LP turbine blade root cracking modal testing

[Inthevalley]

We recently discovered cracking in our LP turbine last stage blades (L-0). These blades have a finger attachement system. From a previous unit, we have seen that it is possible that these fingers can crack off - although this is after many years of faithful operation. Unfortunately, due to access, we can only inspect the exhaust finger using ultrasonics in-situ (profile issue). We thus thought about modal testing. Based on the FEA models created, there is not a big shift in blade frequency if the fingers crack off - theoretically 2-3 Hz for each additional finger(natural frequency around 110 Hz on a 50 Hz machine). My worry was that this may be within the tolerance resulting from manufacture and blade wear over time. I did not believe we can get sufficient differentiation on blade frequency to clearly identify cracked versus non craked blades.
On the last outage, we went in and did modal testing on the blades anyway. As the models predicted, the scatter in natural frequency was not wide enough to say anything about balde cracking based on the natural frequency. What we also saw was that on some of the blades there was a second natural frequency emerging around the 250 to 260 Hz mark. We also went in with ultrasound on the exit finger and found a few blades of concern. It seemed as though there was a fair correlation between the blades showing a 250-260 Hz peak as well, and the blades with the exhaust finger showing a crack indication. To confuse matters, we also saw a number of blades with a 250-260 Hz second peak but no cracking on the finger we could see with ultrasonics. It is always possible that one of the internal fingers could be cracked...
Now for the question - is there any experience out there using modal testing to identify the extent of blade root cracking on an LP turbine?

 

[cincibcats]

Valley,

There has been a lot of work in the modal testing of turbine blades for defect detection. The most sucessful methods I know of used a change in the damping from a good blade to a bad blade rather than the natural frequency. However these methods also used microphones instead of accelerometers to increase the accuracy of the damping calculation by not adding additional damping by placing an accelerometer on the blade itself.

 

[GregLocock]

The accelerometer does not add damping. A handy trick to evaluate the approximate effect of mass laoding by an accleerometer is to measure the response with and without a second accelerometer installed alongside. This is not definitive, but it does give a good idea of whether you need to consider a non-contacting measurment system.

So far as the crcked blades go, I doubt that any reasonable crack size would be detectable by frequency, but the damping idea may work IF the mounting of the blade to the disk is sufficietly rigid -any looseness or compliance will add relatively large amounts of damping, masking the change caused by the crack.



Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[cincibcats]

Greg,

Have you ever suspended a simple beam from fishing line or dental floss and then struck it to see how long it rings. Then add a single accelerometer to the beams and compare the ring down times. The accelerometer certainly affects the system.

 

[GregLocock]

No I haven't, but what you say sounds reasonable. Perhaps I should have said 'for typical structures with both structural and interface type damping, appropriately sized and mounted accelerometers do not add appreciable damping'.

I'll get the lawyers to check it...





Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.