Hello my dear colleagues!

We have a GE Turbine Machine LM6000 PF Plus since 2019. Unfortunately, we had a strange event that broke all the blades on HPC from 3rd stage to 14th stage. Here is some review of the event.

01 Nov. CT was online with 48 MW in normal conditions.
02 Nov. CT starts to decrease load from 48MW to 45MW, the operator check parameter and notice that the CT increase automatically the load, but the vibrations starts to increase a little bit without activate alarm limits. in the afternoon, the field operator sees in the exhaust a lot of fire after an explosion, but the turbine still Online but this time with 38 MW. Finally, after the explosion and 1 minute without alarms, it was activated the turbine trip condition.

An borescope inspection shows a completely destruction in all the baldes but the thing is that the control didn't show any kind of alarm and it was working 100% operational.

27800EOH
Aero derivate Turbine.


I need your kindly help with ideas for a first RCA. I supposed that the LPC stage lose some a particle from the blade and destroyed everything or maybe a huge stall event occurs or maybe bleed valve deficiency, but it was a suppose.

Thank you for your comments.

 

PLease Help Me with your kindly comments

 

I strongly suggest three things:

First, read your original post; it's a little confusing. (I fully realize English is probably NOT your native language, but there is some more detail required and some grammar issues which make it difficult to follow and understand.) It's also extremely difficult to understand why operators and their supervisor(s) didn't shut down the machine when the outside operator saw smoke and flames in the exhaust.

What fuel was the machine burning at the time of the smoke and flames?

Did an operator reduce the load on the machine, or was that a result of some external/automatic control system reducing load for (what?) reason?

If the machine lost 10MW why did the operations department not do some investigation, especially if smoke and flames were noted in the exhaust?

What was the frequency of the site/grid (if the machine is synchronized to a grid with other machines) during the entire period of the operation you have covered in your original post--was the frequency stable, or unstable, and if it was unstable was it wildly unstable or mildly unstable and what were the minimum and maximum frequencies?

Does the machine exhaust into an HRSG (a boiler to produce steam using the hot gas turbine exhaust) or is it a simple-cycle machine that exhausts directly to atmosphere?

So you see, the original post is unclear and there is a lot of information which would be helpful that is missing. I have seen an LM run for several hours after losing turbine blades during operation that reduced load by 50%. The site decided to shut the machine down for an off-line water wash, but when they tried to re-start it the vibration was too high, and that's when the missing blades were discovered. And the machine was shut down, not tripped by the operators, though it did coast to a stop much faster than normal....

Second, tell us what turbine control system is in use for the machine. And, if there's some other external control system which sends lead control signal(s) to the turbine control system, tell us a little about that (what is the setpoint (frequency;

Third, if you have any historical operating data from the time prior to the shutdown--INCLUDING alarms (if the turbine control system is a Mark*, include both Process Alarms AND Diagnostic Alarms).

Any RCA (first, second or third) MUST use actionable data--actual operating data and values and the alarms, even if no one feels the alarms were relevant.

That's about all I can add. I'm not a fan of LMs and can't really offer too much in the way of any assistance with any degree of certainty. By reviewing running/operating data from the time prior to the event--AND the alarms which were annunciated during the period (and if you have a list of active alarms PRIOR to the event that would also be very helpful) it might be possible to provide some suggestions/assistance.

Help us to help you by supplying the requested information. There is too much unknown and too many possibilities to make even an educated guess as to what happened. But, it is hoped that there will be an operational review of the event that leads to changes in the operation. Just because a turbine control system can detect most (but not ALL) faults and trip or shut a machine down automatically DOES NOT mean that's a substitute for conscious operators paying attention to operating conditions and taking appropriate action to protect the machine AND plant personnel. If there was smoke and flames coming from the exhaust and the machine was allowed to continue to run because it wasn't automatically tripped by the turbine control system then the operations department (operators AND their supervisor(s)) don't understand basic machine operation and know when to make a decision to "over-ride" the turbine control system and shut the machine down. It's entirely possible--from the information provided--that the catastrophic damage suffered by the machine was avoidable if someone had understood the situation and made a decision to shut the machine down (or manually trip it) to protect the machine AND plant personnel.

Finally, this is a controls/automation forum. It's not really a mechanical forum, though we do discuss mechanical issues as they relate to controls/automation from time to time.

 

Thank you very much for your answer. I apologize in advance for my English which is not my main language since I am in Colombia; however, I try to do my best.
Regarding your feedback, I make the following clarifications:
The damaged condition of the machine was not reflected in the DCS until the explosion of the machine was felt. The field personnel saw the instantaneous fire coming out of the turbine exhaust; however, the control system was still 100% operational with no alarm or trip values displayed. The sequence of events prior to the engine catastrophe shows an increase in vibration (not reaching alarm limits) and a gradual load shedding. All log information and sequence of events was sent to GE and is under analysis; however, within our own analysis I have been able to detect 24 hours before the event, an abnormal behavior in which the vibrations do not correspond to the actual load and from there, it is my point of reference.

After reviewing all the information both operational and in the control system as well as in the ST Toolbox, I have so far two scenarios.

1, detachment of material from the low stage which subsequently caused the damage.
2. Uncontrolled stall effect (i.e. the blowoff valve was not able to evacuate the air during the last output of the machine) which caused the mechanical failure which gradually increased.

Relevant points: it is a single cycle, 60hZ, 50MW.

 

This is more information, but far from complete.

If the machine uses ToolboxST, then unless the GE Mark VIe HMI is unique (which is possible these days… unfortunately) there should be both live data files and Trip History display files available on the HMI. Even if the DCS has historical data archival and retrieval capabilities the data may not have been at a fast enough capture rate to be really useful.

As I wrote before, an LM (which was severely abused by the operations and maintenance departments) experienced a similar situation. Vibrations increased but not to the trip levels, and the machine steadily lost half its output, this over a six-hour period if I recall correctly. The night operator made a decision to shut the machine down and perform an off-line water wash (the machine had experienced previous episodes of compressor fouling, each one was (more or less) resolved by an off-line water wash). To most (poorly trained) operators every problem is similar to previous problems, hence the decision to wash the axial compressor.

When the last rinse was com, they attempted to start the machine—but it had EXTREMELY HIGH vibrations just during cranking. Tens of thousands of dollars of compressor and turbine blades that had been literally ground to small pieces were found in the gas turbine exhaust. I had a particularly enjoyable time throwing shiny bits of expensive metal into the air and listening to them tinkle as the fell to the floor of the exhaust plenum.

If you don’t want to share information, and that’s your prerogative, then there’s nothing else to add here. Let the GE review run it’s course.

“Learn from history, or be doomed to repeat it.”

¡Adios!

Buena suerte.

 

Nothing you could receive from any World Wide Web forum is admissible in any court of law. It may provide suggestions, but YOU will ultimately have to provide actionable data for review and analysis. And that’s really and truly beyond the scope of this controls/automation forum. AGAIN, trained and experienced operators would have sensed something was amiss LONG BEFORE the machine cratered (crashed) and taken appropriate action to protect the machine and personnel.

 

Thank you, my friend, for your kindly help. I have a question... in the past when you saw the event that you mencionated : "
Vibrations increased but not to the trip levels, and the machine steadily lost half its output, this over a six-hour period if I recall correctly. The night operator made a decision to shut the machine down and perform an off-line water wash (the machine had experienced previous episodes of compressor fouling, each one was (more or less) resolved by an off-line water wash). To most (poorly trained) operators every problem is similar to previous problems, hence the decision to wash the axial compressor.
When the last rinse was com, they attempted to start the machine—but it had EXTREMELY HIGH vibrations just during cranking."

Which was the main cause of vibrations increased during the machine running, did GE tell you a detailed RCA?

 

Broken turbine blades. The more that broke the more other blades were broken. And the machine can't make rated power with broken blades, even if the fuel flow is the same.

GE's detailed RCA was a bill for rebuilding the machine. In the end, that's what has to happen--if you want to generate power/money again any time soon.

GE has a far, far better chance of explaining exactly what happened. They have a LOT more experience with these issues.

Especially when they can see the damage AND have actionable data to look at.

 

Hello my dear colleagues!

We have a GE Turbine Machine LM6000 PF Plus since 2019. Unfortunately, we had a strange event that broke all the blades on HPC from 3rd stage to 14th stage. Here is some review of the event.

01 Nov. CT was online with 48 MW in normal conditions.
02 Nov. CT starts to decrease load from 48MW to 45MW, the operator check parameter and notice that the CT increase automatically the load, but the vibrations starts to increase a little bit without activate alarm limits. in the afternoon, the field operator sees in the exhaust a lot of fire after an explosion, but the turbine still Online but this time with 38 MW. Finally, after the explosion and 1 minute without alarms, it was activated the turbine trip condition.

An borescope inspection shows a completely destruction in all the baldes but the thing is that the control didn't show any kind of alarm and it was working 100% operational.

27800EOH
Aero derivate Turbine.


I need your kindly help with ideas for a first RCA. I supposed that the LPC stage lose some a particle from the blade and destroyed everything or maybe a huge stall event occurs or maybe bleed valve deficiency, but it was a suppose.

Thank you for your comments.

Hola Diego.
Me gustaría hablar contigo sobre la falla que tuvieron. Acá tuvimos una falla similar y podríamos compartir experiencia para encontrar la Causa Raíz. Mi correo es [email protected]

 

I have a few comments.

First, you mention a blowoff valve. A PF1 does not typically have a blowoff valve but they typically have ST8 and CDP bleed valves. Does this engine have a dedicated blowoff valve or are you referring to either the ST8 or CDP valve?

Second, where exactly did they see the flames come from the engine? When the engine is installed and running in a package you typically cannot see the exhaust as it exits the engine. Was there a camera installed inside the package? Were the flames that they saw actually at the exit of the engine or at the exit of the exhaust stack? The exit of the exhaust stack would be quite a distance from the combustor.

Third, I have seen a number of engines that have been severely damaged but still managed to run. So just because an engine can continue to run doesn't mean that there hasn't been significant damage. Losing 10MW from peak power after seeing flames visible in the exhaust is an obvious red flag for severe engine damage.

Fourth, I am a little surprised that you can damage that many blades and not trip on vibes. Most of the time, but not always, that much damage results in the engine tripping on vibes.

Fifth, I am also surprised that an "explosion" of sufficient magnitude to result in flames out the back of the engine (or stack) would not be enough to trigger the stall detection logic inside the control.

Sixth, I agree with WTF? Data of the event, particularly if you have fast data, would be invaluable in determining the sequence of events. If you don't have data, then you have to examine the hardware (even if the hardware is in pieces) to try to reconstruct the event. It is much harder to reconstruct without data.