GT Liquid fuel hunting

W

Thread Starter

Wajid29

Hello Guys,

We are facing an issue during our GT startup on liquid fuel (Low BTU gas so can not start on gas fuel). Want to clear few things before my question. We have GE frame 9E gas turbine with MARK VI controller.

Gas turbine is fired successfully every time but as the rpm increases i.e >2600 rpm then huge fluctuation in liquid fuel flow is observed. however AUTO Synch command is initiated as soon as turbine reaches 3000 rpm regardless of fuel hunting. This hunting is observed reducing after 25 MW and almost zero after 40 MW.

Actions Taken:
1. Liquid fuel Bypass Servo valve was replaced two months before and only first startup after replacement was hunting free.

2. Liquid fuel system venting/priming.

We can not understand the issue that why it is starting hunting reaching a specific rpm and load range. If there is a problem in servo valve then it must fluctuate whenever it is commanded. It is here to be informed that no LVDTs are installed at liquid fuel bypass valve rather flow divider magnetic pickups are used as feedback of the valve.

Our machine is now synchronized and on Gas fuel but we do not want to face this issue in future. I have received positive and beneficial comments from this forum before and hoping for the same once again.

If you require further information to help us then please let us know. I will visit the forum continuously and if anyone want to give answer me on my email id so it is provided below.

[email protected] is my email id

Thanks
 
Wajid29,

What Process- and Diagnostic Alarms are being annunciated during starting? Please be as concise as possible with your list of alarms (both Process- and Diagnostic). Use Toolbox to check the Diagnostic Alarms of the cards which have the orange Status LED lit on the card edge.

When you replaced the servo-valve, were the polarities of the servo currents being applied to each of the servo-valve coils checked? This is EXTREMELY important, and commonly overlooked.

Can you tell us, when the liquid fuel flow-rate is stable how much current each of the processors is supplying to the LFBV (Liquid Fuel Bypass Valve) servo-valve? This may be a big help in determining what the problem is.

I looked at some of your past posts, and one very interesting one was regarding replacing the hydraulic fluid which resolved a sluggish ACV problem. Could this be the problem here? Servo-valves don't just fail--the usual culprit is poor oil quality. It's becoming a more well-known fact all the time that oil refiners have changed their turbine lube oil formulations in the last 10-20 years to improve the lubricity characteristics, but for those machines using lube oil as a hydraulic medium it has resulted in lots of problems because of the way the oil holds solids and varnishes.

Having said that, if the only hydraulically-operated device with a servo-valve that is experiencing the issue is the liquid fuel control valve, then it's probably best to focus efforts elsewhere.

Liquid fuel systems usually have a forwarding pump system with pressure regulators which must function properly to maintain the desired pressure--and to limit the pressure to the GT liquid fuel system. These pressure regulators frequently get air in the actuators which causes problems.

There are also usually strainers upstream of the forwarding pumps which can get clogged ("choked") and cause problems.

Since there are no LVDTs on the LFBV, the Liquid Fuel Flow Divider feedback is the main feedback for liquid fuel control. The reference, FSR1, is converted to a liquid fuel flow-rate, and the feedback is from the flow divider's speed pickups. It's not very likely the problem is the liquid fuel flow-rate reference--but it could be, since to an extent it's affected by turbine shaft speed. But you say the instability continues after the unit synchronizes--and the presumption here is that the grid frequency is stable so the turbine shaft speed is also stable (not always true in some parts of the world, so gotta bring this up).

Have you checked the tightness of and the gap of the Liquid Fuel Flow Divider speed pick-ups? Many Mark VI and Mark VIe units have extremely sensitive speed pick-up inputs these days, and the old specification of 0.010 inches for the liquid fuel flow divider is just too small. Gap settings of 0.030-0.040 inches seem to work a lot better.

Liquid Fuel Flow Dividers also get worn, and the bearings need replacing from time to time. How long has it been since the Liquid Fuel Flow Divider was refurbished? Has anyone looked at the toothed wheels of the Liquid Fuel Flow Divider to see if one or more of them is loose?

It could be that the Liquid Fuel Pump Clutch is slipping--not likely, but it could be a problem that should be eliminated. You can use a strobe light to "freeze" the coupling between the clutch output shaft and the High-pressure Liquid Fuel Pump input shaft (when the turbine shaft speed is stable) to see if the clutch is slipping or not.

It could be a problem with the hydraulic actuator of the LFBV allowing hydraulic fluid to leak by the piston seal ring and contributing to instability. Many times the cylinder walls of the actuators can get worn in certain areas and cause problems like this.

You have a pretty powerful troubleshooting tool in the Mark VI Trend Recorder. You can set up a high-speed (40 ms) trend to monitor liquid fuel flow Divider feedback from the three pick-ups (some newer machines seem to be using only two pick-ups; we don't know how new your machine is), the flow-rate reference, the speed, the actual flow-rate calculation, turbine shaft speed feedback, etc. Most sites don't have pressure transmitters on the liquid fuel supply system, but if yours does that would another useful thing to trend and analyze. I would also recommend trending TNHA and TNHAR (actual turbine acceleration, and turbine acceleration reference). You should also be trending the amount of current being applied by each processor to the LFBV servo-valve. Sometimes one has to create new pins for signal names for each controller's servo current output to be able trend them easier, but that's not too difficult, though it does require a download and re-boot.

Another thing which you could try is to use Manual FSR ("FSR GAG") to try to reduce the pressure fluctuation during loaded operation (when the turbine shaft speed should be stable and the required liquid fuel flow-rate should be stable. You can even start the machine with the FSR gagged at say, 25%, and then lower it slowly as the unit approaches FSNL to try to limit the speed fluctuations. You should be recording signal values using Trend Recorder to be able to analyze the results. This will tell you if it's the LFBV and/or it's actuator, and possible even if it's the servo. If, by limiting the travel of the LFBV you can reduce the speed fluctuations--presuming there are no Process- or Diagnostic Alarms which may be indicating other problems--this can help with your decision to try replacing the LFBV and/or its actuator (or the servo-valve, again--because if it's poor quality oil which is the culprit here, which doesn't seem likely if it is stable at low- and then again at higher flows, then replacing the servo might be a good idea).

Troubleshooting is many times a process of elimination. Sometimes problems are the result of two or more stimuli or reasons, so one has to logically eliminate things until the cause(s) are found and resolved. The liquid fuel system can be daunting, but study the P&IDs for the Liquid Fuel Forwarding System and the Liquid Fuel System and do your best the eliminate each and every possible cause in your quest to resolve the problem.

It's also a good idea to keep a written record of the tests and checks you make, and the results, so that you can go back and see what was done and what the results were--especially if you get to a point and you haven't solved the problem. By going back through what was done, you can see what may still need to be checked or some flaw in the testing or results which you may have missed.

Hope this helps! <b>PLEASE WRITE BACK</b> with answers to the questions above. And, if you can collect some Trend data, please post it to a file sharing/hosting site like tinypic.com or something similar and then post the link to the file(s) here so we can download and analyze them.
 
Wajid29,

Also, what is the Hydraulic System pressure doing during the instability? Is it stable, or fluctuating? Does it seem like the Hydraulic Accumulator is working correctly?

How long since the charge on the Hydraulic Accumulator was checked?

Are the valves on the Hydraulic Accumulator set correctly? (One is a block valve, to block the flow of Hydraulic fluid into the Accumulator; the other is a bleed valve, to bleed pressure off the accumulator when charging or testing.) The block valve should be OPEN, and the Bleed Valve should be CLOSED under normal operation. If you open the Bleed Valve when the Block Valve is open, the Hydraulic System pressure will start to drop. Unfortunately, the valves are not usually marked, though sometimes the valve handles are different sizes. (If I recall correctly, the large handled-valve is the Block Valve, and the small-handled valve is the Bleed Valve--but I may be wrong.) You can find the vendor's instructions in the Manuals provided with the unit, or look them up on line.

I would not recommend checking valve positions with the unit running....

<i>[Edited last line per CSA's reply below. - Moderator]</i>
 
Dear CSA,

Thank you for your very detailed reply. We are working on the points which you have described. Will write back here very shortly.
 
Wajid29,

Thank you for the progress update.

By the way, the last sentence of my second reply should have read:

"I would <b>NOT</b> recommend checking valve positions with the unit running...." If one opens the bleed valve too fast while the block valve is open, hydraulic pressure will drop quickly--with potentially tripping results. <i>[I changed it for you. - Moderator]</i>

By the way, it's difficult to use the Manual feature of AutoCalibrate to stroke the LFBV when it doesn't have LVDTs--difficult, but not impossible. By using a zero-crossing (no DC offset) frequency generator to simulate flow divider feedback, one can use the Manual feature of AutoCalibrate to get the LFBV to move. You only need a minimum of approximately 5 VAC--but you need to disconnect the flow divider speed pickups from the Speedtronic when you connect the frequency generator output IN PARALLEL to the Speedtronic flow divider inputs.

And, by carefully controlling the frequency generator output one can get the LFBV to open and hold a stable position. This could be useful if one can gently stroke the LFBV across it's entire range of operation to see if at some point the valve is unstable--which would indicate a problem with the actuator. But, if you have a spare hydraulic actuator for the LFBV, that might be quicker and easier in the long run. If the instability goes away when you replace the actuator (or the valve and actuator), then it's likely it was the actuator (or the valve, if you replace the entire assembly--you'll never know if it was the valve or the actuator, unless during refurbishment it becomes obvious what was failing).

Lastly, I did hear sometime back about a broken spring on an LFBV assembly on a Frame 9E. VERY odd and unusual occurrence, and it did have similar effects as I recall--instability at some regions of operation, yet stable operation at others. The LFBV spring can be visually inspected very easily.

If you can get some Trend Recorder data and post it to a free web-hosting site we can download it and analyze it. (GE has released a stand-alone version of the Trend Recorder which can analyze trends files without having the entire .m6b file.) I hope I included in my list the flow divider feedback signals--it's very important to the analysis. [No, I can't send anyone this application.]

Finally, has anyone visually observed the LFBV when this instability is occurring? How much is the LFBV moving--a lot, a little, fast, slow?

Also, what are the exhaust temperature spreads doing during this instability? Are they high, low?

What are the flame detector intensities during the instability?
 
D

deePATIDAR92

Dear CSA,

We are operating a Captive power plant with three 20 MW GE Frame 5 Gas turbine(GT1,GT2,GT3) running in droop mode. From last 15 days we are experiencing MW hunting of different magnitude(around 0.5 to 1.5 MW) at different load and at some random load (on decreasing or increasing load) hunting reduces automatically.

GT-2 has been running smoothly since March-2018 after its CIBI. GT-2 is presently running on Naphtha.

Following action was taken:
From now onwards, I will use GT instead of GT-2.

Naphtha forwarding pump changeover was done and forwarding skid control valve was checked. No hunting observed in control valve.

GT was taken to base load, and base load command was given. Subsequent to that Turbine status was coming as "unloading" whereas it should come as "base load". FSR and FSRT was matching and during base load temp FSR has taken over. GT liquid fuel pump discharge pressure was fluctuating form 41-46 Kg/cm2 almost on all nozzles.GT liquid fuel pump speed was measured by tachometer and was found okay and steady as well.

On base fuel pressure before stop valve was fluctuation was from 4.2-5.6 Kg/cm2 on the On Base Fuel oil pressure gauge. GT LFBV servo current was varying from -4.5% to +6.8% and sign of servo current was changing, indicating movement in LFBV valve. Base load trend was recorded.

VSVO Card diagnostic alarm LVDT #4 excitation voltage out of range - was persisting since start up however no LVDTs are installed in LFBV.
Flow divider magnetic pick up value were found same in R,S,T except Fql_pr1 voting mismatch diagnostic alarm came thrice at different time and got reset immediately.

On giving Preselect command; speed raise and lower command was getting actuated continuously.

Matter was discussed with OEM and it was suspected that LFBV servo has drifted over the time and may need proper setting / calibration now or replacement.

Null Bias was 1.5 and Regulator gain was 1.5. It was suggested to reduce the gain to 1.4 for 10-15 minutes and then 1.3 and TNH, DWATT, CSGV, CSRGV, CPD, CTD, TTXM, FQL1, FQL_PR1, FQL_PR2, FQROUT, FSR1, FSRN, FSRT, TTRX, TTRXS, TTRXP, FAL, FAL_NVR, FAL_NVS, FAL_NVT was recorded @40 msec.

It was found that even at lower gain fluctuation was on higher side. Hence machine was taken offline and servo coil and filter was replaced. Gain was again set at 1.5 and after checking polarity of servo coils and null bias setting(2.67), again machine was started.
At 1.5,1.7,1.3 regulator gain , hunting was still persisting.
As per OEM advise, null bias was set at 2.8 and trends were recorded at 1.3 , 1.2 gain and FSR gagging also.

Also trends were recorded at null bias equal to 2.75,2.7,2.6.
Still MW hunting was persisting.

It was then advised to replace LFBV assembly and verify the null-bias with new valve. After LFBV replacement and after checking servo response by OEM experts and the following settings were made:Null Bias - 2.6, Gain - 2.5.

After start up, now hunting of higher magnitude was observed at load above 10 MW . Below 10 MW very gradual hunting of around 0.6 MW was observed.

Then gain was reduced from 2.5 to 1.4 in step of 0.1,in all cases no final conclusions could be drawn, still hunting is being observed
The only thing now comes into mind is flow divider. It is observed that both Fql_pr1 and Fql_pr2 both increases or decreases simultaneously during load hunting, however the difference between them increases (more than 3 %) during MW hunting.

Can flow divider be the reason for MW hunting and if yes then how? Also why is the magnitude of hunting is different at different loads?

Please suggest next course of action.
 
D

deePATIDAR92

Sir,

It is to inform you that hydraulic oil pressure is stable during load hunting. Also no significant change in flame intensity or exhaust spread was observed during MW hunting.

After LFBV replacement, MW hunting did not start just after start up . It was also loaded up to base load without any MW hunting. MW hunting started after two days. You can check the trends below after LFBV assembly replacement at different gain and null bias:

At null bias 2.6 and gain 2.5, Trend data link for load hunting @14.0 mw

https://drive.google.com/open?id=1PMriBNXMlDgDtr0G1EEZNHkjQeoJWHZZ

At null bias 2.6 and gain 2.5, Trend data link @ 12 MW & 10 MW ( It is observed that hunting has reduced at 12 MW, But GT-2 load decreased to 11 MW and Increased to 13 MW without any change in total MW load of all three GTs )

https://drive.google.com/open?id=140fOSvD2htGz244AaWxH-i4bEXT0ZYAH

Trend data for the gains from 2.4 to 2.1,

https://drive.google.com/open?id=1qr8IXdDddyHXlvtyeiurZ8q6FnEH-Vah

Gain value was reduced upto 1.4 and it was observed that the problem is getting worse at lower gain values. The trends at different gains is as attached:

https://drive.google.com/open?id=1ZSmSy0Zya5k1mtWbV0o_Lu470NZjg3E0
 
deePATIDAR92,

Yes, indeed, the flow divider can be a cause of hunting. It's presumed the unit has the horizontal liquid fuel flow divider, where all the wheels are connected with each other, and to one or two toothed wheels, which the speed pick-ups are monitoring.

The bearings of the flow divider can be failing, causing increased friction to result in unstable flows. Naphtha is crazy corrosive.

The toothed wheels can also be worn, and the speed pick-ups can be worn, or loose, or misadjusted.

There must be something amiss with the configuration of the Mark VI for that LVDT Diagnostic Alarm to be present. That's for another thread, though.

Servos and valves can't be calibrated--not the ones used on GE-design heavy duty gas turbines. Only LVDT feedback can be calibrated. And, since the liquid fuel bypass valves on GE-design Frame 5 heavy duty gas turbines don't have LVDTs--there's NOTHING to calibrate.

I don't have access to any Frame 5 documentation at this writing, but there should also be no reason for changing the servo current gain from the value listed in the Control Specification. Unless the servo is not the one specified for the application, or the actuator or valve is other than the ones typically supplied by the packagers of GE-design heavy duty gas turbines there should be no reason to use any other gain value than the one listed in the Control Specification.

There should be NO reason for the null bias current to be anything other than 2.67 %/%. Especially for the liquid fuel bypass valves used on Frame 5 and -6 GE-design heavy duty gas turbines. Full stop. Period. End of discussion

I have seen dirty ("choked") fuel filters cause similar problems to the one you are describing, also.

Finally, if the liquid fuel supply piping was disturbed (taken apart and restored), or if the filters where changed and the filter vessels weren't properly bled of air, air in the piping can also cause similar problems as the one you are describing. Most liquid fuel supply piping was never designed or installed correctly, lacking pitch and high-point vents. And getting air out of piping without pitches and high-point vents can be VERY difficult.

Check valves can also be failed or failing and cause similar issues.

I have also seen fuel pump clutches that intermittently slip and can be very hard to troubleshoot without a knowledgeable person using the speed-measuring instrument. A LOT of time (and money) has been wasted by people misreading the speed-measuring device when trouble-shooting a fuel pump clutch.

As hard as it is to believe, problems like this are almost never the turbine control system. There are SO many components in a liquid fuel system that are NOT controlled by the Mark* and which have to work perfectly together in order to work trouble-free. Think about check valves, and flow dividers. And, unless there are Diagnostic Alarms pointing to some issue in the Mark* it's actually easier (usually) to use a process of elimination on the external components until the problem is found.

Another really difficult problem which can crop up at any time is poor twisted-shielded pair drain wiring terminations used for the servo coils and the liquid fuel flow divider speed pick-ups. Poor construction practices and poor maintenance practices can combine to cause intermittent problems. Wiring is so often overlooked, and can cause so many "nuisance" and intermittent problems. It should be the FIRST thing that is verified, and it usually isn't.

The feedback from all the speed pick-ups (whether it's two or three) should be identical at all times. If not, best to understand why and resolve that problem.

Hope this helps! Please write back to let us know how you fare in resolving this problem.
 
D

deePATIDAR92

Dear CSA,

First of all, Thank you very much for your reply.

Regarding,
>I don't have access to any Frame 5 documentation at this
>writing, but there should also be no reason for changing the
>servo current gain from the value listed in the Control
>Specification. Unless the servo is not the one specified for
>the application, or the actuator or valve is other than the
>ones typically supplied by the packagers of GE-design heavy
>duty gas turbines there should be no reason to use any other
>gain value than the one listed in the Control
>Specification.

In Frame 5 documentation, 2.1 regulator gain is mentioned for LFBV servo regulator.

>I have seen dirty ("choked") fuel filters cause similar
>problems to the one you are describing, also.

Both low pressure filter before stop valve as well as High pressure filter at Liquid fuel pump discharge were checked and cleaned during LFBV replacement.

>Check valves can also be failed or failing and cause similar
>issues.

During start up after LFBV replacement, during liquid fuel line purging with distillate fuel at crack speed, distillate fuel was found coming through false start drain which clearly indicates passing of one or more check valves. After firing and check valve pop up, the problem in check valves will be reflected in spread which is already running on a higher side for our GT at higher loads.

Please explain how come problem in check valves will cause hunting in speed/MW.
 
Problems with flow at the nozzle can cause problems with the stability of the liquid fuel system, especially if things like the servo gain are not correct. Just because a check valve isn't, or multiple check valves aren't, working doesn't mean it is not, or they are not, working in a smooth fashion; one or more could be oscillating. And it doesn't take much to get a larger oscillation problem started in the liquid fuel system.

Check valve problems should be VERY evident in uneven pressures using the manual selector valve and gauge at the liquid fuel flow divider. The general rule is: Any individual pressure that is 10% more or 10% less than the average of the other pressures is cause for concern about something associated with that combustor. That's the easiest way to narrow down a check valve problem to a particular combustor--using the manual selector valve and gauge at the liquid fuel flow divider. It's easiest if you have two people checking the pressures--one rotating the handle and reading the pressures to a second writing the pressures down.

You didn't say if the unit has natural gas capability. If it does, then there are liquid fuel purge check valves which could be causing problems, and that would still be evident by uneven pressures measured using the manual selector valve and gauge at the liquid fuel flow divider.

You should check all the recommended things--even the ones you don't think are relevant. Like wiring and shield drain wire terminations.

If the feedback from both liquid fuel flow divider speed pick-ups was the same and both were oscillating then the problem would not likely be the flow divider or the speed pick-ups or the wiring. BUT, since the feedback from the two liquid fuel flow divider speed pick-ups is NOT the same, it's most likely the problem is most likely related to something associated with the liquid fuel flow divider: slipping shafts or couplings; bent speed pick-up wheels; internal clearances out of tolerance; worn bearings; mis-adjusted (improper gaps) and/or loose speed pick-ups; bad speed pick-up wiring.

Some of the horizontal liquid fuel flow dividers have toothed wheels and threaded holes at both ends of the assembly. Many were shipped with three speed pick-ups, sometimes two were at one end and one was at the other. If there are toothed wheels at both ends you could try moving/reconnecting the speed pick-ups to get one from each end (if that isn't already how it's currently wired) and see if that fixes the problem, or move both speed pick-ups to the other end to see if that fixes the problem.

You could also use a frequency generator (the output must be zero-crossing, not zero-based) to simulate inputs to the Mark* (one frequency generator can usually power two or more inputs if it's a good frequency generator). The output voltage doesn't need to be more than 6-10 VAC RMS, and again the output must be zero-crossing (the output waveform MUST cross the zero axis; it matters not whether the output is sinusoidal, or a square wave, or a sawtooth--it just has to cross the zero axis with as little DC offset as possible; and polarity doesn't matter, either).

But, the inputs from the liquid fuel flow divider speed pick-ups should be the same at ALL flows and speeds/loads.

Have you tried monitoring or trending the liquid fuel bypass valve servo output current during operation? If so, what is it doing? Are any oscillations before or after the load swings?

Are you using Pre-Selected Load Control? If yes, don't. See if that helps reduce the severity of the load swings. (Don't worry; the unit won't go haywire if Pre-Selected Load Control isn't enabled and active!)

Hope this helps!!!
 
Top