area has grid frequency excursions--and the actual load changes (as it should!) because of grid frequency changes? If the "back-up" Pre-Selected Load Control was enabled then it would try to return the load to the setpoint, which is the exact opposite of what should be happening during a grid frequency excursion.

Just about anything is possible with programming changes--but the problem to be solved needs to be defined properly first. Why do you (they) want to "cancel" Pre-Selected Load Control once the setpoint is reached? Why do you (they) feel the need to have back-up Pre-Selected Load Control? What do they think should happen when there is a grid frequency disturbance--do they think the unit should maintain its load during the disturbance (because it shouldn't!)?

Why can't the operators just use the normal RAISE- and LOWER SPEED/LOAD targets/buttons to raise or lower load until the desired load is reached? Is that so hard? Yes, it may take a lot of clicking, and RSI (Repetitive Stress Injury) is always a possibility. Or, the easier alternative is to set a Pre-Selected Load Control setpoint, enable Pre-Selected Load Control and once the desired load is achieved click once on RAISE- or LOWER SPEED/LOAD to cancel Pre-Selected Load Control and then just monitor unit operation <i>which is their job, anyway.</i>

Again, logic/application code would be written to do just about anything--but, it's the definition that is the hardest part of the task. Understanding precisely what the problem is that needs to be solved--in its entirety--is the really hard part.

Tell them to imagine Pre-Selected Load Control is not there and try operating the unit for a week or two. They will learn a lot. (But most operators don't think they need to learn anything, so that idea won't get very far.) And, have them form a committee to define what it is exactly they are trying to accomplish--and to take all the different scenarios they can think of into account. And, then sit down with them and review their problem definition and their scenarios and arrive at an agreed-upon "fix" to be implemented.

Automation is not the answer to every problem. Automating some tasks can create unanticipated problems (just look at what Pre-Selected Load Control has done!). If the only thing they want is to make load changes without having to click multiple times on RAISE- or LOWER SPEED/LOAD that's doable. If they have to be operating on "free governor mode" (Droop speed control) when operating at any steady-state load, that's doable.

There's always the GE option of Primary Frequency Response, but it always costs at least USD10,000.00 minimum. And, it isn't even perfect, either.

What other tasks do operators have to do that they can't click once on RAISE- or LOWER SPEED/LOAD after the unit reaches the desired load, and watch for 5- or 10 seconds to make sure the load stabilizes where it needs to be, and then go back to eating their biryani and reading their newspaper (or surfing the World Wide Web)?

Your post asks a very narrow question without stating the reasons for the request. What is it that they (or you) are trying to do--in the overall big picture scheme of things? Answer that, and you are on your way to a solution.

I've had a chance to think your question over and I don't think you were looking for a solution.

Rather, you were questioning why the designer of Pre-Selected Load Control didn't automatically cancel it when the desired load was achieved, or give the operator the option to leave it enabled once the desired load was achieved.

Again--it goes back to "problem" definition. Most likely it was developed because it could be--with the introduction of digital turbine control systems. It seemed like a good idea at the time and could be rather easily implemented with a keypad and a CRT display. (I have heard some earlier Speedtronic turbine control systems, Mark II's I believe) had a thumbwheel which could be used to enter a load setpoint, but I never saw one of those panels with a thumbwheel for a load setpoint.

And, back when Pre-Selected Load Control was implemented in Mark* controls (they were called Speedtronic controls back then) there wasn't the issue there is today with grid frequency disturbances--at least not in North America, where the majority of GE-design heavy duty gas turbines were sold at that time (early 1980's). And, gas turbines weren't so prevalent as "primary" generation back then, except for small developing islands and countries.

The problem that using Pre-Selected Load Control "causes" when the grid which the unit is synchronized with is unstable is that it can make the grid frequency excursions worse--because it induces hunting in the Mark*, which does NOT help a problem with insufficient generation to match load (which is the most common cause of grid frequency disturbances--a unit, or units, have tripped off-line and there's not enough generation to match demand (load)).

That's why in many parts of the world, grid regulators/operators are telling power plants that they can't use Pre-Selected Load Control--once they learn what it's effects are. And, those same grid regulators/operators are, if they haven't already, telling power plants they can't use Power Factor Control or VAr Control for much the same reason--they can make grid stability issues worse instead of allowing a proper response.

So, because the only ways the designer of Pre-Selected Load Control were to: 1) click on RAISE- or LOWER SPEED/LOAD, or, 2) initiate a STOP, or, 3) click on BASE LOAD or PEAK LOAD, that results in the need to do one of the three to cancel Pre-Selected Load Control once the desired load is achieved.

It would be great if the original designer of Pre-Selected Load Control had included the method(s) you suggested--but, again, there wasn't really a problem to be solved at that time. It was more that digital control systems made this an easy "option"--and it wasn't ever really given another thought.

Also, most GTs back in the day were used for peak load power purposes or Base Load operation--not really for Part Load operation. Since they really weren't very efficient back then (less than 33% efficient in simple cycle mode; less than 50% efficient in combined cycle mode) they were usually operated primarily at Base Load or Peak Load--not at part load. Economic conditions have changed greatly around the world since then, and efficiencies have improved a lot. And, in many cases electricity is not the primary product in many combined cycle plants--it's usually steam.

Finally, changing anything which has been around for three or four decades in GE control systems is nearly impossible. In many cases, the person/people responsible are no longer working or available for consulting and those currently working in control system design and implementation just won't change something that's been around "forever" (which means before anyone can remember).

After thinking, and writing, about this today I think Pre-Selected Load Control should NOT be included with GE-design heavy duty gas turbine control systems--unless supplied with the options you propose.

In the interim, operators and their supervisors and plant management are going to have to rethink their operating practices in light of present-day realities. And, either discontinue the use of Pre-Selected Load Control, get someone to modify the sequencing/application code to implement your suggestions, or pay GE to do it.

Which do you think is most likely to happen? ;)

Thanks for that response. You got it right that I was thinking hypothetically and trying to conjecture why there has never be a re-inventing of the wheel for this type of unit operation.

I have been looking at the software change required. And it doesn,t seem such a difficult one to implement. I will probably write up a proposal as an exercise but likely it will never be approved or downloaded.

If you ever hear of a breakthrough in awareness somewhere in the world for PRESELECTED LOAD let us know will ya

thanks and regards