Dear stress fraternity,

I need to perform local stress evaluation for large diameter thin walled pipes at restraint locations. I have been using "Roark's formulas for stress and strain" for different loading patterns and calculating various stress quantities.But I am doubtful about how to combine them.I think B31.3 explicitly doesnt give any view regarding this.How should be the various stress quantities resulting out of local stress evaluation be combined?
An advice regarding this would be greatly appreciated.
Thanks in advance.

I imagine most people use the kellogs methods
for calculating the various local loads due to pipe attachments.
We had a 'Engineer' do work here to compile swl's for pipe supports at the start of a Project.
It was quickly evident that the tecniques used were complete rubbish and he left fairly quickly afterwards.

I believe that the kellog method is a simple addition of the axial,circumfrential, and longitudinal bending stresses, and are compared to simple allowables.

(i have though seen these three stresses added in umpteen different ways.)

Personally, simple FEA would be preferable
(i haver used this though)

My favourite method is from 'Piping and Pipe Support Systems' by Paul R. Smith & Thomas J. Van Laan par. 8.3.

The correct way to design is to a vessel code. I have used BS PD5500 Appendix G but no doubt ASME VIII does the same thing. The problem comedown to whether you can get away with "thin saddles". If not stiffening rings are needed. You might find you are putting in stiffening rings [ASME VIII UG-28, UG-29 and UG-30] anyway if your large diameter thin pipe must withstand full or partial vacuum. Beyond this abandon the "thin saddles" and use ring supports. With the BS PD5500 approach you must substitute beam analysis moments for those in its "2-support vessel" model. You also have to wade through a lot of "dimensionless chart" look-ups. [Unless you have correllated them previously for a spreadsheet.] But buck up it can be done!

Michael

I got a look at ASME VIII Appendix G for Divisionn 1. It suggested look to BS PD5500 for support design as main reference.

Thanks Michael. I will check out ASME VIII Appendix G Div 1 and BS PD5500. Thanks for the valuable time you took for replying.

Regards

API RP 579 has a section for evaluating local thin areas on piping and pressure vessels.

Hi Larry,

Please insult our mutual friends for me.

I don't think that Ramkumar was referring to in-service piping with local thin areas as much as he was new piping with a large D/t ratio. I agree that they should be evaluated using the ASME B&PV Code rules for evaluating local stresses in the wall near supports. The legendary Brownell and Young tome also addresses this topic and it is addresses in some of the AISI nomograms. Maybe even AWWA Manual M-11 addresses it.

Regards, John.

Ramkumar,

You can use local stress formulas for pipes sitting directly on steel using Roark's stress and strain formulas ( as an example:5th edition of this book,Table 31, Case 9b).Then use the relues of ASME SEC VIII DIV 2 APPENDIX 4 I.E. STRESS CATEGORIZATION AND SUMMATION.

In lieu of calculations, a thumb rule good design guide can be to use SHOES WITH RF PAD for all 24" and above thin walled ( 10S )pipes supported at 20'-00" span.

Hope this answers your question.

Regards

To Superpiper or anyone else who uses the Kellogg's method:

With regards to simplified methods per Kellogg (only applicable to trunnions and nozzles), it is not clear whether and how the local longitudinal bending stress and the local circumferential bending stress are combined. (I have the second edition.)
Are they independently compared with 1.25(Sc+Sh) or combined as SRSS and then compared with 1.25(Sc+Sh).

With regards to the source that I previously recommended, it covers local stress at welded attachments (based on WRC107) as well as at lose supports like line contact and lose saddle support. These are based on same formulas (Roark and Young).

In case of the line contact longitudinal and circumferential stresses are independently summed and the greatest of is compared with code allowables

In case of a lose saddle circ. stresses (hoop and circ. bending) are summed and compared with code allowables.