Dear All,

We are operating frame-6B and frame-9E gas turbines.

What is difference between calibrate the SRV/GCV/ICV/LFBV from auto calibrate manual mode and user defined display?

Which is more safer?

Thanks
G.Rajesh

 

Neither.

 

Can you explain in detail?

 

A question that has been asked for many years. What is better, using AUTOCAL or manually calibrating regulators using the Manual function in the User Defined Display pages.

In the past and in past versions of Speedtronic, we didn't have AUTOCAL and had to do everything manually and it worked.

I believe AUTOCAL is more accurate and have pretty well proved it by doing it both ways and comparing the position accuracy. If you do it manually, you are depending on getting the regulator end positions accurately and recording the feedback signals at these positions before inputting to IO Configuration. To me, AUTOCAL gets these positions much more accurately and therefore gives you a more accurate calibration.

I guess this is open to discussion and I would be interested to hear other users opinions.

 

Thank you Johnston, I too early waiting to hear others opinion.

Thanks
G.Rajesh

 

In reply to Bob Johnston,

The accuracy of LVDT calibration using AutoCalibrate is dependent on how well the actual measured positions are calculated using the Control Specification values and then those calculations must be entered into ACALIB.DAT.

So, if an SRV is supposed to be at 100% position when it's at 1.500 inches, and when it's at full open mechanical stop it's at 1.625 inches, then when it's at the full open mechanical stop it's at 108.33% of stroke. That value must be entered into the POSITION_NEG_SAT field in ACALIB.DAT.

That way, when AutoCalibrate runs the device up to the full open mechanical stop it knows that position is really 108.33% of rated stroke.

So, if AutoCalibrate is properly set up it works very, very well. The only time I've seen AutoCalibrate not work well is when there is huge "slop" in the IGV actuator for Frame 5s and Frame 6s. If the device can't go to the same start and stop points because the pins are worn or the bolts are loose, then AutoCalibrate can't reliably calculate the voltages it gets from the feedback.

But, as you say: Either method will yield reliable results if used properly. It's all in understanding the process that's required to use either one.

Unfortunately, GE has never really written a proper document for either method. And, if they did, it would only be good for the machine it was written on; there are a lot of different configurations out there, and people generally don't want a generic procedure or one that's not specific to their site. But, they don't want to write a site-specific procedure, either, using a generic procedure and their understanding of their site and its configuration.

 

I haven't calibrated a gas turbine, but I have calibrated a lot of LVDTs in other applications. The most common reason for recalibration that I have seen has been due to some sort of mechanical change. The most common causes seem to be something mechanical has broken, come loose, been moved or replaced, or some such similar cause. The LVDTs or signal conditioners never (or almost never) just change on their own.

It has always been very critical to observe and inspect the mechanical action associated with the LVDT. I have investigated discrepancies between machine measurements and QA inspections and have seen cases where the (gauge head style) LVDT passed calibration without even coming into contact with the part it was supposed to be measuring. "Calibration" can cover up a host of problems, but unfortunately some people tend to treat it as some sort of magical incantation that will make the real problem just "go away".

I'm not familiar with the calibration method used with GE gas turbines, so I can't comment on that. However, what I have found to be the most effective means of dealing with calibration "problems" has been to make the calibration procedure as transparent and methodical as possible. In addition, the operator should be able to conduct the complete calibration from a position where he can actually observe what is going on.

Furthermore, there should be written procedures for calibration, and those procedures should separate the concepts of "checking the calibration" from "adjusting the calibration". If the calibration check is within a stated tolerance, then it should be left alone. The operator is more likely to make things worse by trying to adjust it. Only if it is out of calibration should the operator adjust the calibration, and only after inspecting the system to make sure everything is OK (e.g. nothing has come loose, etc.).

Again, I can't say how the above relates to the specific application under discussion here, but the general principles do apply in most situations.

 

In reply to M. Griffin,

Whenever there is any kind of fuel or starting or Inlet Guide Vane (IGV) problem on a GE-design heavy duty gas turbine, real or perceived, one of the first things that most sites do is "calibrate" the LVDT feedback. To one of your points, "calibration" can cover up a myriad of problems. (The second thing that's usually done if that doesn't work, sometimes the first, actually, is to change the electro-hydraulic servo-valve, and a lot of people think that if the servo-valve is replaced the LVDT feedback has to be recalibrated, and that's not true.)

As you stated, about the only time it's really necessary to recalibrate LVDT feedback is when something has been which might change the physical stroke of the device, such as a refurbishment of the device, or a repair procedure or if the device has had problems or the LVDT has been replaced. But most sites "calibrate" when there is a problem (real or perceived) and "calibrate" during every maintenance outage, whether the feedback needs calibration, and without ever checking to see if the feedback needs calibration. LVDT Calibration is put on the outage schedule so it must be done, whether it needs doing or not. (Most valves only get refurbished or repaired once every twenty years or so, except the Woodward valves being used these days.)

To another of your points, calibration of LVDT feedback cannot be done locally, at the device where the LVDTs are mounted.

And, to another of your points, most people never actually measure physical position when calibrating LVDT feedback, just assuming that minimum mechanical stop (fully closed) is 0% stroke and maximum mechanical stop (fully open) is 100% stroke. And, for the majority of GE-design, non-F-class gas turbines (those without DLN combustion systems), that's just not the case. But, people don't know where to find the information to properly verify the LVDT calibration or how to properly calibrate the LVDT feedback.

They just "calibrate" without ever checking the LVDT calibration before they "calibrate" to see if it needs calibration. I've never even seen anyone compare the results of the calibration being done to the last calibration results to see if there is any difference, and if there were, I don't know what would be done. The numbers just blindly get put into the I/O Configurator (in this case), and downloaded, and the processors get rebooted.

And most of the time the only check that's done is to manually stroke (position) the device to some position with a quick check to see if the LVDT feedback is approximately equal to the reference, and if it is, then calibration is deemed successful! No measurement of actual position is ever done at most of the sites. And if feedback doesn't match reference then people start changing regulator gains and null bias current values.

You're absolutely right: LVDT feedback almost never changes on its own. And GE-design heavy duty gas turbines generally have two redundant LVDTs, so if one drifts, it should be alarmed and annunciated. (The prevailing LVDT failure mode is to produce no feedback, so the position feedback would be less than zero. The Speedtronic panel uses the high-select of the two feedback signals, so that if the device fails in the predominant mode, the turbine will continue to run. And if for some reason the feedback goes high then the regulator output will try to drive the device closed which would be the safe direction to drive the device if the feedback is greater than the reference. Reliability and availability; it's the name of the Speedtronic game.)

Explaining how to calibrate SRV and GCV LVDT feedback is very complicated even if you can point and demonstrate how to do it right there on site. Written procedures are difficult to read and produce without being able to include charts and graphs and photos (which is why it hasn't been done here).

There is some closed-end over travel on both of these valves on the majority of non-F-class gas turbines with conventional combustors, and that can make calibration very difficult if it's not clearly understood and eliminated during the calibration process.

Measuring IGV position is even more difficult, as it requires a specially machined ruler in a machinist's protractor and crawling into the compressor bellmouth and measuring the angles; it's not usually done on most sites because they don't have the proper equipment and don't know how to use it. There is usually a pointer on the side of the compressor case, but it's not very accurate, especially since it's usually in a great place to be used as a step when climbing around the turbine, and it's never adjusted to accurately indicate position; never. And most sites just assume that minimum mechanical stroke is 34 DeGrees Angle (DGA), and full open mechanical stroke is 84 or 86 DGA. And that's just simply not true. And the only verification that might ever be done is to compare the feedback to the reference, and it's close, the calibration is deemed a success.

The really amazing thing is that most turbines can be made to run with these incorrect calibrations, and they run pretty well, too. They might not start easily, and they may have trouble shutting down, and they might not make rated power, and in some cases they make more power than they should, but when they are at rated speed and producing torque they run pretty darned well.

So, obviously this isn't really rocket science because these turbines do run. And it's a testament to the design of the turbines and control systems that the turbines do run and are still protected against major catastrophic failures (exhaust over temperature; overspeed; etc.).

But I'd venture to say that there are a lot of units which aren't making rated power, and a lot of units that are experiencing premature hot gas path failures and decreased hot gas path parts life because of inaccurately calibrated IGV LVDTs, primarily. Also, starting problems (including high exhaust temperatures, high exhaust temperature spreads, flickering flame indications) can be traced to poorly calibrated GCV LVDTs.

Most of what is done at the sites is all by trial and error and tribal knowledge. Even most of the OEM field services people have had very poor training with these older turbines and their gas valves. And there is almost no documentation on how to do this anywhere, not by GE nor any of its packagers. It's really mind-boggling. And is the cause of all of this tribal knowledge and confusion.

The basics of calibrating LVDT feedback are the same for any LVDT: it's all about making sure that the LVDT feedback is very close to the actual device position. But, most sites don't know what the actual device positions are, should be, or where to find them. And they don't take the time to actually measure the feedback. There is this "feeling" that the AutoCalibrate feature of the Speedtronic panel knows about the particulars of every device with LVDTs, but it just isn't so.

Every time I've written a procedure for a site, it gets shared with other sites where it's not directly applicable and then people say, "It's no good! It doesn't work!" Instead of gleaning the information that's applicable to their site and modifying or adding the information that makes the procedure appropriate to their site and producing their own procedure, they just wad it up and throw it in the round file, and go back to doing what they've always done.

A written procedure is very long if it properly details every step, with explanations so that if something goes wrong with some portion of the procedure that people can understand what was supposed to happen and actually troubleshoot and correct the situation. And, most people just don't want to read such a long procedure, preferring instead to do things the way they've always done them. They just want a step-by-step "cookbook" procedure that assumes every time it's used there will never be a problem. Keep it short and simple. But the problem is it's not a short procedure, even if it is pretty simple once it's understood.

But understanding it requires a working understanding of piping schematics (P&IDs: (Lube Oil, Trip Oil, Gas Fuel, Liquid Fuel, Hydraulic Supply, IGV), how motor starters are controlled, which signals to force and when and why, servo-valve basics including null bias current settings, what Diagnostic- and Process Alarms to be ignored and which ones to pay attention to (which is pretty inexcusable that there be any erroneous alarms at all, but there are).

So, it should be evident that LVDT calibration on a GE-design heavy duty gas turbine is not without it's idiosyncracies and it's issues, but it doesn't relieve anyone from the basic principle which is actual position versus position feedback. Not position feedback versus reference. But, back to your point: because people aren't out at the device when calibrating, they can only see reference versus position feedback. Which must mean that actual position isn't important.

But, these turbines do run and do produce power. How well they run, how much power they can produce, and how long their hot gas path parts last are all another matter.

 

Thanks for all for detailed reply.

G.Rajesh