We have GE Make 6FA Gas Turbine equipped with DLN 2.0 combustion technology which is being controlled by Mark V Control System. Recently we had finished our Combustion Inspection (CI), as a part of CI we had replaced the hot gas path parts like

Flow sleeves, Combustion Liners, Transition pieces,Fuel nozzles, etc.,with reconditioned pieces. DLN Tuning was performed by GE and following constants were changed,
FXKPTSP_1 85 --> 85.5 %splt (Premix Steady State Split %Split IGV Temp control on Array)
FXKPTSP_2 89.5 --> 89 %splt (Premix Steady State Split %Split IGV Temp control on Array)
FXKPTF_4 2350 --> 2330.6 deg F (Premix Steady State Split TF IGV Temp control on Array).

We had changed all our Moog valves (SRV,IGV, GCV's) with same part no (there was no problem in existing moog valves-it was working good but eventhough we had changed it) & its filters. Before replacing the servo we checked the resistance of new SRV and Splitter and it reads 1.084, 1.085, 1.091 KOhms for SRV and 1.05, 1.05, 1.06 KOhms for Splitter.After the replacement of new servos, stroke check were performed and found feedback is under the limit. No cards were replaced.

Due to GT Inlet gas pressure problem, we had requested GE for Auto unload sequence and implemented the same. Compressor inlet TC's were changed due to false reading and new TC's are working good.

The problem is that we had started the unit and reached the baseload (66MW) and found the MWATT was hunting between 2 to 3 Mwatts. All control valves like SRV (54.00 to 56%),Secondary (62 to 65.5%), Splitter (82 to 83%), Quat (25 to 27%) are fluctuating , due to that the parameters like FSRN, FSRT, TTXM, TTRF1 (+/-8deg C) & all Spreads are fluctuating (+/-6deg C).

CPR, CTD, AFQ are reading good and no fluctuation. IGV LVDT calibrated due to high difference between command and field position (measured locally) and no fluctuation. Checked the Prevoted values for MWATT Transducer and found all are reading almost the same values.

We suspected the DLN Tuning constants would create this problem and we had put back the old values and found no change in fluctuation (not helpful, keeps on fluctuation). Physically checked for Hydraulic Oil leakage in Gas compartment and found no leak.

We had taken the unit to preselect load just to check whether it is fluctuating or not, but it was fluctuating in that mode also.

Again taken the unit to Baseload and selected FSR MANUAL Control to determine whether the load hunting is caused by SRV or GCV's, in that we found Quat and Secondary valve stops fluctuating, but Premix splitter (FSGX) and SRV continues the fluctuation. (primary valve was in purging)

This issue had been taken to GE and they were suspecting the Servo of SRV and suggested to change the gain of SRV which they are thinking it too high (10.1), but the same gain was there in earlier Servo valve and we did not face any issues with that.

Can anyone help me to solve this issue?

1. What is making the MWATT to hunt?
2. What we need to do in offline to solve this issue?
3. Is there any temporary solution (like changing the constants in online, etc.,) to avoid this hunting until we plan for shutdown?
4. If we operate with this load swing what are all the parts may affect, is it advisable to operate with this load swing?

I am happy to provide you more inputs if you require.

Thanks in advance,
A2A.

 

To all reading this post, <b><i>THIS</i> is how to present a problem to control.com!!!</b> Tell us when the problem started, what's been done to troubleshoot the problem, and what the results of the troubleshooting are!

A2A, you didn't mention checking servo current polarity--which should be the FIRST thing done ANY time a servo-valve is replaced. Even new, out-of-the-box servos from Moog can have incorrectly color-coded coil wiring!

My money is on the SRV gain. It may have been high with the previous servo because of some problem (real- or perceived problem). It's certainly worth a try to reduce the gain and check operation--AFTER verifying servo current polarity.

My other suspicion is the new sequencing. I'm always suspect of new sequencing, because lately GE has been unnecessarily complicating simple control schemes, just because they can. The sequencing may be very simple, but it may also be very complicated if it's looking at supply pressure and using that as some kind of secondary reference for P2 pressure. It may be as simple as a slow gas fuel supply pressure transducer or the gas fuel supply pressure transducer needing to have some damping added (presuming it can be adjusted) or something similar. Again, without being able to see the sequencing running in the panel it's very difficult to say.

Are they any gas fuel valve-related Diagnostic Alarms? That may be helpful to look and resolve if any do exist.

Load fluctuates when fuel fluctuates; it's just that simple. If P2 pressure isn't constant it can play havoc with load. There may also be some gas fuel supply line dynamics/interaction with the SRV fluctuations ("harmonics") that are making a small problem worse.

But, based on the information provided (which was VERY good, by the way--Thank You!!!) if you haven't verified servo current polarity (to all the servos which were replaced), that's the best place to start (must be done off-line). Then do check the SRV gain; it does seem high. Then look at the new sequencing and any analog inputs to the new sequencing especially, if it is used as some kind modifier or secondary reference to the P2 pressure reference.

Operating with this load swing isn't really going to hurt too much if it's left for a week or two, but it probably isn't going to be good for the turbine hot gas path parts if the unit is operated at Base Load (fuel will be cycling). Also, fuel valve actuator oscillations can increase wear rates on cylinders and seals, not to mention the effects of cycling on the servo spool pieces. It might get worse, also; and it might even get better, but probably not. Best to do the troubleshooting at the earliest possible opportunity.

And, be sure to write back to keep us posted on the situation and its resolution!

(Changing servo gains on-line can be very risky, even scary, and I almost never recommend doing so unless there's just no other choice. If you decide to do that, I recommend doing so at low load--as low as possible, maybe even at FSNL.?.?.? And making small changes if you do so while the unit is running.)

 

Hello CSA,
Thanks for your quick reply.

1. We have verified the new gas pressure software and there is no link with P2 Pressure or reference. This software will start unload if gas pressure goes less than the limit, in our case the upstream pressure is healthy and well above the limit.

2. We have not performed the polarity check but we are very much sure about the coil wiring we connected, (we checked it twice) but as you are saying that the new Moog may have incorrect color coded coil wiring, so we will check it offline.

I have the screenshots of stroke check for our old servo and new servo, can we get any clue by comparing those screens for polarity?

3. There are no diagnostic alarms.

4. I am really very sorry, i had mistakenly typed the gain of SRV as 10.1 (which is gain for IGV) SRV has 4.5 only.

After posting here, just checked the GCV's feedback and found the Splitter (FSGX) valve is not fluctuating, remains constant in 82.3%. (compared the earlier trend and yesterdays trend and found almost stable on 82.3 to 82.2%), i am not sure how it could happened, but rest of the valves are fluctuating like before.(no changes were performed in Mark V or rest of the systems).

During offline if i find the polarity is good, then what else i can do/check for?

Thanks in advance,
A2A.

 

A2A,
Thank you for the good collection of information as CSA stated relying on the color of the servo valve wiring to be correct is not a good practice. If one or two of the tree coils are wired improperly then you can have opposing drive which will cause a servo valves pressure outputs to be unstable creating behavior similar to what you are commenting about.

The gain value of 4.5 looks normal as compared to the other sites of similar equipment and vintage that you are using. This should not be the issue. Another characteristic of a servo valve is the null bias setting which if not adjusted EVERY TIME you change a servo can also cause behavior similar to this in a control loop situation. The null bias should be set by positioning the valve in a mid stroke position between 10-90% open and measureing the voltage across all three coils and then summing them and dividing by 30 then multiplying the result by 100 and this will give you your value that should be input into your null bias setting for the valve assembly in the TCQA card I/O regulator definition. Please confirm that this is set properly and what the current value is if you could for us all.

Other than this there are only a handful of things that can really affect fuel valve position which then affects turbine performance and that is fuel supply pressure changes, air inlet temp changes, load demand changes, etc. So verifying the health of the control loop itself is a critical part of the troubleshooting process of an issue like this.

Thank you again for the post which I feel many people will learn alot from including myself.

 

A2A,

Usually one can see from the calibration screens if there is one servo with reverse polarity; it's pretty obvious because it will cause one or both of the other servo currents to be well off normal and usually causes a larger error between reference and actual (feedback).

You said you measured coil resistance; in my experience coil resistance isn't really an indicator of problems unless the coil is open. I can't recall ever encountering a coil with low resistance (say, less than about 900 ohms).

Some units around the time yours was produced used a rotary cam vee-ball type of SRV with an RVDT (Rotary Variable Differential Transformer), with two RVDTs in a single housing. These RVDTs usually would not work properly with a zero-stroke voltage setting 0f 0.700 VAC RMS because they would over-range and the output would be non-linear before reaching 90 degrees (100%). It was found that the zero stroke voltage for those RVDTs was to be set at approximately 0.50-0.055 VAC RMS, resulting in the 100%-stroke voltage of approximately 1.2 VAC RMS, but still linear with respect to stroke. The range (0.5-1.2 VAC RMS) is pretty small, meaning the resolution is not very good, and this resulted in some instability of the SRV on many machines. And, the hidden, built-in dither of the Mark V servo-valve outputs also exacerbated the problem.

If you have a right-angle Woodward SRV, with LVDTs, they have been proven to be a little more stable.

If the load was stable before the outage but is unstable now, you have to look at what changed. Are you certain the new servos installed are exact replacements for those removed? Flow-rate differences and internal passage sizing (usually reflected in part number differences) can cause problems. Many servos look remarkably alike, but have very different operation characteristics.

Were the gas valves refurbished during the maintenance outage? I've seen incorrect plugs and seats installed in valve bodies cause similar problems.

I've seen P2 pressure transducers replaced with SMART transmitters cause similar problems. SMART transmitters cannot be used for P2 or CPD sensing; they do not have a sufficiently fast response time for these applications.

I don't have access to a Mark V I/O Configurator printout, but I believe there is a second screen for the SRV with an integrator gain value. Verify all the settings on that second display with GE.

If the gas valves were not refurbished then there's a very slight possibility that the actuators are developing problems, but that's a slight risk which might be accentuated by the installation of a new servo. Not likely, but possible.

I can't think of too much else to try. Again, if it was working before, then something has changed. It might be worth checking the servo wiring loops by disconnecting them from the Mark V and checking for continuity and grounds all the way to and from the servo coils.

If the site uses intrinsically safe barriers (diodes), I've seen them cause problems--lots of problems. Intermittent problems, as well, which were very difficult to pinpoint. While they may test okay, when current is actually flowing they can pose intermittent problems.

Lastly, I've seen loose wiring terminations cause problems. Many times people have tried to crimp terminals on servo coil leads, and it just doesn't work very well unless one strips about 25-40 mm of insulation off the wires and then folds the wire a couple of times before inserting into the terminal and crimping. This has been a problem many times when servos are replaced. Sometimes, a very slight tug on the servo wires pulls the wire out of the crimp terminal. It's very difficult to crimp a terminal on a single servo wire and make a proper crimp with sufficient surface contact.

That's about all I can think of. I don't have access to any 6FA servo gain settings, but that may be an issue. But, it wouldn't make sense that it would be necessary to change the gain--unless it was changed because of a problem with the previous servo (again, a real or perceived problem).

It should be possible to reduce the load oscillations, especially if it wasn't hunting before the outage work was done. When troubleshooting a problem like this, remember to list all the possible causes (even loose terminals and wire crimps!) and methodically eliminate every problem one by one. You seem to be doing that now; don't stop.

Again, keep us informed, please!

Oh, one last thing which can only be done AFTER an AutoCal, is to use the 'Verify Current' and 'Verify Position' features of AutoCalibrate to check LVDT/RVDT linearity and servo current/coil operation. These can be useful tools, and they can also be part of the elimination process of troubleshooting.

 

Ohhh, this should/could be fun!

I don't believe an incorrect null bias would result in load instability.

Do you, HFGTF&AGUY, want to provide a step-by-step procedure for adjusting null bias current, please? Tell us how to begin the procedure (what should the null bias value be at the start of the procedure), how to check to see if the null bias needs adjusting, how to adjust the null bias, and how to determine if the adjustment produced the desired result, and how to make any necessary changes to produce the desired result.

The emphasis is on the "desired result". What should be the desired result of changing the null bias value? In other words, when do we know that a proper and sufficient change has been made to the null bias current?

What does the null bias value do, and when does it need to be changed? We need to know when the null bias is set to the proper value and when to change it, as well as how to change it. So, tell us how to tell when it's been properly changed, not just that it needs to be changed and how to change it.

I will preface any further remarks I have by saying, first, I take exception to the statement that it's ALWAYS necessary to adjust the null bias current value EVERY time a servo valve is change/replaced. Second, I have NEVER had success with trying to calculate null bias values from measuring voltages and/or resistances. NEVER.

But, I'm always wanting to learn from others, so, let's get my education started! I'm sure many people can benefit from this discussion, and we could even end up with a control.com wiki article out of this!

Let's roll! The ball is in your court, HFGTF&AGUY. You started this; let's keep it going.

 

A2A... the following Control.Com thread might be helpful:

http://control.com/thread1026226787

Regards, Phil Corso

 

Phil Corso,

The thread (the link is actually: http://control.com/thread/1026226787) was solved when the damaged liquid fuel pump clutch was replaced.

This thread is for a Frame 6FA burning gas fuel, so there would be no liquid fuel pump requiring a clutch.

The referenced thread does bring up a problem I hadn't considered: Hydraulic accumulator problems. However, if there were problems with the hydraulic accumulator it would likely be affecting all hydraulically-actuated devices, and the original poster has stated the IGVs are stable and one of the gas fuel valves is also stable.

It's still worth a look at the charge on the hydraulic accumulator(s), and reviewing the hydraulic system (air bleed check valves; relief valves; etc.). From the information provided, specifically the results of gagging the FSR, it would seem that the problem is most likely something to do with the SRV (Stop-Ratio Valve). Gagging the FSR until the MW fluctuations stopped and still seeing SRV fluctuations sure seems to point at the SRV, it's actuator, the position feedback, or something like that.

 

CSA... it is clear to me that not all posters have the acumen or experience you have.

My intent was to illustrate an approach for solving "Chicken or the Egg" problems. That is, compare associated variables on the same time scale! For example, fuel-flow vs MW, etc.

Regards, Phil

 

I agree that a lot of data was given to assist in troubleshooting. But since it is a DLN machine, some data in regards to CO and NOx from the continuous emission monitor system might reveal a problem with the tuning on the splits. Attempting to lower NOx too much can cause combustion problems that cause CO to go up.

 

Belief is a wonderful thing and as diverse as the people of this earth! I respect everyone's beliefs but this forum in my opinion is for facts and here they are.

In response to the procedure for adjusting the NULL BIAS servo card setting GE TIL 1813 can be referenced by anyone (and should be as it is a great read that illustrates issues with any device using a servo output card) who has access to the GE TIL system.

My reason for bringing up the topic of NULL BIAS is that it is a setting that should be accurately configured otherwise why would it be a part of a OEM GE procedure for installing and calibrating a device such as this on the gas turbine? It is a way to clear the servo valve from the cause/contributor list and move on. I am not saying it is the root cause or that is a co-contributor to the symptom only that it is good practice to verify all parameters of operation for the device in question are correctly set. As stated by another poster below I am simply assisting with some strategy.

The NULL BIAS is a characteristic of the servo valve that is factory set to a specific value and should not be adjusted in the field. The offset is there to allow for a fail safe response should a power loss situation occur within the servo valve command loop.

In MOOG servo valves, installed on GE gas turbine auxilliary equipment, the NULL BIAS is set by offsetting the servo valve spool bushing from center which in turn causes leakage from one output port of the servo valve to the drain thus providing in IGVs & GCVs a fail closed response. The state in which you will find the current control settings varies as the NULL BIAS characteristic is specific and different from servo valve to servo valve.

The default setting for the turbine regulator should be 2.67VAC. The way that an incorrectly set null bias will cause a device to behave is by allowing for an offset between commanded position and actual position. This will be constant throughout the range of operation. This is particularly menicing to DLN machines specifically DLN 2.0 and higher machines where the tolerances on temperature and thus fuel control are much tighter than conventionally fired machines. So this in turn could cause issues with combustion, fuel flow, and could cause fluxuations in P2 pressure as well if the affected servo valve is installed on the SRV. That is why the NULL BIAS should be accurate dealt with per the OEM instructions.

As stated before in the previous post you simply position the actuator mid stroke anywhere between 10-90% open and measure the current delivered to each coil via the correct location on the terminal board and sum the values and divide by 3 and then multiply by 10 to achieve the servo valve specific null setting characteristic which should be 10% of the rated current +/- 1% of the rated current, so 2.4VDC to 2.93VDC REGNULLBIAS value. The dsired result should be an accurate following of the position command reference with the valve actual position feedback within 1% of the setpoint. This can be checked with an interval plotting of 10% increments from 10 to 100 percent stroke using a dial gage to verify actual mechanical stroke at each interval. For more in depth information please see TIL 1813 or contact MOOG.

If there is someone that cannot gain access to the TIL please reply to this post and I can supply you with instructions directly for setting the NULL BIAS.

Additionally so all are aware you can install a feature for your Mark system that can calculate/monitor the null bias automatically and alarm you if the setting drifts into unacceptable ranges (Very Handy).

 

Hello,

Do you have the MkVI feature for alarmimg in case null bias drift?
Could you send it to [email protected]?

Thanks.
Regards.

 

Dear All,

Thanks for your reply, here is my inputs

The current value for nullbias is 2.9 and there is no fluctuation in air inlet temperatures, Fuel gas pressure and frequency

(all are pretty stable).
I have compared the stroke check screens of old servo and new servo and found the servo currents are more or less similar.
Earlier (before outage), the load was stable and normally vary in the range of decimals only and gas fuel inlet temperature

(to GT) will have +/- 1deg C fluctuation, now it is fluctuating by +/-1.5 deg C (but the gas fuel discharge from boosting

compressor is stable) and we did not do any work on FTG's.

We are having Young & Franklin make SRV, no maintenance work was carried out in SRV/GCV's during our shutdown. All our

Pressure transmitters and switches were calibrated and no adjustments were done.

Before shutdown at 62.6 MW, the NOx was 11.11, CO-3.0, O2-9.9%, now at 62.5 MW NOx is 12.5, CO-3.0, O2-9.7%.

We have planned just to replace the new servo with old servo during our next opportunity.Even after replacing the servo we cannot come to know the fluctuation at offline (the unit must be online to see the fluctuation, this is the problem with this kind of issue).If the old servo solves the fluctuation issue, happy, but the load swings again, then we have to wait for another opportunity, but we dont want to take this issue on our every opportunity.Hope changing the servo will solve the issue.

Please let me know if anyone needs more input.

Thanks in advance,
A2A.

 

A2A... can you at least provide answers to the following questions:

1) the rate-of-change?

2) does it occur randomly?

3) does it appear to be sinusoidal?

4) does it occur during a specific time interval. That is, during a shift, a day, a week, weekends, etc!

Regards, Phil

 

A2A... Have you looked into the phenomenon, "Which Came First, The Chicken or The Egg?".

Have you seen any corroboration between MW deviation, and local load change?

Search Control.Com Archive for a thread involving a vibration problem that took weeks to solve. It is:

http://www.control.com/thread/1319907715

Phil

 

A2A,

Thanks for the update! "Feedback is the most important contribution!"(c) as it makes the threads valuable to others for a long time after a single problem is solved for the original poster.

I have seen cases where the o-rings of the servo-valve were not properly fitted and which caused hydraulic oil leakage leading to some small instability.

Unfortunately, the Mark V does not update the servo-current values to the CDB (Control Signal Database) faster than 4 Hz (if I recall correctly) so using one of the VIEW tools to capture data doesn't really have very good resolution. Even though the VIEW tools can run at a 32 H rate, because the Mark V operating system doesn't update the servo-current values faster than 4 Hz, the data values won't change at a 32 Hz rate if the servo current is fluctuating. And, the TCQA compares the feedback to the reference and adjusts the servo current output(s) at a 128 Hz rate. So, gathering data for servo current fluctuations with the Mark V is very difficult.

The Young & Franklin valves and -actuators seemed to be more robust than the Woodward valves and -actuators GE is supplying these days. That being said, all valves and actuators can fail.

Anyway, please keep us advised of the progress.

 

If you provide more details to my email, we may be able to help you.

Rgds
John

 

> 1) the rate-of-change?

>2) does it occur randomly?

>3) does it appear to be sinusoidal?

>4) does it occur during a specific time interval. That is, during a shift, a day, a week, weekends, etc!

Dear All,

Here is my inputs,

1. the rate-of-change? --> its hunting to the maximum +/-3.1 MW/s (most of the time hunting in the range of +/-2 MW/s)

2. does it occur randomly? --> Its continuous

3. does it appear to be sinusoidal? --> Yes,

4. does it occur during a specific time interval? --> No, swings all the time

I could not correlate with any parameter to find which comes first, but there is a fluctuation in Gas Fuel Temperature in the range of +/- 1 to 1.5deg C, but not sure whether this may contribute to huge load swing. I just started to check the Gas flow to GT, if i find anything i will keep posted.

Thanks,
A2A.

 

A2A... thank you for your response!

Q5) Regarding your comment about the change of temperature... although seemingly "rapid" does it also occur at the same rate as the deviation!

6) Is there a corresponding change in stator Amperes, PF or kVAr?

Regards, Phil

 

A2A... further to my earlier questions:

7) How about providing Generator parameters?

8) Is Generator operating in Island-mode?

9) Is Generator operating in parallel with another gen'r, or a utility?

10) If in parallel with another Generator, is there load sharing?

11) If in parallel with utility, what is relative size compared to grid capacity?

12) If in parallel with utility, what is droop-setting?

13) If in parallel with utility, via a GSU Xfmr, what is xfmr's size and impedance?

Phil