Overpressured Vessel

[loilfan]

A vessel overpressured and blew out a gasket. I am attempting to determine what minimum pressure the vessel must have seen using the formulas in Div 1 App 2.

To determine Wm1, do I use the total area of bolts, or just one bolt since each bolt is torqued individually?

If I have a flange with 4 bolts on it, do I increase the compressive load on the gasket if I add 2 more bolts (using the same torque)? Or is the flange just more evenly distributed since adding the extra bolts wouldn't compress the flange more than the 4 already have?

 

[SnTMan]

loilfan, recommend you review the definitions of Ab, Wm1, Wm2, etc.

Also, if by "blew out" you mean the gasket exited the joint, I don't think this method will get you there.

Regards,

Mike

The problem with sloppy work is that the supply FAR EXCEEDS the demand

 

[racookpe1978]

First, if the gasket was "blown out" (from between two flanges?) then that merely says that particular gasket was the weakest gasket located at the single point where the flange compression was lowest.

You DON'T KNOW the previous torque on those 4x bolts, nor the actual tension in the bolts (studs ?) clamping that particular flange set together. Merely that that flange was clamped the least of any other flange in the system. Are you trying to estimate acvtual failure pressure using some assumed strength of the four studs and the strength of the gasket?

 

[TGS4]

Appendix 2 is a method for designing flanges, not for the purposes that you're intending. Try PCC-1. What kind of handset blew out?

 

[LittleInch]

I agree with racookpe. I don't think this is the way to find out the pressure.

If the pressure got so high that you blew a gasket out, then it's beyond 150% of the flange rating minimum and possibly up to 200% or more, but there will be too many unknowns and assumptions to make the calculation any more accurate that that - IMHO.

How come a pressure vessel got to that level of overpressure without relief systems kicking in??

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[loilfan]

I fully agree that we will never be able to determine what the blow-out pressure was given the unknown variables. We are trying to determine the what the minimum pressure is that the system must have seen to have caused flange leakage to help give us something to work off of. The SCADA system was recording too infrequently to even see the event. Haven't heard the details about why the pressure relief didn't prevent the overpressure; we are still in the early stages of this investigation.

The flange was in operation without leaks, so I'm not looking at Wm2.

Nobody answered my question in the last paragraph, but after sleeping on it, I've come to the conclusion of using the total bolt area. Although I can use the torquing of a single bolt to determine what my compressive force is (where adding more bolts wouldn't increase the force if torqued the same), I need the total bolt area to determine what pressure is required to separate the flange and cause leakage.

 

[SnTMan]

Done...

The problem with sloppy work is that the supply FAR EXCEEDS the demand

 

[LittleInch]

But you still don't know the bolt load.

you also don't really know how much stretch you needed on the bolts to get sufficient reduction of pressure on the gasket face.

The min pressure is basically just a few percent above whatever that joint was pressure tested to.

Anything else is pure supposition and the minimum pressure could be double that pressure.

BTW, how can you add two more bolts? Were some bolts removed??

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[TGS4]

I've performed bolted flange joint failure investigations before. Never used the Appendix 2 calculations, because they're not appropriate. Most likely, something caused one or more bolts to lose their load (vibration, fatigue, ratcheting, etc), which caused the gasket blow out at a pressure quite likely lower than the PSV set-point and possibly even at or below your normal operating pressure.

Remember, Appendix 2 is for designing FLANGES, not for designing the bolted flange joint system.

 

[loilfan]

LittleInch, I know the bolt prestress along with the torque. Using Bickford's book, I can assume some losses from the original loading for embedment relaxation and tool/operator accuracies to estimate the bolt load.

I asked that hypothetically.

Keep in mind everyone, I am not doing this to give me an exact solution. I want to see what the flange should start to leak at with the known information that I have.

 

[r6155]

There is only one cause of leak: wrong design, fabrication and inspection.

Regards
r6155

 

[LittleInch]

Lolifan. You keep stating that you"know" the bolt stress/torque. Can you explain how you know this. Do you have as built torque figures? Can you say for certain it didn't change?

If so great,but otherwise you will generate a figure which could be put by a factor of 2

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[DSB123]

LI,
Could be out by a factor of greater than 2. Too many parameters to get a sensible answer!!!

 

[loilfan]

LittleInch, I know what the torque should be from the standard torquing tables that were used. I am doing this exercise to show what the minimum pressure should have been using the as-expected scenario including some expected installation and embedment losses. If the gaskets blew out below this expected pressure as confirmed by the PSV lifting, then it tells me that either the PSV is undersized or the bolts were installed improperly.

 

[LittleInch]

Ok, but of you're doing a theoretical calc you should already know that the joint is good for at least 1.5 times the flange rating as an absolute min. That should be a long way from the mawp of the vessel which should be protected by a relief valve or similar device.

So I can't figure out why you're bothering.

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[loilfan]

The overpressure event was transient and the overpressure protection is downstream of the vessel. The PSV would not have been able to protect from a surge of pressure and there is no way of knowing if the PSV even lifted. I'm using this calculation to show that a high pressure event is what's required to blow out the gaskets as a justification to service the PSVs and perform additional NDE. It's possible that the joint wasn't tightened the way it should have been and it failed at a lower pressure - we will never know. But at least I can use this as an argument to doing more than just assuming that the PSV caught it all.

 

[SnTMan]

Those that are still participating (I'm not) might try to determine if the gasket BLEW OUT or merely LEAKED. Not the same thing. Not at all clear from any of the foregoing...

The problem with sloppy work is that the supply FAR EXCEEDS the demand

 

[LittleInch]

OK,

I can't understand why you're having to go to that extent, but if it helps then go for it assuming the bolts were all originally tightened to the correct torque. I wouldn't be surprised to find the pressure is x2 of the MAWP of the flange, never mind what the MAWP of the pressure vessel is.

Some sort of severe transient event could be enough to allow enough fluid out to destroy the gasket before the relief valve had even had the chance to lift properly - if you don't have good recording or the event was so sudden the DCS didn't properly capture it then you are running blind.

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[loilfan]

LittleInch, that's my point of view on it too. Thanks for the discussion.

 

[chicopee]

How do you know if the vessel overpressured? didn't you have a pressure gauge on the vessel?