SW Gasket Failure, why????? 3

[Giuss]

Hi all,
please help me!!!
On a 42'' #150 line a flanged joint showed a leakage, the flange connection (Flg per ASME B16.47 A, A350Lf2, Blt A320L7, gasket: Spiral Wound AISI 316/Graphite with inner AISI 316, Outer CS) was opened and the gasket appeared destroyed.
It happened two times more (3 in total)
I inspected the gasket of the 3rd failure:
It clearly seems ''Inward Buckling''.
The gasket dimensions are ok (according to ASME B.16.20).
The istallation, the torquing procedure and the bolt torque values were correct (at least the third)and witnessed by Contractor Inpector, Client Inspector, Our Inspector (we are an engineering company).
Torque values have been also verified against gasket manufacturer's suggested values, furthermore values specified by client are quite low compared to other companyies spec.
In the three failures, gasket from different manufacturers were used.
During the Hydrotest (with a temporary gasket) no leakage was observed.
The line is no subject to the vacuum.
The flanges were visually inspected, no apparent defect is present on the mating faces, no apparent leak of planarity is observed.
Many Identical Joints are present on the plant and no leakage was observed, following the same installation procedures.
One could say that failre of SW gaskets on big sizes is common, and i will agree, but why with same gaskets,flanges and bolting, identical procedures, only 1 joint fails 3 times while the remaining about 40 are all ok.
It seems that they are ''sw gasket killer flanges'' but why?
which kind of problem (not detected by a visual ispection) could these flange have to damage the gasket?
Could the roughness affect it? i bet no

Please could someone help me, basing on his own experience or on literature basis.

thank you in advance

Giuss

 

[Artisi]

Have you checked the flanges for flatness?

 

[Artisi]

Further thought - are the faces parallel to each other?

 

[Giuss]

Artisi,
yes (at least apparently)
I was told the first time the line was a little ''pulled'' then there was made an adjustement, by cutting and welding, then the flanges were aligned

Thank you

 

[Artisi]

Being from the old school of "hands-on" engineering, the first thing I would do is to check or witness this myself.

 

[Giuss]

...me too, unless the project does not involve 60000 tons of piping.....
me too, unless i was not here (in this cases i say ''i was told'')

 

[gerhardl]

Since you have tried with different gaskets and same type of flanges with gaskets works elswhere, and reapeatedly mounted (eliminating wrong mounting methode and wrongly selected sealing) the answer is obvious:

at this joint something mechanical is different.

Correct torque on bolts does not guarantee correct pressure on sealing if skewness is present.

Most likely the cause is as already mentioned flatness or parallellity, other type of unevenness, alternatively support or clamping of pipeline making movement/stresses possible which again gives uneven thightening and media trapped and pressuring on parts of sealing.

Gas or liquid bubbels forming/reforming?

Vacuum possible? (Sucking sealing in incorrect fastened parts in?)

Temperature variations?

 

[Giuss]

Gerhardl, thank you,

unfortunately as i wrote, after the first failure, a stress analysis has been carried on and the spool refabricated, now the line is not stressed;
Start-up ensured me that for no reason the line could have worked under vacuum conditions.
Operative conditions (both temperature and pressure) are very low with no strong excursions.
Furthermore the leakage happened soon, i suppose the gasket was damaged during tightening.
Anyway thanks a lot, i will further investigate your comments.

Giuss

 

[rconner]

What is the service?

 

[unclesyd]

Two things that bother me and one is the use of a Class 125 flange. On certain sizes of Class 125 flanges it is almost impossible to seat a SW gasket because of rolling or cupping the flange. If a SW gasket is properly seated and the flange faces stay parallel they are amazingly forgiving. I don't know if a 42" Class 150 falls into this problem range.

Second in your OP you mention inward buckling which is one sign of a cupped flange. This can be easily remedied by using a an inner ring on the SW gasket. There has been an awful lot of work in respect to inward buckling of SW gaskets so ask you supplier about getting some of the papers.

You also post that the gasket was destroyed. I think if you look around you will find the spirals.

 

[Zapster]

Giuss,

Does the flange that is causing the problem have a different surface finish than the others? What is the surface finish of the flange. Did you state what the design pressure and temperature for this line is? Are there any excursions in temperature or pressure that might be taking place?

 

[unclesyd]

Though not exactly your case but here is a report on SW gasket failure.
As I posted above there are some good reports on inward buckling of SW gaskets. A lot of work was done under the auspices of the Pressure Vessel Research Council.

http://www.hse.gov.uk/foi/internalops/hid/spc/spctld07.pdf

 

[HeviiGuy]

Giuss,

I've come late to the party. Nevertheless, I'll throw something out here as well:

You've indicated that everything is identical on this joint as it is on others which haven't failed. You've also mentioned that the bolting procedure was the same. The trick, I suppose, is to determine the "hidden" factor as it's apparent that all is not equal. One key factor that's not been addresed yet is actual bolt stress. Although you say that the torqing procedure was witnessed and accepted, I contend that it was a moot exercise in this regard. Perhaps something was indeed awry, thus causing inconsistent bolt loading. Too much lube, maybe? Different stud/nut/flange friction factors? Verifying that the proper torque was applied in a certain sequence indicates little to nothing about the residual bolt stress. Measuring the elongation of "torqued" fasteners should always be considered as part of the process if the application is important enough to warrant monitoring.
We're commonly involved in projects after a number of scapegoats ad been slaughtered in attempt to lay blame for the joint's failure, the gasket always being the first to succumb. Alas, it's often not the gasket's fault nor the flange's but rather bolt preload scatter around the flange. Two ways to handle this:

Measure the resultant stretch and "tune" the flange accordingly.

The second is to consider, particularly with a 36 bolt flange like this, the use of bolt tensioners with at least 50% of the suds being tensioned simultaneoulsy. Not only would this enable even gasket crush thereby eliminating a major factor of potential damage but, it would also tend to eliminate some of the unknowns affecting final bolt load.

Ciao,

HevïGuy

http://www.heviitech.com

 

[saplanti]

If you get excessive external forces at the flange connection this type of failure might occur. You need to investigate this option in your system if everything else is OK.

Hope it helps

Ibrahim Demir

 

[gr2vessels]

The crushing of the gasket changes the physical alignment of the assembly, right? It might be worthwhile to measure the initial gap between the flages, perhaps at two stages of bolt tensioning, all around the flange perimeter, then at/after start-up and during operation, particularly if deformation is noticed at any stage. Smaller gap than initial means crushed gasket, off-loaded bolt(s), overstressed joint, no equivalent pressure taken into account. Perhaps the rating could be exceeded if equivalent pressure is included? Also, any larger and skinny flange should be checked somehow for rigidity, that's way is now mandatory in the code.
My two cents...
gr2vessels

 

[unclesyd]

On the failed gaskets did you notice any distortion of the outer ring when first installed?

 

[BillBirch]

You should consider the flange facing. As you know, the spiral wound gasket is a series of alternative "chevrons" of metal and filler material.

Prior to installation, the "Vees" are wide, and upon compressing become narrower - i.e the angle of the vee decreases. This requires the contact point (on the flange faces) of the vees to slide away from the apex across the flange faces (the cosine of the angle increases with a decreasing angle).

If the contact face is not sufficiently smooth, the tip of the vees will snag on the faces and be prevented from sliding, causing the chevron ribbons to bulge (or buckle). A standard gramophone finish may provide sufficient obstruction to flow to cause this failure.

Incidentally, the gramophone final is best suited to flat gaskets e.g CAF where the clamping force is always normal to the gasket surface, and the peak-trough configuration of the flange faces allows micro concentration of clamping pressure to bite into the softer gasket, creating interference and a longer leak path.
Regards,
Bill