Is there any difference in result if same piping analysis problem is run in Caesar-II & in ANSYS with PIPE16 element ?


regards,

sam

Theoretically no - a 3D beam element is a 3D beam element. However, you have to watch the implementation. For example:

- When you specify a restraint in CAESAR II, the default stiffness used is 1E12 lb/in. What is this value in ANSYS? Depending on the model, this can affect the results.

- Read about how CAESAR II models rigid elements in the Technical Reference Manual. You'll have to finesse one of the programs to get these to match.

There are likely other differences as well. Benchmarking is an art.

Thanks for such a quick reply.

Is it possible to verify the Flex factor & SIF of any piping fitting using beam element in any general purpose FEA software ?

Are we not to use FEA elements like shell or other multi-node FE elements to achieve this ?

regards,

sam

Sam,

Yes it can be done.You can find a series of discussion on this topic in one of the threads ( I cannot recall which one) where I participated and you can check Dr.Becht's response on the same thread.You can get an excellent overview on use of element types in the WRC Bulletin by Hechmer and Hollinger with reference FEA and requirements of SEC VIII DIV 2, SEC III "Design by analysis" philosophy.

Regards

Regards

Sorry, Anindya; I beg to differ. With 3D beam type elements, you can't. If you can, please explain.

We know with other FE elements like shell element, we can.

regards,

sam

Sam,

With plate and shell elements, stresses are naturally categorized in membrane and bending, but that is not the case with 3D elements , say 8 or 20 noded brick. So for these elements we use the concept of "linearization".So where else is the problem?

Regards

Anindya,

You had misunderstood me. I didn't talk about brick element; I talked about 2 node beam element - PIPE16 in ANSYS - with that one can't verify SIFs at fittings, but use APP-D SIFs of B 31.3.

regards,

sam

Sam,

Yes, I misunderstood you.

To properly use FEA to correlate SIF with Sec VIII Div 2 appendix 4 or SEC III SUBSECTION NB requirements, use either brick or shell elements.For additional details, you need to use the paper by Hechmer and Hollinger, which in case you don't have and need, I can send it to you.However the paper by Hechmer and Hollinger won't throw any light on how to get the B31 type SIF values.The philosophy for doing so is something like this : get the "peak stess" and use the correlation 2i= CAESAR II*K2.Boundary condition that you need to use in your FEA model and the subsequent fine tunings need very good judgement.

Regards

Further to my earlier response, I wanted to write CAESAR II, which I don't know how got to CAESAR II.

The character string "c 2" is an "auto-text" string that gets converted by the Forum to "CAESAR II".

Yes, as we wanted to verify B 31.3 bend/miter SIF & flex factor, one way could be analyze the same say one end anchored 90 deg elbow subjected to some inplane/outplane force at free end problem using Caesar-II and using FEA with plate or brick model both and compare displacement at free end and peak stress at same nodes and verify both flexibility factor -i/o and SIF-i/o separately. CAESAR II has user defined flexibility factor & SIF facility - which will also be useful in this activity.

regards,

sam

My guess is that you will find many ASME papers over the years that have performed this and similar exercises. Maybe you will find your work already done or at least some guidelines for proper evaluation.

I have found one in ASME JVPT Vol-110 Nov 1988 sht 374 by Fujimoto & Soh for D/T >100 fittings. Thank.

regards,
sam

To

Mr. Dave Diehl & Our other learned forum members,

As you have experience & access to ASME papers on related topics, please share with us any useful info on SIF & flex factor of D/T >100 for B 31.3 piping in your possession, if you have.

regards,

sam

Sam,

What happens is that the geometries start experiencing localized buckling and kinking effects. So instead of your elbow bending and ovalizing in a manner as predicted by Markls test at some point the higher D/t reatio elbow will locally kink or buckle leading to a sudden drop off in resistance or drastic increase in flexibilty along with a large increase in "effective SIF".

Because these phenomena are so localized beam elemnt models such as CAESAR II simply cannot be counted upon to come to a reliable solution. And the codes basis itself breaks down as well hence the note in App D. Due to the 3 dimensional nature of these kinks usually FEA is the only way to look at these high D/t ratios with confidence.

Sir,

In the paper of Fujimoto & Soh cited above, FEA studies have shown higher flexibility factor than predicted by 4-cut 1.5D miters with D/T >> 100, but similar SIF.

regards,

sam

I understand your desire for more information on these subjects but I cannot simply reproduce and distribute copyrighted material.

Who knows, if we all paid for our papers, ASME might be able to put more revenue into maintenance & update of the Codes.

I respect your integrity. I don't want coyrighted material, can't I ask for your opinion - as you are our mentors here ?

What I wish to get some corrected correlation for thin (D/T >> 100) 4/5-cut miter flexibility instead of 1.52/(h^5/6) as available in B 31.3 App-D. Welding elbow correlation in B 31.3 is 1.65/h.

4/5-cut miter is a close approximation of welding elbow and Fig-7 (c) & (d) & table-2 of Fujimoto & Soh's paper shows the same for flexibility in 0.01 to 0.2 range of h.

regards,

sam

I am unfamiliar with the paper you cite and I have no opinion to offer on your specific interest here.

I believe there is still resistance to establishing SIF's and flexibility factors through analytical methods. Correct me if I am wrong but aren't these values to be established through physical testing? I believe the soon-to-be-released B31J will once again affirm that position. (Notice I said "I believe" a few times here. Perhaps others can confirm or correct.)

And isn't that why there is a lack of data? No one wants to run the tests to develop the new relationships?

And isn't that why Markl's work is so amazing?

"I believe there is still resistance to establishing SIF's and flexibility factors through analytical methods." Yes the benchmark gold standard is still prototypical testing.

"I believe the soon-to-be-released B31J will once again affirm that position." B31J is a B31 standard under deveolpment it is titled "Standar Method for the Determination of Stress Intesification Factors (i-Factors) for piping components by test.

Currently the codes are "silent" as far as FEA only Sifs is concerned.

"And isn't that why Markl's work is so amazing?" Markl et. al. were either very lucky or very good... my money is on very good.

Sam...

"What I wish to get some corrected correlation for thin (D/T >> 100) 4/5-cut miter flexibility instead of 1.52/(h^5/6) as available in B 31.3 App-D. Welding elbow correlation in B 31.3 is 1.65/h.

4/5-cut miter is a close approximation of welding elbow and Fig-7 (c) & (d) & table-2 of Fujimoto & Soh's paper shows the same for flexibility in 0.01 to 0.2 range of h."


Your Statement "4/5-cut miter is a close approximation..." well yes and know the difference between the miter and a wrough elbow are substantial BUT maybe their peak stresses (at different locales mind you) and hence their fatigue strength may be similiar if so then as i correlates to h their flexibilties would also be linked. Unfortuanately you have only two sure ways to be 100% confident testing or FEA.

Lets look at the problem from another view point.... can you design the system with the SIF and h being off by 150-200% and still get things to work out? If so your margin of unknown things amy still be tolerable.

Sir,

Most of the cases FEA studies are validated on simple real prototypes. Even, the paper cited here had FEA element type validated through static stress testing. Also, isn't Caesar-II also a beam element based FEA ? Moreover, we will be building real plant - if it does not work for 7000 thermal cycles, with f=1.2, we can go for 3500 cycles and then replace the fittings.

Now comes the situation of 150-200% additional safety factor on thermal expansion stress - is it salable in today's marketplace where civil & layout people don't provide even the just needed flexibility loops citing various reasons like cost reduction.

regards,

sam

Sam,

My suggestion was based on the facts as I understand them namely...

You are working in a D/T ratio beyond 100.

You have a mitred elbow.

You have a paper in hand discussing a welding elbow above D/T 100 and wish to use its cited SIF and h for you relatively thin high D/T elbow.


My Suggested "Now comes the situation of 150-200% additional safety factor" is really a margin for ignorance. We don't know how the miter will behave exactly but we are absolutely positive that a mitered elbow behaves far differently then a wrought elbow. How great a difference can only be adressed by three dimensional analysis such as testing or FEA.

You are in unknown territory your choices are....
1)Perform FEA/ Testing to determine the correct SIF / h

2)Fudge the SIF from the code with a factor of ignorance

3)Ignore the whole problem, build it, and do a full sized fatigue test (either with or without informing the owner I guess)

Obviously number 1 is the best choice, followed by 2, before I would ever go to 3 I would probably walk away from the mess.

Having had this actually problem myself in the past I opted for either 1 or 2 at times, and one time washed my hands of the deal... I told the designers and my boss go ahead with your scheme but when it breaks its all yours. (The components did break a few years later (fatigue).

I hope this helps its a dicey problem!