we have ge frame v turbine. mark v control system.

two months ago we faced problems with intervalve pressure fluctuations. we found the problem related to a leaky power cylinder of gcv. also in the process since srv did not open during start up, we had to replace a moog valve for srv. the turbine was running well since then with stable p2 press and load. even the mvar fluctuations subsided.

yesterday we had to shutdown the turbine for air inlet filter replacement. after the stop command was given the turbine unloaded but it tripped later with the alarm 'combustor flamed out during shutdowm' and 'gas ratio valve position servo trouble' at some value between 3000-3500 rpm.

the filters were replaced. and the turbine was restarted. following is the SOE. request to kindly help us broadly understand where the problem could be.

first start command :
the turbine came to firing speed. ignitors were powered but flame didnt appear. intervalve pressure was nearly 0. we checked flame detectors. it was ok.

second start command :
the turbine again reached firing speed. we measured 110vac at inputs of ignition transformers when ignitors were energized. but the flame again didnt appear. so we decided to watch srv/gcv and 20fg.

third start command :
the turbine reached firing speed. the gcv opened. srv did not. we changed the servo moog valve (despite the fact that the same was changed just two months ago).

fourth start command :
the turbine reached 14HM and we were waiting for 5 minute purge timer to time out and firing command be sent. but just two minutes after 14HM signal the turbine tripped on alarm 'gas control valve servo trouble'. we could not verify whether the newly replaced moog worked or not. so a next start was given.

fifth start command :
the turbine reached firing speed. srv opened. srv and/or gcv was hunting. p2 was very unstable. flame was very unstable. the turbine went into warmup mode. but soon both the flames were out. and the turbine again tripped on 'flame failure'. but we were sure that srv was opening. the newly replaced moog worked.

sixth start command :
the behaviour was same as fifth start command. we stopped for the day.

the next morning we found that during the night two alarms appeared. 'start-up shutdown <C> comm failure - trip' & 'common IO communication loss'. the status was '0' for both the alarms. we decided to change a pall's hydraulic filter in the morning and take a trial start. before the turbine cud reach the firing speed, it was seen on the main display that 'fsg' momentarily fluctuated between 25% and 0% thrice. 'fsr' was fixed at 0%. soon the 'gas control valve servo trouble' alarm appeared and the turbine began to coast down.

we tightened lvdt terminals. checked excitation values. it was between 6.5 to 7.0vac for all lvdts. the feedbacks were 0.245 vac and 0.256vac for gcv's lvdts at the zero position.

we wanted to replace lvdts but did not know the calibration procedure of mark v so could not do it. we decided to take another trial. this time the turbine started without any problems. reached fsnl and synchronised.

given the above sequence of events what are the areas we should check? is the problem related to field or mark v control system? is the failure of moog valve in a short span of time a coincidence or there could be some serious grounding problems? pls help

regards

 

You've quite a problem. Made even more problematic by the fact that the unit started and synchronized after you considered recalibrating the LVDT feedback. (Threats always work with the Mark V!)

You haven't told us whether or not you have a TMR Mark V or a SIMPLEX Mark V, though it probably doesn't make a lot of difference (but it might).

It is presumed that high-pressure Hydraulic Oil is at normal pressure for your unit.

In general, the SRV will not open if 20FG-1 and the dump valve it operates, VH5-1, don't work properly. Are you monitoring Gas Fuel Trip Oil pressure during firing? It should be greater than approximately 20 psig (which is the usual pressure below which VH5-1 will not permit high-pressure Hydraulic Oil to pass to theSRV servo and SRV actuator). You need to review the Trip Oil System Schematic Piping Diagram and then check each device, along with the operation of VH5-1, to make sure that high-pressure Hydraulic Oil can get to the SRV servo and the SRV actuator. If the SRV won't open, then having spark won't make any difference because there isn't going to be any fuel to ignite.

You didn't mention verifying servo-valve current polarity when you replaced the servos. This is *critical* whenever replacing servos as many Moog servos, while they all have the same colored leads, are *not* all internally wired identically and so verifying the operation of each coil independently of the others is critical (even for SIMPLEX panels, which should be using two-coil servos). You can perform a check while the unit is running by checking the servo-valve current output of each processor. The easiest way to do this is by looking at the servo-currents using AutoCalibrate. (*DON'T WORRY!!!* Just using AutoCalib to look at running values of servo current and LVDT feedback won't affect turbine operation!)

The LVDT excitation voltage should be approximately 7.0 VAC RMS. It's not clear where you measured the feedback voltage, but in general, anything less than approximately 0.600 VAC RMS would be VERY suspect. You mentioned 0.245 VAC, but you didn't say if the measurement was taken with a True AC RMS voltmeter, which is what must be used when checking LVDT feedback voltages. If the readings were taken with a non-RMS voltmeter, they mean nothing. If they were taken with a True AC RMS voltmeter, then they are very suspect. Many applications have zener barriers (IS, or Intrinsically Safe, barriers) installed on many of the analog I/O connected to the panel. So, where you measured the voltage is also important; in other words, between the barrier and the LVDT or between the barrier and the Mark V terminal board? (Zener barriers can reduce the LVDT feedback voltage inputs to the Mark V, so it's important to know if they are used at your site, and if so, where you are measuring the LVDT voltages, as well as the voltmeter being used to measure the voltages must be the correct type.)

As soon as I finish writing this, I'm going to look into purchasing Moog, not just a few hundred thousand shares. I'm going to purchase the company outright based on what I've observed over many years and what I've read on this forum. People so readily change servos any time they perceive a problem, and these things aren't cheap! The main reason servos fail is poor L.O. quality (lubricating oil is used as high-pressure Hydraulic Oil on GE-design heavy duty gas turbines). Either the only thing done to "maintain" L.O. purity is filters are changed (and only when high differential pressures are annunciated), or the L.O. is replaced on some subjective basis (like every 20 years, whether it's necessary or not!). There are countless trade journals (magazines) devoted solely to L.O. purity and maintenance! L.O. analysis is a course at some universities! Changing filters is *NOT* the entire extent of proper L.O. maintenance. It's not even the minimum requirement for proper L.O. maintenance. What does your site do for L.O. maintenance? Does your site ever send samples to a lab for analysis? (I worked at a site that was obsessive about sending samples for analysis; however, they never spent a dime to implement any of the remediation proposals from the lab! I think they just figured paying the annual fee for the quarterly analysis "protected" their L.O.) Has your site ever used some kind of L.O. centrifuge or "polisher" to improve the L.O. cleanliness?

Are you experiencing any Diagnostic Alarms? What Diagnostic Alarms are always annunciated? (Come on; be truthful about this.)

Has your site implemented TIL 1480-2, which is supposed to take care of the 'Comm I/O Loss' problem (but doesn't seem to work every time)?

I think you need to make sure 20FG-1 works and that VH5-1 works and that any other devices in the Trip Oil system work properly. If 20FG-1 is working properly, when it's energized there should be NO trickle of oil coming from it's drain (in other words, it's a normally-open valve that is energized to close). If 20FG-1 works properly, then you need to make sure that VH5-1 works properly. If neither of these have been serviced or replaced for a while, it's recommended to do so. Neither are complicated devices and both can get "gummed up" with poor quality L.O.

You also need to check, and correct if necessary, the servo-valve current polarity (there are several threads on this site about how to do that; use the 'Search' feature, it's very good!).

It would also be helpful to know a little about your site. How old is the turbine? Was the Mark V provided when the unit was new, or was it a retrofit application of an earlier Speedtronic turbine control system? What are the L.O. maintenance procedures at your site?

Please write back and let us know what you find.

 

Is there any news on this case?

 

Our problem isnt solved, if I may say, completely. I checked with maintenance persons and they say that they do use a centrifuge for oil. But the turbine in question is due to its MI for last 26000 hours (Instead of MI at 48000 hours it has run for 64000 hours already without MI). So naturally, the Oil quality isnt checked as you suggested. C&I has spoken to the OES and the OES doesn't seem to believe that it's a problem with any hydraulic circuit. (The OES never visited the site for any onsite observations).

No, we dont use any IS barriers. We use True RMS 4 and 1/2 digit Fluke Multimeter for all our measurements.

The LVDT voltages are indeed 0.245 - 0.250VAC and this is the situation since last eight years. we never stroked the gcv and srv after the unit was commissioned. These values are, I believe, since the time of commissioning.

We have an OES engineer at site for another job and 'fortunately' the problem of MW fluctuations has re-occured in his presence. We spoke to the OES and the OES still is of the view that the Moog valve may have to be changed again. (This will be the third Moog Valve since Dec 2007 for srv). We told this to OES and asked if it is normal for Moog valves to malfunction so frequently given that our other units have never needed a moog replacement in last 25 years. (The other units have two coil servos). The OES still insists that replacing the moog servo valve again would 'probably' solve the problem. OES is of the view that in Liquid fuel turbines its very common for Moog to malfunction but sometimes in Gas Fuel turbines too, it can be a case. However, with no offence to the OES, personally I remain unconvinced in light of the moogs of the other turbines. The OES' engineer is also not a GE 'certified' engineer. So that makes the diagnosis all the more less reliable. The MI is planned to take place in the month of November or early January 2009. C&I will have to insist on asking for a GE certified engineer to look at the problem.

Tommorrow, we intend to replace the Pall's Hydraulic Oil Filter just at the entry of the Servo valve blocks. Then run the unit and see if the MW fluctuations have eased. The idea is to make sure that its not the servos that are troublesome but the quality of oil, which gets ok for sometime when a new filter is put in place. If this filter is clogged then it will affect gcv and that would in turn affect the srv too, ultimately resulting in a visible P2 fluctuation, as we now see. If the problem doesnt get resolved by changing the filter alone then we may have to make do with another moog, as the OES suggested, to carry on till the MI.

Ours is a TMR Mark V simple cycle frame 5 unit. yes we do have diagnostic alarms but none other than those that were always present since the commissioning time. Our turbine was new when it was commissioned. the whole package turbine and hrsg were commissioned by one packager.

Meanwhile, I thank you for answering all of ours quesions so patiently and elaborately. Hope I have given some more details in addition to what I had already written earlier to assist you in exploring more possibilities for the MW fluctuations.

thanks n regards

 

Back to basics, If you measure (confirmed RMS voltage) the LVDT feedback around 0.256V in closed position of the valves, MK5 will not open more then 50% of the demand value at full load (the LVDT voltage will be appr.1.1 to 1.2V)! At full load SRV should be open appr. 70% and the GCV 45..50%.

If this is the case, recalibrate the LVDT to 0.7V (at closed pos. of the valves)and your problem will disappear.

Good Luck.

 

Actually, the Mark V can use *any* LVDT feedback voltage *as long as* the feedback is linear over the range of travel.

For years the spec that GE has used to purchase LVDTs says the output from the LVDT should be linear from approximately 0.58 VAC RMS to approximately 3.75 VAC RMS over the expected range of travel of the LVDT when the excitation was approximately 7.00 VAC RMS. *THEN* GE chose LVDTs that, when the zero stroke voltage was set to approximately 0.70 VAC RMS, +/- 0.02 VAC RMS, the range of travel of the device to which they were attached would not cause the output to exceed approximately 3.50 VAC RMS. This meant that the output would certainly be linear between 0.70 VAC RMS and 3.50 VAC RMS *when connected to the device*.

That's difficult to write, but if the LVDT output is linear from 0.58 VAC RMS to 3.70 VAC RMS over 2.00 inches, for example, and it's attached to a device that only has a range of travel of 1.50 inches and the zero stroke voltage is set to approximately 0.70 VAC RMS, then at 1.50 inches the output would be less than 3.70 VAC RMS (proportionally). By setting the zero stroke voltage to 0.70 VAC RMS, it is *certain* that the LVDT output will be linear at zero stroke, and by putting the LVDT on a device that would make the output go no higher than approximately 3.5 VAC RMS, the output would be linear over the range of travel of the device. (Most LVDTs have non-linear outputs at very low stroke and also at very high stroke, but output is linear over a large range of stroke between the "bottom" and the "top" of the stroke. So, it's critical to adjust the LVDT such that at zero stroke of the device to which it is attached the output is linear, and remains linear over the expected range of travel of the device to which the LVDT is attached.)

By the way, contrary to popular belief, the output of an LVDT will *NOT* always be 3.50 VAC RMS when the device to which it is connected is at 100% stroke and the zero stroke voltage is 0.70 VAC RMS. Sometimes it might even be slightly higher than 3.50 VAC RMS, but never greater than approximately 3.75 VAC RMS, at which point the output would probably start to be non-linear. If the output is only 3.14 VAC RMS, or 2.89 VAC RMS, then that's what it is! 0.70 VAC RMS and 3.50 VAC RMS are just the maximum values, not the values which will magically appear. Every device has a slightly different stroke, and that will be reflected in the LVDT feedback, hence the need to calibrate the feedback!

Some SRVs purchased from a couple of vendors were rotary cam vee-ball valves and were provided with "non-standard" RVDTs (Rotary Variable Differential Transformers), which had a very different linear output range than the LVDTs typically provided the GE-design heavy duty gas turbines. These RVDTs, and even an LVDT or two, had very low output and very low resolution. Sometimes the LVDT maximum linear output was approximately 1.50 VAC RMS, and the zero stroke had to be set at approximately 0.035 VAC RMS or less. It was pretty critical to check the output and ensure it was linear over the 90 degrees of travel of the SRV. But, with a range of 0.035 VAC RMS to 1.50 VAC RMS, the resolution was very poor, and it takes a larger change in position to cause a small change in LVDT feedback than if the device had an output span of 2.8 VAC RMS instead of 1.15 VAC RMS, which caused some instability. If the gas fuel supply pressure was not extremely stable, the instability of the SRV could even be higher.

Also, the actuators of the rotary SRVs never seemed to be properly "understood", meaning that the flow-rate through the servo required for actuation of the device was never properly analyzed in order to determine the proper regulator gain to put in the I/O Configurator. And people would not always enter the proper I/O Configurator gains for the devices, even if they weren't rotary devices.

And, any gas fuel supply pressure instability along with the poor resolution from the RVDT contributed to lots of problems.

So, to Ali Oztas' point, if a typical LVDT (one with an output which is supposed to be linear over 0.58 VAC RMS to 3.75 VAC RMS) is adjusted such that it's zero stroke voltage is approximately 0.24 VAC RMS, the Mark V probably won't be able to drive the device full open when, or the feedback may not be exactly proportional to position.

I'm *NOT* saying that's what's happening here, but what I *AM* saying is:

1) The zero stroke voltage, while commonly 0.700 VAC RMS, +/-0.02 VAC RMS, is dependent on the position feedback device (LVDT or RVDT), *NOT* the Mark V.

2) The Mark V can calibrate the feedback at any voltage *as long as it is linear*.

3) The resolution of the feedback can be problematic.

4) The regulator gain may not be set correctly for the device provided.

So, what kind of position feedback device is installed on the SRV and GCV on your site?

What kind of SRV and GCV are on the unit at your site?

You've told us what the zero stroke voltage is; what is the full stroke voltage (when the devices are full open)? (To do this, you need to manually stroke the devices; *BE SURE* to isolate the gas fuel supply before opening the SRV!!!!) While checking the output of the position feedback devices on the SRV and GCV, please try to plot the reference and the feedback; you're looking to make sure the feedback is linear over the entire range of travel.

What are the gains in the I/O Configurator for the SRV and GCV? (You can find this info on the respective SVO Output screens of the TCQA Card in the I/O Configurator; the SRV is usually SVO1 and the GCV is usually SVO2.)

How stable is the gas fuel supply pressure?

Oil quality is still pretty important. It would be VERY interesting to know what the condition of the current oil is versus it's condition when new. It has been reported before on control.com: L.O. manufacturers have changed the way they make lube oils and the effect has been to cause more problems for servo-operated devices. GE has actually done some research and is now offering a "lube oil polisher" to help with the problem, though it's not clear how the polisher will affect the oil longevity.

P.S. I'm in negotiations to purchase Moog, and when I complete the transaction, I won't be trying to help anyone understand servo-valve outputs! I just need about $300 Million USD more and I will complete the transaction, but the bank has stopped talking to me. However, I don't give up easily.