Stress Intensification Factors for Weldolets 1

[DSB123]

Can anyone answer this one. We have received stress analysis results from a Contractor who has used Caesar. The situation is a 1"nb Weldolet onto a 6"nb header. The SIF's are in accordance with the Code ASME B31.1 for the header branches but the SIF for the 1"nb branch at the node is indicated as being 1.00 and not the same as the header. Whilst I know the branch SIF will be less than the header due to the ratio of the branch to header the Code, as far as I beleive does not allow this SIF reduction so why does Caesar not follow the Code. If the analyst wants to then reduce the branch SIF he may do this at his own discretion in subsequent analyses.

 

[richay]

B31.1 contains an alternate equation for the computation of the branch SIF, assuming the input meets several criteria. I suspect that this input met the criteria and the alternate computation was performed. Now, if the result of the SIF computation yields a value less than 1, it is changed to 1 (which is the minimum allowed SIF).

If you can't resolve the issue, e-mail the input file to COADE for review.

Richard Ay
COADE, Inc.

 

[DSB123]

richay,
Thanks for the reply. I know of the alternative equation for the branch SIF but unless all of the requirements are met this eqn cannot be used as I see it. In the case concerned there are no specific details of the branch and hence the internal radii and the external radii are not specified so may not meet with the B31.1 requirements. In which case as I see it the SIF should be set to the same as for the main. It is worth noting that I have modelled the branch in both Autopipe and PSA5 and both default to the same SIF for the branch whereas it appears that Caesar has defaulted using the alternative equation in B31.1 Surely this is wrong as this is not conservative for the branch connection.

 

[richay]

Then go to "tools\configure-setup", click on the SIFs and Stresses tab, then click on the option for Reduced Intersection and set this option to Pre 1980.

Read the help text [F1] on this item for additional details.

Richard Ay
COADE, Inc.

 

[DSB123]

richay,
Thanks for the reply but I do not have Caesar. I refer to a Contractor's stress results.My point is that in reality the Code (B31.1) does not allow the use of the alternative branch SIF unless the geometric qualifications are met. When designing unless the Contractor prepares a specific detail drawing calling for the dimensions to meet the Notes in B31.1 then the "Alternative" branch SIF values are not valid. My question is why does Caesar seemingly utilise this Alternative equation from B31.1 irrespectively and by default??. Surely the conservative approach by Caesar would be to no use this equation.

 

[richay]

When a code implements a new option that most of our user base wants, we implement it in the software as default behavior.

I agree that the geometric conditions must be met in order to use the branch equation. However, the only way to insure this is to cut the fitting and take measurements, which destroys the fitting. So the only requirements which we check are those that can be obtained without destroying the fitting.


Richard Ay
COADE, Inc.

 

[DSB123]

richay,
Agree with your response but my point is that if the geometric requirements are not guaranteed how can the alternative branch equation be used. for conservative results (i.e. higher stress levels) it must be more prudent to use the base equation for the SIF.

 

[richay]

I'm sorry you don't agree with our default. However, as I stated above, if you don't like this, you can change it in the configuration.

Richard Ay
COADE, Inc.

 

[4Pipes]

Setting aside the wording of Code requirements and just looking at weldolet construction, a sif = 1 for the branch would seem logical. The weldolet is probably rigid enough to prevent the branch feeling the affects of the header (?). I would doubt if the stess distribution local to the branch butt weld on each side of the weld is any different to the hub on a weld neck flange ? I you are going to worry about accuracy, I would think there is more practical benefit in offsetting the branch to the branch butt weld.

 

[DSB123]

4pipes,
You are suggesting that a weldolet connection is similar to the hub on a weld neck flange as regards SIF. I think you really need to read the Code before making such comments. You are saying the Code is incorrect!!! Hope you have a strong basis for discounting the Code.

 

[4Pipes]

DSB123, please carefully re-read the my first line which I thought gave a strong indication that my reply is not intended to be code gospel. Other words including "probably are also a stong indication that the post is intended for discussion rather than making definitive statement about use of the code. The "(?)" also has a similar meaning.

All I am saying is look at the branch end of a weldolet as opposed to the header end. If you took a slice that included the branch butt weld and a bit of pipe and fitting each side, I would doubt the stress distribution is much different to a similar slice through a butt weld on a WNF flange at least from a stress raising point of view. Setting aside the wording of the code, is this a reasonable observation especially for small d/D branches. If not, why not ?

I must admit that at this instant, I only have an out of date electronic copy so I not going to argue wording of the code. I couldn't spot where in App D where the code sif for the branch end was specifically stated. I don't claim to be able to read electronic copies as reliably as paper copies. At the end of the day, the code doesn't prevent the designer from doing his own thing provided he can justify it.

The header end of the weldolet is obviously a different matter.

 

[xhpipe]

Is the code ever incorrect??? Well it is a document based on concensus and sometimes making a change that is required is very difficult to nearly impossible. So yes sometimes the code may ne incorrect.

Having said this however one must follow the requirements of the code which the work is being performed under... there must be some order and basis to your work.

So one should always keep the truth and the code requirements as two separate issues.

From the "truth" point of view having examined numerous FE plots I can assure you that the majority of stresses intensified by the weld on fittings geometry are on the header.

However you have correctly stated the B31.1 requirements. As for any computer programs defaults... well all of them are sophisticated tools and must be properly used and tended to. I suspect that there may be other faults you may not have even caught....

In the final analysis B31.1 and B31.3 in their forewords clearly state that they are not design cookbooks, and that experience and judgement is a prerequisite in doing the work succesfully.

So I personally find no fault in the CAESAR II default, and you should use your experience and judgement in applying the code.

And one other question while we are beating on this dead horse.... does the application of the SIF to the branch over stress the branch? And if so for what types of loads??? i.e., Occasional, displacement, or sustained loads?????

Regards,
XHPIPE

 

[DSB123]

xhpipe,
Who's beating on the dead horse????. Answer to your last question is that the displacement stresses are well over the Code limit when the application of the CODE sif is applied.

My original point is that the Code gives requirements for using the alternative branch SIF equation and if the geometry does not comply then you cannot just arbitrarily use the equation which "makes" the system work. If the use of the lower SIF is based on sound engineering then so be it. I've seen a few weldolets which have "popped off" the parent pipe in operation to be wary about the sif's to apply.

 

[xhpipe]

Your statement "I've seen a few weldolets which have "popped off" the parent pipe in operation to be wary about the sif's to apply."

This clearly indicates the over loaded condition was indeed the header attachment. Therefore the SIF being applied to the branch would not be the limiting factor. Indeed your observations would appear to confirm not applying an SIF to the branch vs applying an SIF to the header.

In any event I am anadvocate for weldment details for these types of fittings. Seldom are weldment details called out in the design documents. In the case of weld on fittings the SIF is very dependent on the weldment. This usual lack of detail versus the application of a SIF numerically imposed to 2 or 3 digits has always been troublesome to me.

Regards,
XHPIPE

 

[4Pipes]

Unspecified olet weldment details and sifs to 3 decimal places - a bit like a pair of worn out shoes or a blanket carried around by a baby. No logic but comfortingly familiar.