Fun with metallurgy ...

Status
Not open for further replies.
Joined
Dec 12, 2002
Messages
43,887
Location
'Stralia
Given that it was a while ago, and the station in question has now been closed ( :sniff: ) RIP, here's some stuff from my old days in heavy rotating equipment.

Day started with an unusual vibration, about 40% of the trip level, but a step change early in the morning, perfect step, not a gradual increase, just a vertical jump, and hold...sister bearing on that rotor (multiple rotors, High Pressure, Intermediate Pressure, and Two Low Pressures, then Generator and exciter train) suffered at the same time a drop, and phase change.

Was convinced that something had come apart in the machine, and arranged for a shutdown that day...problem is that you have to turn them at 4RPM after a trip for like 4 days to get the 1005F metal temperatures down to 300F so that you can stop turning them and applying oil to the bearings.

Came up with a modified cooling arrangement, where we would undo covers, stop the "barring" for 20-30 minutes while we lifted something off with a crane, put it back on barring until the shaft straightened a bit, and repeat.

Get maximum information on this issue as early as possible to help plan a repair.

When we got far enough into the machine, we were greeted with this

Unit7LP1June07005.jpg


The 14" 'L-2' blades had some ends missing, and the shrouding had been broken, and made contact with the stationary spill strips.

Clearly high cycle fatigue, and some interesting things happened :
* on the upstream blade, you can see the line of imprint where the blade that snapped struck the pressure face.
* given how high cycle it was, the failures (and there were more of them) all occurred at the 6:00 position...there's 2g difference between 12:00 and 6:00.

These blades are "packetted", as you can see, with a dozen blades joined with a section of "coverband". The coverband is rivetted to the blades by "tenons" in the blades that are formed from the blade material on manufacture, then the perforated cover band is fitted, and the tenons peened down to rivet form by a guy with a hammer, and a tool made from cut off cold chisel, and many many blows.

The blades are essentially 410 stainless steel, so that's another complicator.

Look at the photo, and you can see that there is a part line at the end of each 12 blade packet...that design, they are continuously coupled, and there are two "short" blades at the end of each packet, and the packets are joined with a titanium "butt strap".

Problem then is how does one repair such an arrangement.

Google Search shows that it's rotors out and down to the machine shop.

Was set a challenge by TPTB that I was to devise a method of repair that didn't include taking the rotors out (4 days of work at either end).

Can see that the blades are held to the rotor disk with a number of pins. Two "fingers" on the blades, and three disks, with a through pin. Blades are machine fitted axially, then fitted on the leading and trailing root faces to get the "pitch" and "lie" correct, so that they fit at the bottom, and the tenons at the top fit into the machined holes for riveting.

The pins are normally drilled with a machine called a Caludon.

Found an old guy, one of the original blade fitters, and we worked up a method of guide blocks where he could drill the old pins out with a right angle, OTC air powered drill, and some staged depth jigs to follow through the pin.

Pulled the two damaged blades, the two between them, and the one either side (total 6) out, and machined the rivetted blade ends, and the top section of 12 blade coverband.

Unfortunately the last of the blades on the lower part of the pic was under a titanium fish plate, so we had to take the band off the lower 12 blade packet as well...leaving a gap of 6 blades, and 18 blades which were missing the tenons.

Tenons, we had arguably the best of the best in the country (and UK and Germany by repute) TIG new blank tenon/rivetts onto the top of the old blades...the two at the top of the removed section, the two short blades, he had to be very very careful not to contaminate the titanium fish plate. While he was lying face down on felt mats, at a 45 degree angle on the blades.

He made new tenons on the existing blades, and they had to be heat treated and metallurgically proven before being hand finished down to the proper shape, and fit for the new cover bands (I found two hiding inside a piece of 3" bar...remove the bit that wasn't cover band, and there they were).

Two new sections of cover band, a titanium fish plate (synchrodestiny works there), and a guy with blacksmith like upper RHS body beat the tenons down.

All up just under two weeks repair, instead of a month...one of my highlights.

Then came the investigation and analysis...

These blades had a chequered past, right from the day that they were designed, they broke...there were many investigations around the globe in the 1970s, and the blades were "detuned" to get them away from resonance and failure.

http://www.sti-tech.com/dl/vibnfr.pdf

There's other papers that I will find and link later.

There was a really high end harmonic that made the blades and disk look like a 20 node wave washer, and in the '70s was deemed insignificant that a long time later became a problem...chloride pit on trailing edge. 100,000 hours to crack initiation 1,300hrs from crack start to failure...

Having found the problem, then it was time to plan replacement.

And refine the repairs.

The first failure took two weeks to fix, which was good.

Next one occurred while the new parts were on a boat from Newcastle, and I didn't need the repairs to go all the way like I did the first.

So had to try to come up with a way of doing the smallest amount of weld repair as I could.

Welding the cover band to replace small numbers of blades was obviously worth a look...problem was Post Weld Heat Treatment of the cover bands with the material present.

Imagine the cover band, being heated to 1500F, it grows, lengthwise, and pushes the blades out of position. The metal is at heat treatment temperature, and the blades start to squeeze the cover band in, lengthwise, which relaxes...when the whole lot cools, the coverband shrinks, and there's a tip load on all of the blades when cold...unacceptable.

So came up with this...well me and a coupe of metallurgists.
IMG_20150817_201609.jpg


It was a "cold" Inconel weld, only needed 300F pre heat, and welded "cold" needing no post weld...minimum residual stresses, but also comparatively weak, which was probably OK with no residuals from the PWHT.

If you look at the lower single blade, it had cover band cracks already, linking tenon holes...had to be stuck together without joining the blade, band, or tenons.

Welder did it, and as can be seen, it didn't crack, but the crack found another path.

Whole lot stayed together until replacement time.
 
Wow, that was a lot of work! Really interesting fix on both. It is amazing what is done in the field to make things work.
 
spasm3, that rotor weighs 30 tonnes.

The missing bits didn't even take it out of the ISO continuous running range, but the step change on #6, and phase reversal on #5 said that it needed investigation.

The welds didn't change anything.
 
Interesting, my Dad does some customer inspections for motors coming out of GE large motors in Peterborough.
I was initially surprised how balanced they have to be, down to an ounce or two if I recall, but then they are 10-20-30 tons spinning on bearings 20-25-30' apart at 3600rpm, with the rotors 6-8-10' in diameter.
If the shaft starts to flex in the wrong way, things go bad fast and the motor stops very slowly!
 
Wow. That reads as interestingly as the failure analysis of the GE CF-6 fan disk that burst over Iowa on United 232. The initiating cause of failure happened nearly 10 years before and the disk flew on 3 airplanes before it finally went. In a way this is MORE interesting because it involves a field fix, not just removal of potentially defective parts from service and replacement with new build parts.

Further confirmation that turbo machinery is a different universe than reciprocating.
 
Originally Posted By: Shannow
spasm3, that rotor weighs 30 tonnes.

The missing bits didn't even take it out of the ISO continuous running range, but the step change on #6, and phase reversal on #5 said that it needed investigation.

The welds didn't change anything.


Also I assume this is direct-drive, so it spins what? 3600 RPM? Or 3000 if the grid is 50 Hz? There's an inherent temptation for people to think "turbine" automatically means insanely high RPM, but that's not the case for power generation. With a rotor that heavy and that RPM range, I'm not surprised that you could do a lot of repair and still be in spec. Its a different set of problems than on a turbocharger or the gas-generator spool of a small turbojet or turbofan.
 
There is not much in the pictures ti give a sense of scale.

What is the aprox width of the band?
 
Originally Posted By: expat
There is not much in the pictures ti give a sense of scale.

What is the aprox width of the band?


about 40mm wide on that one...
 
Originally Posted By: Shannow
spasm3, that rotor weighs 30 tonnes.

The missing bits didn't even take it out of the ISO continuous running range, but the step change on #6, and phase reversal on #5 said that it needed investigation.

The welds didn't change anything.


WOW thats heavy! thanks for the pics! Amazing work.
 
Originally Posted By: spasm3
WOW thats heavy! thanks for the pics! Amazing work.


Fully assembled train is over 200 tonnes spinning at 3000 rpm

Here's the same rotor out of another design, for some shop work.

0416091207-00.jpg
 
Quote:
The blades are essentially 410 stainless steel, so that's another complicator.


Are they too soft or flex too much?
 
Neither really, but they have to have fatigue resistance to last 100,000 - 300,000 hours running.

The weld repairs on the blade tenons had to be ductile enough that when they were peined over to capture the cover band without going work hard and cracking/splitting.

and PWHT was 770C+ causing the coverband problems.
 
This Paper
http://turbolab.tamu.edu/proc/turboproc/T31/t31pg105.pdf

Case study 1 is these two machines
laugh.gif


They had a harmonic problem as described in the paper.

Interestingly, back in the day for model verification, they installed strain gauges on the blades, and ran the leads down the blade, and drilled a hole through the rotor, and into the rotor bore...ran the lead along the rotor, to another hole, then poked them out in the vicinity of the coupling, and formed an excitable antenna, so that they could monitor flex and resonance in real time.

That made that particular rotor "special", in that it had radial through holes in it, limiting overspeed testing on cold rotors, extra metallurgy during inspections etc.

The result was the detuning weight shown in Figure 23 of the linked paper...rivetted onto the blades.

They were replaced in 1995 with the detuning weight forged into the blade aerofoil. as such...
IMG_20151120_173907.jpg


These were the ones that I replaced.

Look at figure 21, and see that the original design had a near resonance effect at 6 through 10 nodes (like a pringle)....the detuning weights moved that.

What they DID do, which wasn't even looked at in the '80s was have a harmonic at a 20 nodal "pringle", which was deemed in the day to impart so little energy as to be insignificant in service.

We had a drought, we had some salt migrating into the condensate, and we had pitting. in the trailing edge. The L-2 is just into the wet zone at low loads.

Had some really good F.E. done, and they calculated that it was 10s of thousands of hours from a pit to a crack, then 1,300 hours from crack to snap....we could never pull the machines apart often enough to find a crack.
 
Status
Not open for further replies.
Back
Top