Knife Blade Cracking on Reverse Buckling Rupture Disks? 1

[Bambie]

Does anyone have experience with cracking of the X knife blade assembly on reverse buckling, large (>24"dia) rupture disk designs in saturated steam service?

The blade ends are fillet welded to the inside wall of the disk holder and the cracks appear parallel to and 1/4" away from the fillet weld toes. The crack at one end of each blade usually goes through wall while the crack at the other end does not, indicating a thermal fatigue mechanism.

The blades are above the disk and see ambient atmospheric conditions in the vent while the holder sees process conditions (60 psig at 306 degF). The warm-up thermal transient would rapidly heat the holder with condensation heat transfer, which would stretch the blades and probably initiate cracks. The blade thickness is constant but the blade depth increases from the holder wall to the X weld.

The holder is A105 and two blade materials have been tried; A505 4130 and ATI PH7-17 and both have cracked. The blade X assembly is heat treated to maximize hardness, then welded into the insert. Perhaps localized annealing is required at the fillet welds?

 

[EdStainless]

What about pre-heat or post weld HT?
If you do this with 17-7PH you first HT at 1400F, then cut, machine, assemble, weld.
And then you age at 1050F to achieve hardness in the 17-7.
With 4130 there would be a pre-heat and then a PWHT.

= = = = = = = = = = = = = = = = = = = =
P.E. Metallurgy, Plymouth Tube

 

[georgeverghese]

What is on the other side of this RD; on the atmospheric side? Is this a common relief stream collection header ? Could steam relief from other sources have condensed and pooled up on the LP side of this RD? If this RD is leading to a common collection header, then the RD exit piping should slope down to the collection header to prevent steam condensate pooling on the LP side of the RD.

 

[Bambie]

EdStainless,

Our Welding Engineers are repairing the blade assembly cracks with the required heat treatments as you have described. The attachment fillets are receiving localized preheat and PWHT with torches. What would you recommend? We have considered pre-loading the blades in compression with an autofrettage technique which would involve heating the insert to 300 degF and shimming the cool blades in place before welding.

georgeverghese,

The rupture disk vent is a single, 70 foot high vertical pipe with 60 feet of it inside the turbine hall, and no condensate collection has been detected in the leak-off line. Our disk is 42" dia, blades are 1/2" thick and the insert is 2" thick.

If nothing can be done to increase ductility in the area of the fillet welds without affecting blade hardness, perhaps the temperature difference could be reduced by:

1) removing insulation from the insert, the spool piece below and the vent pipe above to increase convection.
2) installing a thick, low conductivity gasket between disk and insert to slow heat ingress.
3) installing a thin, highly conductive gasket between the insert and vent flange to increase heat egress.

 

[georgeverghese]

If this is a straight vertical line exit, could polluted acidic rain have pooled up here on the LP side of the RD? I see you're from Ontario, Canada. Bitumen fields perhaps, with H2S / trace SO2 from Claus tail gas treatment units in the atmosphere ? Leak off line blocked up ?

 

[don1980]

Consider replacing this knife blade assembly with a another type of disk design - any design other than knife blade. Many regard knife blade disk assemblies as insufficiently safe. If the blades become dull (or broken as in your case) the disk may not burst when it inverts and impacts the blades. And if that happens, the blades become a mechanical support for the disk, greatly increasing its burst pressure.

 

[Bambie]

georgeverghese,

We haven't done chemical assays of the cracks, however, there didn't appear to be significant deposits on the disk, blades or insert. The leak-off line was not blocked.

don1980,

We are considering replacement with a scored forward bursting design with vacuum support.

 

[georgeverghese]

Also consider replacing troublesome RD with a buckling pin actuated RV - buckling pin can be replaced inline, and a lot more reliable.

 

[Bambie]

EdStainless,

Would you mind commenting on the proposed knife blade to insert weld procedure (attached)?

 

[Compositepro]

I suggest you put a slight curve in the knives to relieve the stresses caused by large temperature changes.

 

[Bambie]

Compositepro,

I agree, however meddling with the blade configuration may influence capacity and since the repair is being done in house without testing, we are obliged to duplicate the manufacturer's configuration.