STG L-0 turbine blade failure

[EPINT]

I have a 130 MW STG in a CCGT plant application that experienced high cycle fatigue failure of several of the L-0 buckets in less than 8700 hours of operation. OEM is stating that 4 separate events of low vacuum opertion caused this failure (a total of 21 minutes in duration), The last low vacuum event occured 2 weeks before the failure (over an 11 minute period of time). Profilometry inspection of the failure surface indicated that the crack growth during this 11 minute period of time was insignificant. We did a whole battery of metalurgical tests and examinations that still support HCF.

One item that we found was that there was more than 20 % variation in the area check (measured at 3 points from the radius) of the L-0 diaphragm and as much as a 12 % variation in the area check of the L-1 diaphragm. We are thinking that the poor area check control by the manufacturer could have created enough pulsations to excite the L-0 blades, considering that the blade design only has a 5.8% margin at the 1X natural frequency. It might be important to note that this is a 60 Hz application with a STG specification requirement for the STG to operate at 60 Hz +/-5% without limitation.

Has any other user out there been able to correlate Area Check variation in stationary diaphragms with resonance excitation of downstream blades? What levels of acceptable area variations are user seeing in STG diaphragms? What margin of safety off of the 1X and 2X natural frequency is typically being seen on new LP blade designs?

Thanks

 

[byrdj]

suggest contacting consultant engineer.

 

[rmw]

How low was your vacuum, and how low was this below design vacuum pressure?

And, how did you get this vacuum excursion?

We must live at opposite ends of the earth. Where I live no one is troubled with low vacuum right about this time of year.

rmw

 

[EPINT]

The total period of time was 21 minutes. The alarm level for the vacuum is 0.2 Bar and the unit was at the trip point of 0.3 bars for a total period of 21 minutes cumulated over 4 separate events (the last event was for 11 minutes which happened 2 weeks prior to the actual failure). Design vacuum is 0.1 bar. Excursions were caused by trip of an MCC that was providing power to the MOV's used on the steam bypass from the HRSG's. The hogger did not come on with the increased backpressure and the STG trip did not occur either, even though is has 2 out of 3 logic and was functionally tested during commissioning.


Plant is located in the Amazon, so it is generally hot all year, plus this is an islanded grid, so there are a lot of issues with power system impacts that always hammer the plant.

 

[rmw]

Does "hammering the plant" mean frequency variations that approach or exceed your 5% limitation?

rmw

 

[jsteverm]

Correlating a diaphragm area check to a component of rotating bucket stimulus will be very inexact. The area check should be done at the throat of the partitions and it is the trailing edge geometry that will have a greater effect. That being said, I have seen large (greater than 50%) variations in trailing edge openning operating successfully.

You mention that crack growth during one event was insignificant. Almost any crack will have a major effect on the response of a bucket surface condition has a major effect on endurance limit.

Vacuum excursions are an easy target for the manufacturer and can cause L-0 failures but design and operation of condenser dumps and low load transients should also be investigated.

 

[abeltio]

is this a double-flow unit?
i.e. did both L-0 stages fail? or just one?

usually the clearance between the diaphragm and the L-0 is pretty high... so there is a lot of steam mixing downstream of the diaphragm.

nevertheless... you could model the situation with CFD and see if there is a significant excitation due to the uneven area.


saludos.
a.

 

[tcs764]

Variation in nozzle geometry, trailing edge thickness and axial gap are important factors in shorter higher frequency blading and can lead to unacceptable blade vibration. In particular for the blading you describe it is important to look for diaphragm half joint discontinuities and possible assembly problems as these can give rise to significant low order excitation.

Low vacuum running has also been attributed to causing failures in this case poor design of exhaust annulus can be an important factor.

Regarding design margins the blading characteristics should have been accurately established and ideally be clear of lower order engine orders over a speed range of +/-6% of design running speed (fixed speed turbine).

As always it can be difficult to establish the specific contributing mechanism of failure however detailed examination of the failed blading can give useful information in establishing the nature and mode of failure.

 

[RhodriEdwards]

Suprised that the OEM allows +- 5% in frequency variation. On 50Hz machines I've seen limits of 2%. I used to work for an OEM and was involved in the repair and investigation of a 660MW machine that had L-0 blade failure. Vacuum was suspect on that unit as well but investigation found that poor operation and poor steam chemistry had caused the crack. Following rebuild blade tip monitors were fitted to investigate non sysnchronous vibration. Diaphraghm effects will cause synchronous vibration. Buffeting by low vacuum causes non synchronous vibration.
You are actually quite lucky the entire unit wasn't trashed.

Cheers

 

[EPINT]

The entire turbine was trashed, except for the HP case! Fortunately the stator was okay. We rewound the field. This is an interesting case, as the OEM allowed the EPC contractor to commission at low loads with no restrictions. During our subsequent failure meetings, the OEM put new restrictions on the low load operation and vacuum curves and specifically added an avoidance area where they indicate that there is a 300 minute life to the L-O blades. Quite intersting, as the EPC contractor was commissioning in this restricted area. There were no warnings in the operations manual and the OEM later indicated that they did not think that it was necessary to provide this level of information initially, as they had assumed that the turbine would operate at higher loads. This plant is in a remote location in an islanded grid in a 3rd world country, this is why we specifically specified +/- 5% on Hz. It is also a system controller requirement.

 

[RhodriEdwards]

It sounds like your owner's engineer or insurance company will have a good case against the OEM/EPC contractor.

I'd still try and press your EPC contractor to fit blade vibration monitoring to ensure the blades are operating in a safe zone (load+vacuum wise). This is quite different from having blades designed for operation at +-5% frequency. The former is a function of the strength in the blade/root design, the latter is a function of how the blade is 'tuned' to avoid certain natural frequencies.

I'd also make sure you're steam chemistry is faultless because the OEM metalurgists will try and prove SCC or poor steam chemistry played a part in the failure.

 

[abeltio]

figure out what additional maintenance/parts will the new restriction require as compared with the original design - that is what you can go for against the OEM.

by all means chemistry MUST be checked and DOCUMENTED...

but if the OEM tries to blame it on the chemistry...

then the NEW limitation on low load operation should not apply, right?
the OEM should have no objections to guarantee, provided the chemistry is within specification, that there are no problems at any load.

also... if chemistry was the problem... there should be indications elsewhere and not only in the L-0.

cheers





saludos.
a.

 

[metengr]

If this was confirmed a high cycle fatigue failure in the L-0 blade(s) as originally stated in the post (with no mention of corrosion pitting on the surface of the blades), water cycle chemistry would not be a root cause of failure.

 

[EPINT]

This was clearly HCF, there were no corrosion issues, but from low load operation, there was a recirculation from the casing sprays that initiated some erosion at the blade root trailing edge. During the remanufacturing the casing sprays were moved back and the angle adjsuted to reduce the potential for recirculation. Additionally, we requried a 100% natural frequency testing and evaluation. These blades are envelope forged and as part of the tuning in the origional turbine, they only tuned NF to the limit, which resulted in a skewed distribution on the high end of the 2x harmonic. The new turbine has more of a gaussian distribution in the NF tuning. A few other things that we required was a much higher tolerance on the LP diaphragm areas. The origional OEM tolerance was +/- 6% on spot checks of only 16 out of 68 areas. We got them to accept +/- 4% on a 100% area check. As we are getting ready to recommission this unit in the next week, I am hopeful that it will be much better, as there is less potential for stationary blade generated pulsations, there is better adherance to avoiding the 2x NF range, the blade weights are more uniform and we also required the OEM to go from Grade 2 equivalent to Grade 7 fasteners on bearing caps and other bolts. We also implemented control protection that will alarm and trip the unit (with a timer) in the event that the unit is operating in the newly identified areas of lower load operation and condenser vacuum where the L-0 blade life is severely limited. This should have been there in the origional installation, howeer the OEM had failed to provide detailed information on the low load restrictions (in the range where they aparently knew that fluttering could occur). We mandated many improvements and controls to help avoid premature failure. Now, we just have to complete our dispute resolution.