I would like to add some more interesting facts and kindly ask for explanation.

We have 4 Frame 5 non-dln turbines running in the island mode. I have found something odd troubleshooting the moog of one of the machines, that randomly changes its bias causing 3% position error - temporary solution we use(unit online) is to recalculate new Null Bias measuring coils voltage and having coil resistance measured before, then downloading the VSVO. This way it's fine for some time, usually until next start-up. I have to mention, that we have already tried three different Moogs with the same result.

Back to the moog story. The odd thing is that each unit SRV, GCV and IGV servo currents change at the same time from one stable value to another one (usually by the step of 3, 4 %). The same time error doesn't change. I have trended hydraulic pressure 96HQ but there is no sign of the event. Also no sign on P125DC and N125DC. I'm running out of ideas what to trend.

Also polarity has been checked many times so far by confiming that MKVI is able to drive the valve by each servo coil separetely.

Although it happens on all four units, the event doesn't occure at the same time looking from unit to unit (see my trends).

Please check the data collected I uploded below:
http://www.speedyshare.com/476835562.html

Do not be surpirsed looking at these trends. One of our machines (G4) also have out of spec, positive Null Bias, since commisioning by GE (MKVI config value -12, changing moogs didn't help, anyway machine is controlled correctly.)

Any explanation on the moog servo current changes? Why all units affected?

 

You say you have four turbines and you're only having a problem with one. Or at least that's what it sounds like. Actually, I can't really tell how many units are having the problem now that I re-read the message.

But, here goes anyway.

You've changed the servos; you say the problem happens with all the servos.

You haven't told us what Diagnostic Alarms are present when this happens, or before this happens.

Do these units have IS barriers?

If you've been reading this thread, you know that the null bias current is the current required to overcome the null bias spring tension. So, if the current required to overcome the spring tension changes, it would seem that either something is making the spring tension change or something is causing the coil resistance to change.

Servo valves are about the most misunderstood device I've ever run across. They're nothing more that electromagnets which produce small amounts of torque. They're just coils of wire through which DC current is applied; the polarity and magnitude of the current causes the direction and amount of torque to change. How can they be so complicated? They're not.

The overwhelming cause of "failure" is oil cleanliness and oil temperature. There's just not much more to them than that. Very tiny, orifice-like passages and clearances which are very susceptible to dirt and varnishing.

Is there possibly a problem with oil temperature?

I have seen units with the relief valves set improperly, used as the "pressure regulators" for the hydraulic pump. This causes excessive flow through the hydraulic pump.

The hydraulic system of a GE-design heavy duty gas turbine is basically a static system, meaning that under steady-state operating conditions there is no flow. There is only flow when a valve or the IGVs are being commanded to move.

I've also seen units with the hydraulic accumulator not properly in service. This also causes the flow through the hydraulic pump to be excessive which could be a cause of high oil temperature.

Have you compared all the hardware ("Berg") jumpers on the TSVOs? And all the configuration settings on the VSVOs? Of the unit with the problem vs. a unit without the problem? What is the dither set to on the unit with the problems vs. a unit which seems to be running well (not the one with the whacko null bias value).

It just doesn't seem like the servos can be the problem here, especially if all of them on one unit are behaving similarly. There's something else that's common to the servos that the problem.

Another couple of months and I should be in a position to make my bid for Moog. The way these things get changed on a whim, they must be raking in the cash! I just want to buy the GE-design heavy duty gas turbine line, and I'll be paying my business loan off very early--I'm sure of that.

The 125 VDC battery supply voltage is converted to voltages by the rack power supplies which are then used by the VSVOs to drive the servos. It wouldn't seem likely that a problem with 125 VDC battery voltage would manifest itself in the servo outputs.

There's something different about the hydraulic system of that unit. Or, possibly even and, there is some configuration and/or jumper settings that are not correct. But, for a Moog servo to be changing its null bias spring tension requiring on-line null bias current changes? That's just pretty not right.

Anyway, that's about all I can think of. To address what I thought the problem was. Now, I'm not so sure.

One thing's for sure though: When I own that piece of Moog, I' won't be contributing to threads like this on control.com. I know better than to kill the goose that lays golden eggs.

 

Thank you CSA for your reply.

>You say you have four turbines and you're only having a problem with one. Or at least that's what it sounds like. <

Actually we have problem with two servos:
1. On one machine the SRV changes it's Null Bias mostly after restart (I have greased the stem already and stroke after many times, result the same, so it's probably not the friction). This Moog creates error >3%, to avoid trip exceeding 5% value we just downloaded recalculated null bias. This really drops the error to zero. Because of production I can't just stop the machine to do it off-line. (I'm aware about on-line software download, but hardware, single VSVO I tried many times with no bad luck).

2. Another machine GCV has positive Null Bias (MKVI RegNullBias =-11.96), and it's not the polarity.

Investigating these two issues I trended all machines and found interesting fact that I shared on previous post.

>You haven't told us what Diagnostic Alarms are present when this happens, or before this happens. <

The only alarm we have is the increased position error that can lead to turbine trip. Again, it's not the machine with positive NullBias value.

>Do these units have IS barriers? <

Yes we have the barriers on SRV and GCV (some MTL-7765ac), none for IGV. I have it planned to run the machine bypassing them, what do you think? Especially, that two other barriers for seismics on the same machine had failed.

>Is there possibly a problem with oil temperature? <

I'm going to trend all servo currents with oil temperature starting this afternoon. Right now (2:42PM) the header temperatures are 57 to 58 deg C, outside ambient 46 deg C (air oil cooler).

>I have seen units with the relief valves set improperly, used as the "pressure regulators" for the hydraulic pump. This causes excessive flow through the hydraulic pump. <

That's another good hint you gave me. I found our technician resetting hydraulic pressure on the PSV instead of using pump PCV. I will check remaining units.

>I've also seen units with the hydraulic accumulator not properly in service. <

No accumulators here.

>Have you compared all the hardware ("Berg") jumpers on the TSVOs? And all the configuration settings on the VSVOs? Of the unit with the problem vs. a unit without the problem? <

All TSVO the same (jumpers), all MKVI hardware configs the same (to be sure, I have also compared m6b with the MKVI).

Dither Amplitude 2.0, Freq 100hz on all Moogs (including IGV), on all units.

The SRV/GCV we have is the old design, containing both valves in one body, don't think we have to disable the dither as someone posted dither disabling is recommended for new types of valves.

Regards!

 

Dear Minister,

I offer this suggestion as something to trend, although from your post it is difficult to tell if this is happening on one gas control valve only or multiple valves. And if the problem is on one unit or multiple units.

Anyway I had an issue recently with a frame 7ea machine. We were intermittently getting an alarm for low P2 interstage pressure, at the same time we noted the SRV valve position was erratic. Trending the position of the SRV in the MKVI confirmed that the SRV valve was not holding a steady position, under constant load and constant main inlet fuel gas pressure. I trended the SRV servo current known as "fagr" and found it was erratic during the event. I then trended the individual servo currents from each MKVI core, in my unit these are called FAGR_R, FAGR_S, and FAGR_T. What I found by running a high speed trend, 40ms, was that intermittently the "T" core would decide to fight the other 2 cores, it would try to close the valve. The other 2 cores, "S" and "T" would respond by opening the valve. Then "T" would decide to go back to normal, and "R" and "S" would have to catch up.

I don't know if this will help, but it would be something else for you to check. Please also note that replacing the VSVO card repaired the problem. Also note that I have seen some sites where the signals were never added at the card level to be able to trend the servo currents individually. Look at your .m6b file, under the VSVO card points, for the given regulator you are having trouble with look at the "ServOut#NVR, ServoOut#NVS, and ServoOut#NVT. If there is no point associated with it, it will say "not used", if that is the case you will need to create a point for each card point and download before you can trend the currents individually.

Again this may not be the problem, but it would be something else you can check. Lastly this assumes that your MKVI is a TMR system, and not a simplex.

 

One more time: The SRV regulator is a pressure loop. Even if the LVDT calibration is WAY off, the pressure loop is going to put the valve at the position required to make the actual P2 pressure equal to the P2 pressure reference. It doesn't make any difference if the valve thinks it's at 35.4% and it thinks it should be at 28.4% or 44.4%, or if it's actually at 31.4% stroke (physical measurement) and it thinks it should be at 37.1% and the LVDT feedback says it's at 35.9%. <b>As long as the actual P2 pressure is equal to the P2 pressure reference, everything is perfectly okay.</b> When it's not perfectly okay is when the unit is a TMR panel and the three servo currents are not fairly equal; THAT'S a problem.

I'll bet any amount of money the differential is related to the calibration method of that combined SRV/GCV assembly.

Would you list the servo gains and null biases for all of the servos? For all of the turbines? A simple chart:
<pre>
GT1 GT2 GT3 GT4
Null Bias/Gain Null Bias/Gain Null Bias/Gain Null Bias/Gain
SRV
GCV
IGV
</pre>
I don't believe you told us if these are TMR or SIMPLEX Mark VI panels; if so, please remind me.

As far as dither goes, my personal belief is that none is needed with most of these legacy-style actuators.

I have seen barriers cause all manner of strange problems over the years. They appear to be fine, but they're not. Also, some seem to be temperature and current sensitive when they start "failing."

 

Dear MIKEVI

Thank your for reply.
I have read your post about the servo current trend 40ms before. I have this it in my plans but have to wait for scheduled shutdown to create new signal PINs.

Good idea with testing another VSVO. I realized I have one unused VSVO (R/S/T) and two terminal boards.

minister

 

Dear CSA,

Thank you this time for your reply!

> One more time: The SRV regulator is a pressure loop.
---- snip ----
<b>As long as the actual P2 pressure is equal to the P2 pressure reference, everything is perfectly okay.</b> When it's not perfectly okay is when the unit is a TMR panel and the three servo currents are not fairly equal; THAT'S a problem. <

Thanks for reminding that. The above was also mentioned many times on this forum. Although it's pressure control loop our concern is not to be tripped by position error.

>Would you list the servo gains and null biases for all of the servos? For all of the turbines? A simple chart:
<pre>
GT1 GT2 GT3 GT4
Null Bias/Gain Null Bias/Gain Null Bias/Gain Null Bias/Gain
SRV 2.4/1.8 3.9/1.8 3.2/1.8 -11.69/1.8
GCV 3.3/1.8 3.1/1.8 3.3/1.8 2.65/1.8
IGV 2.61/6.8 2.67/6.8 2.82/6.8 3.1/6.8
</pre>

The way we calibrate these valves is to lift the stem and insert and leave filler gauge to get rid of the gap between the actuator and valve stem. This way the valve is still closed and LVDT indicates 2.5% difference from its rest position.
Hope this is the way.

> I don't believe you told us if these are TMR or SIMPLEX Mark VI panels; if so, please remind me. <

TMR, sorry, should have mention that at the beginning.

>I have seen barriers cause all manner of strange problems over the years. They appear to be fine, but they're not. Also, some seem to be temperature and current sensitive when they start "failing." <

Can I bypass these barriers? I know it's for EEX zone but I do not recall (I'm not 100% sure) seeing them on 9E turbine in Europe for example.

After trending the servo currents jumps together with Oil Temperature no relation could be observed (during night when temperature decreases, the frequency of jumps seems to be the same)

Can anyone trend servo currents for at least 12hours to observe if these jump like it happens here? I do not understand why three servos jump at the same time and it happen on all machines but at different times. We have running hrs varying from 6000 hrs to 12000 hrs and we have never change the lube oil. I wonder if these jumps happen anywhere else?

Regards!

 

You have a TMR control system. That means that each control processor has it's own gain and null bias for each coil of the servo valve.

Can you please list the null bias for each processor for each servo of each turbine?

One thing all manufacturers do is to "copy" functions to similar applications, like the LVDT position error check. Since the position isn't the problem for the SRV (the inability to control P2 pressure is the problem), do you think it makes sense to use the same settings for the SRV as for, say, the IGVs or the GCV?

I'm still confused about how the position error can start out at one value (even if it's zero) and then change when the unit is running. And I'm confused when you say you download new null bias value(s). Do you download a new value to one processor? Or to all three processors, one at a time?

As MIKEVI says, there might be a problem with one of the processors thinking that the P2 pressure isn't the same as the other processors, and that would cause that processor to change its current output and the other processors would have to change theirs to counteract the other. This happens a LOT, and if you have three P2 pressure transducers and one of them drifts or has a leak or a wiring problem, that can cause a problem like this.

How many P2 pressure transducers does each turbine have? One or three or ????

Have you trended the P2 pressures for all three processors to see what they're doing before and after this problem?

When you put the feeler gauges into the gap prior to performing the calibration of the SRV and GCV, that's to PUT them in their true zero position. When you remove the feeler gauges and the valve stems drop and the indication goes negative, that's what's SUPPOSED to happen.

The purpose for the gap is to ensure that the valves aren't held open by the actuator bottoming out. Some places it's called "closed end overtavel". The true zero stroke position for these valves is <b>NOT</b> with the valve stem fully down when the valve is closed; in that position the valve stem is not touching the bottom of the valve plug.

The true zero stroke position is when the valve is closed and the valve stem is touching the bottom of the plug, and that's what the feeler gauges do: They keep the valve stem in contact with the bottom of the valve plug when the valve is closed and ensure that at the end of the calibration procedure that the valve will return to the same position as when it started (which is one of the checks that AutoCalibrate performs when run).

The use of barriers is entirely a function of the code and requirements of the country or regulatory agency where the unit is installed, OR the policy of the company which is operating the unit. There are thousands of units operating in all manner of applications around the world which do not and have never had IS barriers installed.

So, only you can decide if you can operate without the barriers. We don't know the application or the site conditions so we can't tell you to bypass them. We don't know if the IS barriers are installed on the LVDTs or the servo-valve outputs.

When you're trending oil temperature, you're not trending the temperature of the hydraulic pump discharge. You're trending the oil temperature of the oil at the inlet to the hydraulic pump.

I really don't understand this whole thing; just when I think I'm getting a handle on it, you say "... I do not understand why three servos jump at the same time and it happen on all machines but at different times...." By "three" servos do you mean all the coils of a single servo valve? Or do you mean all the servos on all the machines? You say it's only on the SRV, but then you say it happens on three servos on all machines at different time. Is this happening on the SRV of all machines at different times?

Again, I can't imagine why the null bias spring tension would change during operation, or why the coil resistance would change during operation. Heat would seem to me to be the most likely cause for either, but I've never heard of this problem, or a problem described like this.

And, a lot of times that is a part of the problem: Someone attributes an occurrence to this reason or that reason without any real evidence that this reason or that reason is causing the problem.

You don't know if the coil resistance is changing or if the null bias spring tension is changing.

All you know is that a position error is changing (increasing) and that you can change (reduce) the position error by changing the null bias current value downloaded to "the Mark VI" and we don't know if you're downloading the change to one VSVO or to all three VSVOs one at a time.

I would really like to help solve this problem, but, again, every time I think I'm understanding what's happening then I re-read the posts and I get confused even more.

I really think MIKEVI has suggested a good course of action, and that you need to also trend the P2 pressures of all the processors to try to understand what's happening.

You should be trending the P2 pressure of each processor, the LVDT feedback from both LVDTs as read by each processor, and the servo current outputs from each processor.

I'm also keenly interested to try to solve the problem with the outrageous null bias current. You say you've replaced this servo and the new one still behaves the same way???? And you've tested the servo current polarity under the individual control of each processor for this device (GCV or SRV or ???) and it will closely maintain the position with only one processor????

I think there's a typo in your chart, because you've previously said that the GCV of one unit is the one with the outrageous null bias. We really need to know what the running servo currents for each processor are for the servo with the outrageous null bias value.

Are you sure there's not an oil leak somewhere in the actuator of the device which this servo is installed on?

I apologize if my response seems a little "random" but I'm really baffled by this and at the same time very interested to solve these problems.

You should know this: On every GE-design heavy duty gas turbine operating anywhere in the world, there is likely some error in position feedback vs. position reference on one or more servo outputs. Some more than others. But, they are all running.

And we're only talking about position feedback vs. position reference. I have been to more sites that have incorrectly calibrated LVDT feedback so that the actual physical position is way off from the indicated LVDT feedback. And the turbines still run. And run well.

I heard a former colleague say once a long time ago about GE-design heavy duty gas turbine control, "This ain't rocket science." And he was very correct. If it were rocket science, GE would have been out of business a long time ago because if every servo output had to operate with zero error the units wouldn't run. But they do run, and they run very well.

Sure, in a perfect world we'd all like the position error to be zero, and it should be zero. But, it doesn't have to be zero. What counts is: Is the position error increasing or decreasing?

This has been said many times before on control.com: The value of something today isn't really informative. It's the value today vs. the value last hour or yesterday or two days ago or one week ago or one month ago or six weeks ago, and whether or not that value has changed in that time and how fast it has been changing.

Sure, a L.O. header temperature value of 90 deg C isn't good, but if the unit has been running for four years with an indicated L.O. Header temperature of 90 deg C, would you say it's been running incorrectly? Or would you go to find another way to verify that reading?

In your case, the position reference error seems to be changing relatively quickly and then remaining relatively constant. From what we can understand. Which changes with every reading.

Sorry; I'm both perplexed and intrigued. And, I'm a very literal person (but you couldn't tell that could you?).

 

Minister,

the great thing about GE's control systems are that they have lots of great tools to use for diagnosing problems. The downside is that all the same tools are not given to all the customers (some .m6b have pins, some don't etc.) I am not sure where the ball gets dropped, either during commissioning or straight from Salem. But no matter, we have this forum full of knowledgeable people to take advantage of. And best of all we have a participant such as yourself that provides the information we need to try and help solve a problem, and hopefully we all learn something in the process. I wish you continued luck with your issue, and look forward to you resolving the problem.

 

CSA,

thanks for your post. I appreciate your time spent at the computer.

> Can you please list the null bias for each processor for each servo of each turbine?

The NullBias values I gave are the same in each processor for each servo coil (RST). You confused me a little with this question. Using Toolbox I have (v11.02.09), when I select Download Configuration, it goes directly to all three VSVO cards, with no possibility of selecting destination processor (R, S or T). It could be only possible if all three VSVOs are in Simplex config like for example Thermocouple VTCC cards. To set-up hardware config you only have one common field, not three separate fields for R/S/T.

To have different NullBias values in each R/S/T card, the only way that comes to my mind is to remove Ethernet cables during download. Have you heard about MKVI having different NullBias values for coils on the same servo? What is your way to achieve that?
Gains are of course the same for R/S/T as NullBias and the rest of the HW config.
More on that, I confirmed the config with TSM, gain and null bias values are the same in all three processors.

What I found using TSM checking regulators (command A*pplications -> R*egultors) and then servos (commands S1 for SRV, S3 for GCV and S4 for IGV) : G1/G2/G4 have parameter FIX 1 and FIX 2 equal to zero (for all three GCV/SRV/IGV servos) and MKVI LVDT Min/Max (0/100% position) values at LIMIT HI VOLTS and LIMIT LO VOLTS. The G3, the one that must have had SRV Null Bias corrected after two last restarts has MKVI values of SRV and GCV LVDT min/max in FIX 1/FIX 2 fields and LIMIT HI VOLTS and LIMIT LO VOLTS are different by around +-0.100 (FIX 1 + 0.100, FIX 2 - 0.100). IGV is the same as G1/G2/G4. I'm not sure if you are familiar with this part of TSM. I found it today looking for any unusual thing. It may be that someone here started the calibration and haven't fix Min/Max value - have no idea. Can only check on stopped unit. Also TSM has a lot of stuff that can be checked.

> do you think it makes sense to use the same settings for the SRV as for, say, the IGVs or the GCV? <

For some reason we have it (5% position error trip), I can't think of the situation that this protection is needed, but as it is know that SRV is fuel gas pressure controlled valve, its LVDT is used as a feedback for VSVO card (SRV VSVO regulator type is 2_LVPosMax), so loosing the LVDT the card probably doesn't know how to adjust servo current (what direction to go).
Am I close?

> Do you download a new value to one processor? Or to all three processors, one at a time? <

OK. See the trend I have uploaded below:
http://www.speedyshare.com/470864238.html

The whole story started there
You can see that one day we started having the error. Servo command had to be higher then the position required to drive SRV to desired position. Suddenly either SRV position increased by overcoming the friction or the moog current increased for some reason and then MKVI reduced SRV command and so we got both value (position and command) at the same level. I'm not sure what was the reason I can only speculate as above (as I said I greased the SRV/GCV stems with the same result after restart). Then I started my research and found such Moog current step changes on all machines (but again, on other machines the do not cause an position error). Once again, on each machine. lets say on GT1, all three moogs SRV/GCV/IGV indicates the step change at the same time.
Please see the trend I included in my first post. I highlighted each GT set of moogs with different color.

Again, in my post I wanted to confirm if anyone has ever observed these step changes, I had mention GT3 servo problem as an origin.
So, If you help me with all my problems I will be very grateful.

>As MIKEVI says ----snip----

>How many P2 pressure transducers does each turbine have? One or three or ????

>Have you trended the P2 pressures for all three processors to see what they're doing before and after this problem? <

I have just trended P2 (R/S/T) and see nothing (flat lines during the step change that just happened on GT1 and GT3). Will include it in GT3 trend at next start-up.

> When you remove the feeler gauges and the valve stems drop and the indication goes negative, that's what's SUPPOSED to happen. <

Yes, that's right (the value (negative) is from 3% to 2.5% depending on machine).

>The purpose for the gap is to ensure that the valves aren't held open by the actuator bottoming out. <

That is exactly how I understand the valve. I'm assuming the gap is for safety if the seat is worn and no gap the valve will leak. Also the drawing say to check that gap is within desired limits and if no grind the piston rod to obtain it.

> The true zero stroke position is when the valve is closed and the valve stem is touching the bottom of the plug, and that's what the feeler gauges do: They keep the valve stem in contact with the bottom of the valve plug when the valve is closed and ensure that at the end of the calibration procedure that the valve will return to the same position as when it started (which is one of the checks that AutoCalibrate performs when run). <

No doubt about it.

>The use of barriers is entirely a ----snip----.

> So, only you can decide if you can operate without the barriers. We don't know the application or the site conditions so we can't tell you to bypass them. We don't know if the IS barriers are installed on the LVDTs or the servo-valve outputs. <

Thanks for advice. I will feel more confident doing it.

>When you're trending oil temperature, you're not trending the temperature of the hydraulic pump discharge. You're trending the oil temperature of the oil at the inlet to the hydraulic pump. <

No other option for me right now.

> I really don't understand this whole thing; just when I think I'm getting a handle on it, you say "... I do not understand why three servos jump at the same time and it happen on all machines but at different times...." By "three" servos do you mean all the coils of a single servo valve? Or do you mean all the servos on all the machines? You say it's only on the SRV, but then you say it happens on three servos on all machines at different time. Is this happening on the SRV of all machines at different times? <

Again, as stated the thing above (it's really difficult to me to explain it as English is not my native language).

What I can trend, is jump/step change up/down/up/down (and so on) of the total current of each Moog. But I observed looking at card points ServoIOutxNVR/S/T that it happens also in the same direction on all three coils (second screenshot I attached, you can call it static observation). So, on one single turbine, the change happens at the same time on all 9 coils (3xSRV, 3xGCV, 3x IGV), and you can observe the same on all machines but from machine to machine it's not synchronized. Please see again the trend and color marked GT's.

>Again, I can't imagine why the null bias spring tension would change during operation, or why the coil resistance would change during operation. Heat would seem to me to be the most likely cause for either, but I've never heard of this problem, or a problem described like this.

>And, a lot of times that is a part of the problem: Someone attributes an occurrence to this reason or that reason without any real evidence that this reason or that reason is causing the problem.

>You don't know if the coil resistance is changing or if the null bias spring tension is changing. <

I agree, I don't want to risk and disconnect servo coils one by one online. What I found after stop that the current is different each time (I register in Excel sheet all NullBias calculations), not resistance. It stays on all machines 1200- 1300ohm on all machines (barriers give additional resistance, I think I checked that a year ago).

> All you know is that a position error is changing (increasing) and that you can change (reduce) the position error by changing the null bias current value downloaded to "the Mark VI" and we don't know if you're downloading the change to one VSVO or to all three VSVOs one at a time. <

All three at one time. Simply calculation, have all three resistance noted before the start last month, I measure the voltage for each coil separately, divide V/R, summarize all three results and finally divide by three to get average (negative result must be placed as positive in MKVI config and so on). This is the GE calculation way given by the ControlSpec and it works (offline , for me also online)

>I'm also keenly interested to try to solve the problem with the outrageous null bias current. You say you've replaced this servo and the new one still behaves the same way???? <

Actually It has been replaced by GE TA, that left it saying it's fine. They just wanted to leave the place )))

> I think there's a typo in your chart, because you've previously said that the GCV of one unit is the one with the outrageous null bias. We really need to know what the running servo currents for each processor are for the servo with the outrageous null bias value. <

I wrote G4 - on the trend you will find G4\signal - that was the reason I used it instead of GT4. Affected valve is SRV.

> Are you sure there's not an oil leak somewhere in the actuator of the device which this servo is installed on? <

I will check, I have to re-read your posts guys and prepare a checklist)

Finalizing the story: you can see that we have two major problems - very high positive null bias on one machine running relatively good - position error that disappear after start-up or we have to help it disappears changing NullBias value (we did it last time instead of waiting).

...And one minor that is servos current step changes on all machines that I accidentally discovered.

It took all day to think about answers also have a headache. Sorry if I messed something again.

Regards

 

In the early days of Toolbox, it was possible to download to individual VSVOs. Maybe that's not the case any longer; I haven't seen a Mark VI in a couple of years.

I have never seen trends like this. My initial reaction is that there is something about/with the IS barriers, but that's just because I don't have a lot of experience with them and the experiences I have had have not been very good.

As for the friction comment, when I have seen friction on a gas valve (SRV or GCV) the effect has been that the servo current increases with no change in position until the device "jumps" usually to a position past the reference and then the current changes to try to drive device back to the reference with no change in position until it "jumps" to a position past the reference and this continues and in some cases gets worse. So, the valve behavior was jumpy and erratic. On either side of the sticky portion of the valve stems, the action was normal. But where the shaft was worn and scored (on a couple of turbines) or where the cylinder walls were scored (on a couple of turbines) the behavior was jumpy and erratic.

From what you describe, it seems that the current just "jumps" and without seeing the valve position feedback at the time the current jumps, it's difficult to say what's happening.

I don't recognize the IS barriers you cited. Have you reviewed the manufacturer's manuals/documents to see if these are properly applied?

I would like to know how the *circuit* resistance changes when this phenomenon occurs. It would be very interesting to monitor the voltages of each servo coil and see what happens. A change in voltage would be indicative of a change in circuit resistance. Would it be possible to monitor the voltage across the IS barrier and the voltage across the servo coil?

I do NOT believe the servo coil resistance is changing, but I've been wrong before and I'll be wrong again. I have seen Moog provide some very good troubleshooting assistance when asked to get involved. I know they are slow to respond, but when they do get involved they are pretty helpful.

I don't believe the null bias spring tension is changing.

Have you had your Lube Oil tested for contaminants? And agglomeration (I think that's what it's called)? Some of the newer formulations of turbine lube oils seem to be causing lots of problems for servos. I think BP has a formulation that has been used successfully by a lot of heavy duty gas turbine users.

When these jumps in current take place, does the turbine power output change? Does the unit continue to run normally or does an operator have to make a change to keep the unit running as desired?

If the AHJ (Agency Having Jurisdiction (thanks for that Phil Corso!)) would permit a test with the servo IS barriers disconnected it would be very interesting to see what happens.

I'd really like to know what happens to the position feedback at the time the current jumps.

That's about all I can think of.

 

MIKEVI

The forum is great thing, I agree with you.
I will update my post if I get any new information.

 

The more I think about this the more I think if you're seeing it on all the servos on all the units at your site, the one thing that's common to the gas valves is the IS barriers. I also wonder if they're also affecting the other servo outputs on the VSVO, such as the IGVs.

I did some checking on the MTL site, and it says the MTL7765ac is primarily for high frequency low voltage applications. Hmmmm..... I also wonder about the resistance readings you mentioned, which seem much higher than those listed in the tables in Sect. 8 of the manual.

 

CSA,

I had opportunity to shutdown GT1 and found something interesting.
It seems that "jumps" are not caused by the hydraulic part of the system as these also exist during cool down and after when Auxiliaries are off.

Trend with some description added (trip occured ).
http://www.speedyshare.com/748565606.html

Will try today to bypass barriers and trend again.

> It would be very interesting to monitor the voltages of each servo coil and see what happens. <

The voltage changes for sure, I had usual average value of around -0.320V after last calibration, then when I started the machine and found the position error, I made another measurement of coil voltages and as described before recalculated the NullBias and downloaded new VSVO config (average voltage this time -0.490V).

> I do NOT believe the servo coil resistance is changing, but I've been wrong before and I'll be wrong again. <

I hope barriers is the reason. Could it be changing MKVI cabinet inside temperature affecting them? I wonder if it's sync'ed with the cabinet fan switching on/off (I can simply check it by keeping all fans on for few hours.

> Have you had your Lube Oil tested for contaminants? <

Yes we have it tested, but we are trying to make it more often (monthly). Nothing unusual last time, let's see next time.

> When these jumps in current take place, does the turbine power output change? <

Nothing unusual I have trended DWATT as well.

> test with the servo IS barriers disconnected it would be very interesting to see what happens. <

Yep, this I will provide soon.
I have to find some extra terminals for barriers bypassing.

Cheers!

 

This is troubleshooting: Understanding the possible causes of a problem and working to eliminate them one at a time or to narrow the field of possibilities down to a very few.

I think it's very telling that when these "jumps" occur when the turbines are running that there is no change in power output. The Mark VI is fast enough to maintain steady and stable power output in the face of whatever is causing the servo current to have to change. And, even if the position error increases (for whatever reason) the Mark VI is automatically "compensating" and keeping the turbine operation stable and steady.

You have not provided any information on specifically what's happening to the LVDT feedback when these jumps are occurring, and if the LVDT excitation and/or feedback also have IS barriers. If so, what is the manufacturer/part number?

And here's where I'm going to suggest that the problem is <b>not</b> the changing servo currents, but the changing LVDT feedback. You say that the position error changes, and when that happens the servo current would change to try to maintain the error closer to zero. Right?

Please trend the LVDT feedback as well as the servo current, and the LVDT feedback from each LVDT would be the best to trend, not the high-selected value. Even when the unit is not running.

It would be unusual for the Mark VI enclosure temperature to be changing very much. It should be located in a temperature- and humidity controlled environment, the emphasis is on control. Are you saying the temperature in the Mark VI panel, or wherever the IS barriers are located, changes? If so, by how much during the course of the day? Are there fans controlled by temperature? If so, that's unusual, not unheard of, but not typical.

I think the barriers you listed are considered "passive" devices and don't require bus power, but the manufacturer instruction book is really unclear on these particular barriers. If they are powered, where do they receive their power from?

What IS barriers are used for the LVDTs? Do all the LVDTs have IS barriers?

Another possible test would be to take suitably sized 1000 ohm resistors (heat dissipation) and connecting them directly to the TSVO in place of the barriers and then trending the servo current. (This would have to be done when the unit is not running.)

If the current still changes, then I would suggest removing the IS barriers from the LVDT circuits and monitoring the individual LVDT feedbacks and the servo currents.

This also happens sometimes in troubleshooting: The focus gets placed on the effect rather than the "affect" (the cause). The whole servo regulator loop needs to be considered: the reference, the feedback, the output.

Let's not drop the theory that the servo barriers may be the problem, but let's not lose sight of the other possible contributing factors. You have addressed the hydraulic possibility pretty conclusively. But, servos do get very warm in the environments they are placed in.

Sometimes, the gas valve servos are located at the top of the L.O. reservoir/tank, where the vapors collect. The L.O. in the tank is hotter than the L.O. header (which is cooled).

The IGV servo is usually located in the turbine compartment, which experiences large temperature swings during starting, operation, and shutdown.

 

Thanks CSA for you input:

Post update:
>You have not provided any information on specifically what's happening to the LVDT feedback when these jumps are occurring, and if the LVDT excitation and/or feedback also have IS barriers. If so, what is the manufacturer/part number? <

No barriers at LVDT at all. Summarizing we have only 6 barriers installed for SRV and GCV.

I have used terminal block to bypass the barriers and during last 12 hours "jumps" occurred again. We can exclude IS barriers from servo current change phenomena, but still have to take them into account for two running units, (position error and high positive null bias).

Let's keep in mind, that the offline unit is the one controlled "perfectly" (only servo current jumps" are observed). So let's concentrate on servo current jumps.

> And here's where I'm going to suggest that the problem is <b>not</b> the changing servo currents, but the changing LVDT feedback. You say that the position error changes, and when that happens the servo current would change to try to maintain the error closer to zero. Right? <

Right. This gave in another suggestion as the error appears some time after restart. It can be possible that we loose one LVDT by vibrations while running unit (e.g.. loose wiring) or we are getting the LVDT back by the same reason.

> Please trend the LVDT feedback as well as the servo current, and the LVDT feedback from each LVDT would be the best to trend, not the high-selected value. Even when the unit is not running. <

Trend set-up and running. A hint for people that want to trend LVDT's. You have to turn Monitors on (we have it unused, but I met the feature used on some sites). Select 1_LVposition as monitor type, select LVDT you want to monitor (see your Regulator), fill in Mn and MxLVDT1_Vrms with data you have in your Regulator. Keep in mind that Regulators use usually two LVDT's, and you have to separate them using two Monitors. After the config is downloaded you will see each LVDT value in VSVO Card Point section under MON1, MON2....MON12 (depending of your set-up)

> It would be unusual for the Mark VI enclosure temperature to be changing very much. It should be located in a temperature- and humidity controlled environment, the emphasis is on control. Are you saying the temperature in the Mark VI panel, or wherever the IS barriers are located, changes? If so, by how much during the course of the day? Are there fans controlled by temperature? If so, that's unusual, not unheard of, but not typical. <

We have HVAC unit for the MKVI room so we have quite good environment.
And yes, we have fans controlled by thermostats, I was aware of this as very often we have to clean cabinet ventilation inlet filters and fan outlet filters from accumulated fine desert sand. After trending servo currents yesterday afternoon with all of fans running the servo current phenomena still exists. The average temperature measured using TBTC cold junction is 30 deg C (IS barriers are mounted on the back side of the cabinet where TBTC are mounted, also fans are on MKVI terminal board side only)

> I think the barriers you listed are considered "passive" devices and don't require bus power, but the manufacturer instruction book is really unclear on these particular barriers. If they are powered, where do they receive their power from? <

Passive barriers.

> Another possible test would be to take suitably sized 1000 ohm resistors (heat dissipation) and connecting them directly to the TSVO in place of the barriers and then trending the servo current. (This would have to be done when the unit is not running.) If the current still changes, then I would suggest removing the IS barriers from the LVDT circuits and monitoring the individual LVDT feedbacks and the servo currents. <

Will add to the plan.

> Sometimes, the gas valve servos are located at the top of the L.O. reservoir/tank, where the vapours collect. The L.O. in the tank is hotter than the L.O. header (which is cooled). The IGV servo is usually located in the turbine compartment, which experiences large temperature swings during starting, operation, and shutdown. <

That's the case we have, valve servos are mounted inside the tank underneath the valve. IGV in turbine compartment. As the unit will stay some days, temperature should come down and we will see possible effect on the trend.

After all day trending:
Please find another trends (40ms sample time), this time also one toolbox .trn file.

1) change observed only on GCV/SRV (I was measuring coil resistance a while before it happened. Resistance stays always the same):
http://www.speedyshare.com/727988916.html
http://www.speedyshare.com/314203559.html

2) GCV/SRV/IGV jump (see signal description in first file, scale for some signals is different, color the same)
"Jump" UP
http://www.speedyshare.com/810365505.html
"Jump" DOWN:
http://www.speedyshare.com/340410574.html

I'm trying try do the same with dither off. (As dither is an AC on the DC command signal, can it induct something in the circuit?):

After switching dither off, I observed that all current signals (SRV R/S/T, GCV R/S/T, IGV R/R/T) have moved in the middle of the "jump high" and "jump low" servo current values. It's moved down by 2% from the high value or 2% up from the low value.
Is it coincidence, when the dither amplitude is also 2%?

Please see dither off trend: http://www.speedyshare.com/860219617.html

These 4% current switches seems odd when I have checked servo currents loading the machine by 1MW (around +2% GCV and SRV) and couldn't even indicate the difference on the trend. If it's not the coil resistance that changes, is it whole circuit loop resistance I mean inside MKVI hardware? The dither seems to be the one not explored ))

Another experiment: On running unit I have gradually changed dither amplitude for GCV from 2.0 to 0.2% (by 0.2 % step) , and observed servo current travelling from -7 % to -5% without load or position change.

Again, not to confuse you guys, these are trend taken on the offline machine, the machine that has no problems with position error. I will troubleshoot affected unit ASAP (read when available after shutdown - software modification needed to trend LVDT's and servo outputs separately).

At the end of the post, after 4 hrs of trending servos with removed dither I can confirm no jumps on offline GT so far (have to finish my shift). Tomorrow morning I will updated whether jump occurred or not.

CSA, please share your thoughts and suggestions for next checks.

Cheers!

 

I was thinking in my bed about this changes of the servo current. It's not the resistance that changes, but the voltage and then current had to compensate the change for the servo to do the same work (Power = U*I = const).

(I recall having sometime unusual higher voltage across coils during NullBias verification, I think I wrote the value before).

Ok have to go to sleep

 

Update:

After switching the dither off I haven't observed any more jumps last night.

I'm going to do another experiment, put all three servos on different dither frequency and amplitude trend. As MKVI fastest sample time is 40ms I should see oscillation at least on 12.5Hz and 25Hz dither trend. I want to check if there is a relation between jumps and dither frequency.

I suspect that these jumps are just recorded extrema of the oscillation signal when its frequency is higher than the sample rate. The trending tool gets periodically unsynchronized with the servo current oscillation generator.

Catch you later.

 

Update 2:

I can observe full curve oscillation for servo currents with 12.5 Hz and 33Hz frequency dither.

The 25Hz (40ms) dither as being the same as trend sample could be catched only as shown on attached the screenshot:

http://www.speedyshare.com/660758165.html

Well, now I will focus on the other two issues I have here. I will write more when I have a chance to make more tests on these machines

Regards

 

This explanation would make more sense, especially because there is no upset in the turbine output when these "jumps" occur.

But what isn't explained is the position error that you reduce by changing the null bias.

I'm not a fan of dither for GE-design heavy duty gas turbine applications. The nature of operation doesn't really cause the control valves to require dither from my experience, though I've recently been made aware of some "lore" (because it's not documented anywhere) that there was non-configurable dither built into the Mark V. But there wasn't any dither in the Mark IV or earlier Speedtronic panels because the servo outputs were all analog circuits and there wasn't any dither that I was aware of.

GE Trending tools can be misleading, as can any manufacturer's. Technology is great, but, digital technology has its limits, and this is one of them. An old-fashioned pen-and-ink chart recorder had its limits as well. In general, and this case is probably a great exception because of the variable frequency rates for the dither option, the GE Trending tools are pretty good because the execution rate is not generally higher than the fastest trending rate. But, again, this could be an exception because of the rate issue. Very highspeed toggling of discrete inputs is also hard to catch on GE Trending tools.

This explanation also makes more sense because of the repeatable nature of the "jumps". They never seem to increase or decrease by more than the same amount, at least from the trends you've sent.

But, the position error, which I haven't really seen a good trend of, is still puzzling.