FSR Decreases During Acceleration

Y

Thread Starter

Yasbine

Hello everyone,

This is my first post in control.com. I'll try to be as concise as possible. We are working on MS-3002J gas turbine operating on a Mark6e control system.

After a hot gas path inspection, we started the machine and everything was ok till 51% TNH (acceleration phase) when the turbine trips on "L60BOG_alm HP turbine fails to accelerate". We checked the triplog and noticed that the TNH dropped from 51% to less then 46%. In fact, when we took a look on the logic we found that if the TNH decreases for more than 5% of the last Max TNH value, the turbine trips. Then we checked the FSR. It dropped from 23% to 12%. We thought that there might be a shift in the FSR from fsracc to fsrmin but nothing explains this hypothesis.

Now the turbine is running after we decided to force the logic (production requirement) but if the turbine trips again, we're likely to cope with the same problem.

So my question is, what would cause the FSR to drop?

One last thing we have noticed is that this drop in the FSR happens just after the LP turbine start.

NB:The second stage nozzles are in the full open position 15°angle
The cranking motor is disengaged.

if any other details are needed just tell me.

Thank you in advance.
 
It kind of sounds like your LP rotor is breaking away very rapidly and triggering FSR Acceleration control. Is there any load on the LP turbine during start-up?
 
Yasbine,

I agree with glenmorangie--it sounds like the LP shaft is accelerating too quickly.

The general mode of operation of a GE-design two-shaft heavy duty gas turbine is that--when operating at rated speed (LP speed, that is) that fuel controls LP shaft speed, and the second-stage nozzles control the HP shaft speed. It would be necessary to see the application code running in the turbine control panel to know exactly when the switchover from HP speed control to LP speed control occurs, or if there is some limit from LP acceleration rate that can affect FSR while the HP is still accelerating to minimum operating speed.

I have seen MANY start-ups after maintenance outages when the sequence of operation used after the start-up was not the same as before the start-up. In other words, is it possible that somehow the centrifugal compressor valve line-up (positioning) was different during the starts after the outage than before? To glenmorangie's point, could the compressor have been bypassed for some reason during this re-start, where prior to the outage it was not bypassed and had some load during initial starting?

I have two questions, though. What signal did you force? And if it was the bog down logic signal, did the HP speed actually decrease during starting but then started to increase at some point? Because it's kind of hard to imagine how a machine that had FSR decrease to at or near minimum causing the HP shaft speed to decelerate more than 5% would complete acceleration just by forcing the bog down signal.

Do you have any data from the start when the signal was forced? Did you use Trend Recorder during any start attempt to gather more data than is available from the Trip History?

My second question is: What does NB mean?

Thanks!
 
Thanks Glenmorangie and CSA,

Actually, In the second attempt, we forced the L60BOG signal and made a trend of the FSR and TNH (Unfortunately, We didn't trend the FSRACL, FSRSU, FSRACC to see which FSR is selected.) and you are absolutely right, the FSR decreases from 23% to 12% and then increases again. Otherwise, as you mentioned, it wouldn't be possible to accelerate the HP shaft again.

Nothing was altered in the process (the compressor haven't been by-passed) only the turbine was concerned by the maintenance outage. The only thing which could change is the pressurization of the compressor at start-up and if this one is lower then it should (I guess that's what causes the load to decrease), the LP shaft would accelerate too quickly leading to a decrease in FSR.

NB stands for Nota Bene: a latin word which is also used in english to make a reader pay attention to more or less important information.
 
Yasbine,

Are you saying that during the second attempt when L60BOG was forced to zero that the FSR did decrease from 32% to 12% and TNH did decrease from approximately 51% to something much less (more than 5% less) during the start attempt?

Did you also trend TNL to see how fast it was increasing? Do you have other trends from more normal starts to compare the rate of acceleration of the LP shaft to the current rate of acceleration of the LP shaft? (I'm guessing the LP shaft acceleration rate is TNLA and the LP shaft acceleration reference is TNLAR--but those are just SWAGS (Scientific Wild-Arsed Guesses), for future trending purposes.)

The reason "bog down" protection is in place is that it's NOT desirable to be continuing to burn fuel in a heavy duty gas turbine during acceleration of the rate of acceleration is negative. You said the starting means is disengaged, and if the speed gets too low then it's likely the turbine will not be self-sustaining and will be damaged by combustion at very low speed with no torque assist. It's actually a pretty important protection.

Yes; I would say if the pressure on the centrifugal compressor was less than in previous starts then the LP shaft (which drives the compressor) would accelerate faster, perhaps MUCH faster.

This problem will likely be easily explained/understood once it's sorted out. But, for now, it's unusual.

Thanks for the explanation; that's what I though it was, but wasn't sure. It's somewhat unusual to see in use in a power plant/compressor context--at least for me. I always want to be sure I understand abbreviations and acronyms--each of which can have many meanings.
 
CSA,

during the second attempt the FSR decreases from 23% (not 32%) to 12% and TNH decreases by 5.2% and this last around 3 seconds then the FSR and TNH increase again.

We don't have other trends of normal starts to compare with and it seems that we have missed to trend many signals ( many many mistakes that we made) but we have the TNL from the triplog so we can calculate the acceleration rate ( can't do it for the moment because I'm off work).

You're right the LP shaft acceleration rate is TNLA and the LP shaft acceleration reference is FSKACLR=1%/s so I suppose that the TNLA was greater than 1%/s which causes the FSR to decrease significantly.

Though, I have few questions. Why would the turbine be damaged by combustion at very low speed? wouldn't it trip by high exhaust temperature instead? and what do you mean by very low speed?

thanks
 
yasbine,

Thanks for the clarification.

There isn't much air flow at part speed, and if the unit is decelerating and there's no torque assist from the starting means and the speed drops below the point at which the turbine can sustain it's speed then the liners, transition pieces and first stage nozzles can overheat. As for tripping on high exhaust temperature, that could happen but relying on that to protect the turbine isn't a good idea. I've seen a LOT of turbines kind of "stall" at around 60-70% speed because the exhaust temperature was high enough to cause FSR to be limited by exhaust temperature control. Running at part speed with the IGVs closed and a maximum exhaust temperature is NOT good for the turbine. High exhaust temperature means even higher temperatures in the combustor and first stage nozzles/turbines.

Recently one Frame 5 machines which was "stalling" at 60-70% speed for 10 minutes or so (sometimes longer) at maximum exhaust temperature for a couple of years had a catastrophic failure of the second stage buckets. The buckets weren't that old (less than three years), were OEM (which doesn't mean they're any better than any other vendor's, but....) and there wasn't any other cause which could be pinpointed to be the cause of the failure. BUT, there were about 12 starts in the previous two years where the unit stalled because of reaching exhaust temperature during acceleration. And, there was some issues with the torque converter that caused the unit not to accelerate at all after flame was established and after waiting several minutes the unit was tripped.

So, I'm just saying--high temperatures at low air flows and especially if the unit is decelerating when it should be accelerating can't be good for the unit.

By 'very low speed' I mean less than approximately 45-50% without any torque assist from the starting means. The HP shaft should be accelerating and if it's decelerating that's not good. I understand the thought process about relying on exhaust overtemperature to trip the machine if it gets high enough, but in your case the FSR was already very low so it wouldn't seem the exhaust temperature would be high--unless the air flow just got so low because of deceleration. And, again, the temperatures in the combustor and first stage are usually higher than exhaust temperature, even during starting. And even if the exhaust temperature isn't high enough to trip, the internal temperatures are certainly high (before the exhaust area).
 
yasbine,

the FSR decreases to the minimum FSR during the start-up which is about 12%.
if both GCV, and SRV were steady before, then check your LP speed sensors, missed fuses, or bad connection.
 
belmehda,

Why does FSR decrease to minimum during starting?

It's been a long time since I've worked on a two-shaft, and the sequencing/application code might have changed, but if I recall correctly the transition from HP speed control to LP speed control occurred when the LP shaft broke away from zero speed, but as long as the acceleration wasn't excessive (too fast) the transition was bumpless (meaning there wasn't a big decrease or increase in FSR).

I can understand checking the speed pick-ups (LP shaft in this case), but the unit started with the same behaviour on the next start (when the bog-down protection was forced out) and eventually made it to rated speed (both HP and LP).

Anyway, this is a good example of why making trends of GOOD start-ups can be just as important as making trends of bad ones. With nothing to compare against, sometimes it's very difficult to pinpoint what the problem is. And, it also helps to get people familiar with Trend Recorder--which, while it's pretty intuitive, can be problematic. Technicians, and even operators, should be familiar with Trend Recorder, and should practice setting up trends and getting accustomed to the settings. Also, analyzing trends with Trend Recorder has some VERY GREAT features--such as being able to change pen widths, to zoom in or out, to have multiple trends saved in a file and to switch between them, and my favourite: being able to hide signals that aren't important to the analysis. These are all really great and under-utilized features of Trend Recorder, and using them is the only way to get familiar and comfortable with them so that when a problems arises one can quickly create a trend and then quickly analyze it. And practice and familiarization is about the best thing--and it can be done any time. It runs in the background so it doesn't affect the operators, it can be started and left running, and it DOESN'T affect operation at all--meaning it can't cause a trip or shutdown or runback. It can even be triggered to start on some event (though I've never tried or checked to see if it can be triggered to stop at some point...).

I don't know if I've asked (I've gotten so "accustomed" to getting no or poor responses), but what Diagnostic Alarms are present during starting, and what are annunciated during starting? Diagnostic Alarms, and even Process Alarms--contrary to popular (and false) belief really can be helpful when troubleshooting. Seriously. It's really unfortunate that commissioning personnel don't work hard on eliminating nuisance alarms (both Process and Diagnostic) so that when an alarm is annunciated it is to be taken seriously--and that there are so MANY alarms, most of which are not documented and without training and familiarization operators and technicians can't understand. One of the good things about the Mark VI and Mark VIe is that Diagnostic Alarms were reasonably well described in the System Guides for the two control systems. The descriptions can be brief and obtuse, but at least most are described.
 
Gents,

I've posted the same issue on Oct 2013 in Control.com. after maintenance outage of Frame 3 gas turbine, and during the LP over speed test, the machine reach self sustaining speed suddenly the FSR decrease to FSR min than the fuel gas control start fluctuating,

we've used View1 to track all FSR during start up, until 60% TNL, every thing was steady. then the FSR decrease to FSR min because the algorithm loses the LP speed signal. we did not cutch the change in the speed by View1, but that is the only explanation for what happen.

we checked the speed loops, and fuses were removed during the electronic over speed test simulation. and we did not return them back.

hope this experience can help, Please write back to let us know what you find.
 
Belmahda,

thank you for your contribution and sharing your experience with us.

The SRV and GCV didn't fluctuate. The GCV drops and increases only once and we noticed (from the trend recorder) that at this exact time the LP breaks away. So I kind of think that it is more likely a problem of LP acceleration as CSA and Glenmorangie said. Plus the turbine is running now without any problem of speed sensors.

One more thing about the Mark6 and Mark6e is that they have a very interesting feature which is to display a diagnostic alarm if you have a blown fuses, unhealthy thermocouples and so on (I don't know about the MarkV if it also has this feature).

CSA,

All the features of the trend recorder you describe are really useful. the only problem is that you can't keep it running in the background for a long period of time unless you have a very powerful server otherwise you can't record it for analyses.
 
Top