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

Please see another trend I found in my collection for the position error. http://www.speedyshare.com/900781673.html

The error wasn't so high that time and went back to normal by itself. (I only changed NullBias having the alarm above 3%).
Odd is, the step change is recorded at the same moment, it could be that what caused the error created also a delay (momentary desync) for trending tool (???)

(what happens during that moment can be seen on this zoomed trend: http://www.speedyshare.com/585332895.html)

Looking at the trend and step changes, the overall picture is very confusing - that's why I wanted to raised the step changes issue.
The dither on this machine has been disabled yesterday (machine online).

What I need for my tests is to set-up servo currents and LVDT's to be trended all separately.

Regards

 

An update (sorry for not updating for so long)

1) The turbine with reversed (positive null bias, negative as a value for MKVI config).

I managed to do some tests:

- barrier bypass - no change

- swapping the MKVI hardware, SRV and its LVDT's connected in place of GCV equipment - no change

This gives me a hint to look at the field hardware, as I haven't witnessed the previous MOOG replacement I'm not sure if it's true there was a replacement.

I noticed that error on SRV position (positive one) of the value +1.4% is may cause a fired trip although the turbine is running fine on load as the position is not considered in the process control loop (as we know the SRV reference will try to maintain the interstage pressure proportional to the turbine speed regardless the LVDT feedback, at least on my machine where it's only used by VSVO hardware and HMI).

What has happened here was the valve being open during shutdown by the given above 1.4% while the command was 0% from the MKVI.
This resulted in gas leak through SRV (Exhaust went up to 500 deg C!) with GCV being still open at about 6% and the speed at about 20%. Then the protection tripped the turbine (shutdown exceeded the given period of time)

Funny, next day the position and the feedback
after the restart were correct.
I will have to work on this valve later.

This is just an information (useful I think) what may happened with an error on the valve.

2) The other machine with jumping error that could be corrected adjusting the Null Bias online.

I have the LVDT's and moog currents setup in the MKVI to be able to trend as separate signals for each coil and LVDT, but the turbine is running fine since. Nothing unusual trending LVDT's or the moog servo current.

I will update the thread with any news I find.

Regards!

 

I am in the process of checking the calibration of PM1, 2, 3 and SRV control valves, and have run in to exactly the difficulty you describe in trying to measure null current. Since we are currently offline, I tried disconnecting two of the three servo outputs and driving the servo to 50% stroke, to eliminate the competition between the controlers. It seems that this would allow the one controller to match reference and feedback, and then go to zero output (except for the null current).

For example, with one output active at a time, these are the VDC and calculated servo currents for PM1:

R active: 0.290 vdc, 0.280 mA
S active: 0.280 vdc, 0.275 mA
T active: 0.300 vdc, 0.283 mA

This calculates to a null bias of 2.79%, the I/O configurator value is 2.67%. Should I put the new null bias value in the I/O configurator? Valve position is within 1% of reference when operating.

 

Did you record any as-found conditions, or are these the as-found conditions?

Did you, or based on the as-found conditions do you, feel it necessary to re-calibrate LVDT position feedback? Valve position feedback calibration changes can impact emissions, since fuel splits are converted to valve positions. If you will be DLN tuning after the outage it might not be so critical. But if you won't be DLN tuning after the outage it might be crucial not to change the LVDT feedback calibration.

When two of three controller outputs are disconnected, then two of the three null bias currents are not being applied to the servo-valve, and it will not be able to make the actual position equal to the reference position. All this does (operating the servo-valve with a single controller output) is to verify that the polarity of the current being applied to the coil that is still connected is correct, and does not cause the device the servo is controlling to reduce the flow of fuel (or air or steam) to the unit. Remember, the servo-valves used on GE-design heavy duty gas turbines are bipolar devices and the polarity of the current applied <b>to each individual coil</b> is very, very important.

The three null bias currents are summed in the servo-valve to produce the total null bias current applied to the servo-valve. The summing is accomplished by the magnetic force developed as the result of current flowing through the coils (amp-turns).

2.79% vs. 2.67% is a negligible difference, in my personal opinion, but changing it might make the actual position more closely match the reference position; a 1% error is approaching the limit of tolerance (and that's a personal limit, not a GE-defined limit, since GE never defines one), especially for a unit with DLN combustors.

 

The values given for PM1 are as found. In addition, the actual position value as shown on the valve calibration screen while at 50% stroke ranged from a low of 51.15% to a high of 52.48%.

I followed the same procedure for PM2, and found that with only one core active, the valve would only achieve 44.5% to 45.5% of the 50.0% stroke reference. This now makes sense, realizing that all three cores are needed to supply the required null current.

PM2 is also about 2% off from reference online.
At 50% stroke, the feedback percentages are:
R - 48.17% (LVDT 1) 48.99% (LVDT 2)
S - 49.13% (LVDT 1) 48.87% (LVDT 2)
T - 48.79% (LVDT 1) 48.92% (LVDT 2)

If I replace a servo, is it recommended to change the null bias current value in the I/O configurator to the value supplied by MOOG in the attached specs, or is the procedure that GE gives in the control spec better? Thanks.

 

It would sound as if there's something amiss with the null bias current value(s) for PM1? What is the actual physical position of the valve with respect to LVDT feedback?

I am <b>not</b> a fan of the null bias calculation procedure in the control Specification. I've seen it cause more trouble and lost time than it's worth. If the null bias current doesn't fall within the range that GE specifies (-0.8 mA, +/-0.4 mA total, for a TMR) then there's something wrong with the servo-valve. And many people try to use the null bias current for something it was not intended for.

If the valve (or the IGVs) doesn't behave the way they think it should, then they start changing the null bias current(s) after they spent hours trying to calculate one based on the Control Specification value.

And worse, most people don't have any idea how to change individual servo current values for a TMR Mark V and end up just making things worse in the end.

If Moog is providing null bias currents with their servo-valves, it would seem to make sense to use them as starting points at a minimum--as long as they are in the range specified by GE.

 

I have not measured the actual position.

I don't have the tool that attaches to the side of the cylinder. It rides the actuator spring bushing up & down, and provides a small platform to rest the end of the dial indicator on.

I measured PM1 servo currents at 50%, following the GE procedure, and came up with 1.01% calculated null bias, which is outside the range allowed by the procedure. The calculated servo currents:
R - 0.007 mA
S - 0.160 mA
T - 0.136 mA

 

And what happens when you put the 2.67% value (negative) in the Speedtronic for all three processors and then tell PM1 to go to 50% stroke?

Again, I see more and More and MORE people never bothering to measure the stroke or angle. Speedtronic panels are good, but, I don't know if they're that good. But it does prove a major point: Gas turbine control <b>IS NOT</b> rocket science, no matter how hard people make it out to be. I've seen people spend tens of hours, <b>literally</b>, trying to get the zero stroke reading for LVDTs to be exactly 0.700 VAC RMS, as if that improved the accuracy of the operation of the turbine.

I've seen three-inch stroke dial indicators for about GBP50.00 and mag bases for around GBP10.00 recently. I've even used dial calipers in a pinch with a couple of zip ties to hold one portion stationary and manually moving the other portion. Before GE started using Woodward valves the GE-designed valves used to have scales on them which were very handy. (I still find it simply amazing that GE uses Woodward valves, and the likely major reason is cost. They're certainly not as good as Fisher valves, but they are less expensive, though they do require more maintenance and repair when in service. Guess another old adage is proven true: You get what you pay for.)

 

Great reading, was looking for this information for a very long time. thanks guys.

 

Latest (very late, sorry) update:

1) The turbine with reversed (positive null bias, negative as a value for the MKVI config):

Should have done that test before: I decided to check the behavior of the SRV (I think I made a mistake stating it was the GCV in previous posts) having all three coils disconnected from the MKVI. The valve went fully open instead close. So, either the Moog is installed incorrectly supplying the oil from the opposite port or the hydraulic block between the Moog and the actuator is installed incorrectly (if possible), or something else inside the Moog. Unfortunately I have no detail drawing for the system and have to compare to another machine when possible.

Surprisingly GE left us the machine with servo working in such an unsafe setup.

2) The other machine with jumping error that could be corrected adjusting the Null Bias online.

I think I have finally solved this one. One day we couldn't restart the machine, SRV didn't open. We found that 20HD solenoid didn't work correctly - stuck. The WD40 wash helped. I think, before the trip solenoid valve was closing but being sticky was not providing enough oil to the valve actuator (still draining some amount) when energized. Then after like two hours running on-line the situation changed, the trip solenoid valve fully closed providing all the oil to the actuator (pressure changed) and we had to change the NullBias to correct the new (described above in previous posts) situation.
I suspect, that before the cleaning of the trip solenoid valve, after a shutdown when we were recalibrating the SRV, the sticky trip solenoid made us the same condition (draining some oil to the tank) so we got the wrong null bias again.

Regards
minister

 

Thanks for the feedback!

Have you checked the servo valve part number against the appropriate parts list for the gas valve assembly? Are you sure the servo is the correct one for the application?

A lot of people think that all the servos on some GE-design heavy duty gas turbines are interchangeable, and while some are, many are not.

I've seen incorrect servos been used in dire circumstances, forcing some rash things to be done to make them work. I've also seen them left in service for so long because, well, they're working! And, then when they finally get around to replacing the servo (either because it fails or because of valve refurbishment) everyone associated with the initial effort develops amnesia and a lot of blame gets misdirected. I'm <b>*NOT*</b> saying that's what happened at your site, but I've seen similar things happen during commissioning as well. There is a popular sentiment amongst some TAs, especially when a job is over schedule and/or over budget to do whatever it takes to get finished and get off site. The best intentions are to send the right part back to site, but sometimes others in the decision-making chain have this same mentality, "It's working! If it's not broke, don't fix it!"

Anyway, thanks for the feedback!

 

Thanks CSA,

I haven't checked the part number as I believed the TA during commissioning, that he replaced it with another one taken from the spares, having similar result in operation after. I will try to schedule the check and inform you.

Regards to all of you guys and Merry Christmass!

 

Actually...a 4 VDC source over a 1000 ohm servo-valve force-motor coil will push 4 mA...not .4 mA...