Control.com - Nerds in Control

Hello all. First time poster so forgive any mistakes.

I work in a plant which operates six GE Frame 7EA DLN-1 gas only simple cycle CT's. These units were installed in 2000 and went commercial in 2001. All units have MARK V/<I>/EX2000 for control and use Constant Settable Droop.

The company which owns the plant, and many others, wants to develope a general procedure for testing the "Turbine Governor Deadband and Droop" for the fleet. Each plant then would make the procedure specific to their equipment. I have spent the past couple of days, on and off, searching for a procedure for testing Droop control to no avail. I have searched our manuals and there is no mention of testing once it's set.

Based on the above, I have a few questions:

1. Does anyone else do this type of testing?

2. If so, where did you get the procedure or did you develope your own?

3. Is there really a need for this? It seems to me that once the constants are set, nothing should change.

Thanks in advance for any information.

RichT88

Can you tell us how your units are operated when they are operated? Is this primarily a peaking plant, and when the units are run are they usually started and loaded to Base Load? Or are some of the units started and loaded to Part Load during peaking service? (This doesn't affect the answer to this question, really; I'm just trying to understand how gas turbine plants are being operated these days.)

1. Yes; there are some utility regulating organizations which require this kind of testing in various parts of the world, so it's done. It's very difficult to do without modifying software, and it would require forcing logic and manipulating Control Constants to do so manually, and often the results of such manual tests are not generally recognized by the regulators requiring the testing (see Number 2, below).

2. GE sells "governor" testing software which can be implemented in most Mark V control panels. I'm not clear on the exact details, but I believe it involves temporarily "fooling" the control panel into thinking there has been a frequency disturbance to affect a change in output. That change can then be used to calculate the amount of droop. It's an innocuous test, and one can see how it would be difficult to do such a test without some special software modifications so as not to cause instability during the testing.

3. As has been noted in other recent posts, there can be changes in fuel supply characteristics and machine conditions (compressor cleanliness, hot gas path part condition, etc.) which will have a slight effect on the *exact* amount of droop at any given time. Remember, gas turbine output is affected by ambient conditions (temperature, barometric pressure, humidity), as well as machine conditions (cleanliness, filter condition, exhaust back pressure, hot gas path part condition, etc.). But, generally, if the machines were properly configured and the fuel supply hasn't changed appreciably since original configuration and the machines are clean and in good condition, you are correct: the nominal droop setting won't change. It seems some utility regulating organizations believe that some owners/operators would change their droop setpoints from time to time and therefore require testing; who knows. Most digital control systems don't have any appreciable "drift" like older analog control systems with pots and dials had.

Personally, I have a difficult time understanding why these utility regulating organizations want to know the "actual" droop characteristics of machines which are primarily operated in peaking mode and usually at Base- or even Peak Load during peak load conditions, when a grid frequency disturbance is likely to occur. Droop speed control is not active at Base Load (or even Peak Load for that matter, if the unit has such capability and were being operated at that condition). (Actually, droop speed control is active, it's always active, but when Base- or Peak Load is selected it's biased out of the way so that it doesn't interfere with exhaust temperature control which effectively takes it out of the control loop.) When being operated at Base- or Peak Load the turbine control panel is trying to put out as much fuel as it can for the given operating condition in order to make as much power as possible and it's presuming the grid frequency is at rated and stable. If the speed of the axial compressor changes because the grid frequency changes, then the power output of the unit will be adjusted accordingly in order to maintain exhaust temperature so as not to "over-fire" the machine and cause undue stress on the hot gas path parts.

Gas turbines being operated at Base- or Peak Load with typical as-shipped software will *NOT* respond to grid frequency disturbances as they would if they were being operated at Part Load. It's a nasty little secret of gas turbines that when they are operating at Base- or Peak Load and the grid frequency decreases, the axial compressor speed decreases, which in turn causes the exhaust temperature to increase, which the turbine control system responds to by decreasing the fuel to maintain the exhaust temperature setpoint, which results in power output decreasing, which is exactly the opposite of what one wants to occur when the grid frequency drops. If grid frequency increases, the axial compressor speed increases which causes the exhaust temperature to decrease and the turbine control system then increases fuel to try to maintain the exhaust temperature setpoint, which is also the exact opposite of what one wants to occur when the grid frequency increases.

The UK mandated special software for all combustion turbines to detect over- and under-frequency conditions when the units were being operated at or very near Base Load to cause them to "over-fire" or "under-fire" to help maintain grid frequency as appropriate, again *while operating at or very near Base Load (or Peak Load)*. These over-fired conditions are closely monitored and recorded by the control system and a high maintenance factor is applied to the time of over-fired operation when determining maintenance intervals. It kind of goes without saying that these under-frequency conditions are "expected" to be of short duration, so as not to unduly over-stress the hot gas path parts.

But this droop testing, and annual droop testing, of combustion turbines which are primarily operated at Base Load (even if they are not peakers) doesn't make any sense, at least to me. I would think it would be more important to ensure, as the UK does, that combustion turbines will respond as desired to grid frequency disturbances when they are being operated at Base Load. Part Load operation, on Droop Speed Control, could be tested and demonstrated once, during commissioning, and unless there were unusual changes to fuel supply or operating conditions the droop characteristic, as you say, should never appreciably change (particularly for digitally controlled machines).

Hi CSA. Thank you very much for your detailed and informative reply.

As for how our machines are operated, we are a "peaking facility". That is to say we operate during peak demand times. Usually, we are dispatched to Base Load. We do not have the "Peak Load" option. When we were owned by a different company we would be dispatched to meet a certain amount of MW's and have to operate some units at "Pre-Selected Load". Since we are now owned by the same company which owns the transmission lines (which was almost always the reason we were limited) we are rarely called on for anything other than Base Load. By the way, we are in the US.

I was not aware when operating at Base Load that Droop speed control isn't really active (or in control may be better). But now that you pointed it out I understand why that is. This also adds to my question of "do we really need to do this?". We are a 500 MW nominal plant connected to very large grid and we do not have AGC. It just seems to me that our plant just kind of "rides along" the grid and can't really do anything to change the frequency of the entire grid. Furthermore, just as you stated, it was set up and tested during commissioning and nothing of significance has changed. I'm not sure if this some regulatory agency test requirement or the company's testing requirement. Our company owns and operates primarily coal units and in the last few years started adding gas fired CT's to their fleet. As you know, there are big differences. What applies to them doesn't always apply to CT's.

I will check into the software that GE sells. It was my understanding when they asked me to help with this that it was to be an off-line test for reasons you stated. We don't want to be running and cause a needless trip or damage the units. But if it has to be on-line to meet requirements, then so be it.

Hi Bruce. Thanks for sharing your procedure. Do you check the positions of your gas valves as well? That is something we will be having to check if this procedure is implemented.

Thanks again to both of you.

RichT88

On the old Mark I turbine controllers, we routinely calibrate the droop values from the analog Mark I. This is merely a simulation of what we hope will happen during operation.

Perhaps, the Mark V digital controller could also be fooled by placing a signal generator on the proper speed pickup and viewing the droop responses/digital setpoint values through demand display. By increasing/decreasing the observed machine speed with the frequency generator, one can observe the response of the machine's setpoint.

Of course, CSA also pointed out that the machine might not respond the same while operating.

Your 7EA DLN machines are also in danger of this phenomenon.

"Generator Owners and Operators who own or operate large frame combustion turbines should be aware of potential “turbine combustor lean blowout” under certain frequency excursions."

Turbine simulation and turbine operation are two entirely different worlds.

Please let us know how it works out.

Hi CTTECH,

Thanks for the input. When you do your testing do you also check the positions of your gas valves? That seems to be the main point of the procedure they are wanting to implement. It sounds simple... input a speed signal and check the position of the valves at various speed points. But, as with a lot of other things, it's easier said than done.

Thanks again,
RichT88

A procedure I have used in the past which relies on a calibrated speed setpoint - With machine off-line, run the speed setpoint through its range and note the no-load speed. With machine on-line, note the speed setpoint values as the load is increased from 0% to 100%. At 100%, note the speed setpoint and find the corresponding no-load speed.

Bruce