SIF's in Branches to check max loads 3

[rjstephens]

Hi all,

I have a highly stressed welded tee(ASME B16.9). The stresses are not within the normal stress range allowables, but pass to the liberal stress range case.

Pullman Kellog states that the stress intensities added at the tee as per the code(ASME B31.3) cover the local & secondary stresses of the component to ensure that no failure occurs. Is this true for the liberal stress range case as well?

In the past I have used FE Pipe to check any components of concern, but do not have the same facilities here but will have them again after this particular contract.

My real question is, if one calculates the beam stress forces/moments and subsiquent stresses using Caesar II, does one have to check the branch components for failure due to excessive thermally induced loads as well? Or does ASME B16.9 cover this type of loading as long as the code compliance stresses(incl. lib. case) are not exceeded?

Thanks in advance

Rob

 

[rjstephens]

Hi all who may be interested,

After a bit of research, I have found that if one apllies the SIF and allowable design stress criteria of ASME B31.1, then the external loads applied to the branches do not need to be checked.(As per the B31.1 code). This is not the same for ASME B31.3, which uses higher allowable design stresses and different SIF's at branches.

In the absence of FEA programing, I have resolved to insert a B31.1 branch into the B31.3 calculation to assess if further calculations are required. The flexibilties for the bends are calulated the same way for both codes, so there should not be any major differences in the bending moments calculated at the branch. Another way of determining if the branch requires additional calculations is to limit the expansion stresses calculated at the branch to approximately 70% of the liberal allowable stress range for B31.3. Please note that the above comments are only for ASME A106 Grd B below a temperature of 200 deg c.

If any one has any comments or further data, information to share, would be greatly appreaciated.

Cheers

Rob

 

[bvi]

You are not correct. No additional calculations for the branch are required by ASME B31.3. Further, the so-called (by Coade, not by the Code) liberal stress allowable is actually the more precise calculation. The other allowable simply makes the assumption that the sustained stresses are equal to the allowable stress, a simplifying but conservative assumption.

So, with respect to thermal loads on branch connections, all that is required is the flexibility analysis, such as is performed using CAESAR II. This is the same in both ASME B31.1 and ASME B31.3. Note also that the criteria for thermal expansion is also the same in both codes, although the so-called liberal allowable is included via text rather than as an equation.

The can be a difference in allowable stress, depending upon material and temperature, between B31.1 and B31.3 but this does not affect how the calculations are performed.

 

[rjstephens]

Hi bvi,

The sections from the code (B31.3 & B31.1) that I am refering to are 304.3.5(a) & 104.3.1(F), the design of the tee's as per code are purely for pressure containment, further applied loads need to be considered by the designer and in the case for B31.1 are covered by the use of SIF's to that code's requirements. If you read the respective codes, they clearly have a different consideration on how the designer is to take into account the additional loads other than pressure on the branches.

In general there is a lot of 'fat' left over to cover the other applied loads, but if one has any concerns for an individual component, then it must be checked to assess if failure will occur, using FEA or other recognised methods(Kellogg etc). If you have had a highly stressed/loaded tee and not calculated weither the other forces that the code needs one to consider may cause failure, and never had a failure, is it pure luck due to the 'fat' in the material & code???

I may be entirely incorrect, but that would not explain why, when some branch connections that have been a cause for concern by the calculated high beam stresses in CII, but do not fail, are then analysed using FEA methods & do require design modifications for the loads being applied to them to ensure that failure does not occur within the limits defined by the various codes. This may be explained by the different SIF's calculated and applied to the branch by the different methods.

Don't mean to ramble on so much, but there is a lot more to this subject than just apply Caesar II & hope for the best.

Rob

 

[KernOily]

>My real question is, if one calculates the beam stress forces/moments and subsiquent stresses using Caesar II, does one have to check the branch components for failure due to excessive thermally induced loads as well?

Why would you have to do this? If the applied load < allowable load, you are fine. The code SIF at the tee will cover you here.

>Or does ASME B16.9 cover this type of loading as long as the code compliance stresses (incl. lib. case) are not exceeded?

From the ASME website, re: B16.9:

"This Standard covers overall dimensions, tolerances, ratings, testing, and markings for wrought carbon and alloy steel factory-made buttwelding fittings of NPS 1/2 through 48. It covers fittings of any producible wall thickness."

B16.9 is a dimensional and manufacturing code. As far as I know it says nothing about loading, be it applied, allowable, fatigue, etc. Please correct me if I'm wrong here.

>I may be entirely incorrect, but that would not explain why, when some branch connections that have been a cause for concern by the calculated high beam stresses in CII, but do not fail, are then analysed using FEA methods & do require design modifications for the loads being applied to them to ensure that failure does not occur within the limits defined by the various codes. This may be explained by the different SIF's calculated and applied to the branch by the different methods.

Read the Markl et al. papers about how the SIFs were developed and implemented into 31.3. Note that his work was done WAAAAY before computer FEA even existed. There is work going on right now to restudy and revise the SIFs.

I'm still not sure what your question is. Why are you concerned about this? You said your tee is 'highly stressed'. Is that per the C2 calculation? What does that mean exactly - 98% of allowable?

As long as the applied stresses in the tee are below the allowables, and fatigue and other loadings are not a concern, you are fine. Yes I am sure your FEA will show that the stress level in the tee is higher than the stress calculated using C2 and the SIF. If it makes you sleep better at night, reinforce it.

Thanks!
Pete

 

[bvi]

Paragraph 304.3.5(a) is in the pressure design rules of the code, and it states that in addition to pressure design (e.g. area replacement), external forces and moments must be considered in the design. This is done elsewhere in the code via flexibility analysis rules.

With respect to doing some other calculation and finding reinforcement is required, there are many open issues in that statement. How were the stresses evaluated, for example? Was a limit load analysis done, or was some conservative stress classifcation scheme with elastic analysis done? Just because someone did an analysis and concluded additional reinforcement beyond what the code requires for external loads was necessary, does not necessarily mean that the reinforcement was in fact required to prevent failure.

In any case, the ASME B31.3 does not require detailed analysis of branch connections for external loads other than the use of SIF factors (although the issue of what SIF to use for sustained loads is presently not stated in ASME B31.3, B31.1 uses 0.75i).

This aspect of the rules has been unchanged since long before people could realistically do FEA or other detailed analysis of the branch connections for external loads. Further, if you think about it, imagine having to do detailed analysis of all the branch connections in a large plant, it would be an unrealistic and unduly burdensome requirement. The rules are deliberately simplified.

 

[JohnBreen]

Hello,

Wow, there seems to be some confusion here in the use of SIF's. I notice that there are several threads active here on the "first page" that in one way or another ionvolve SIF,s.

When you find that SIF's are involved the best thing to do is do enough reading to increwase your "comfort level".

Now, haveing said that, I am afraid that I am about to reinforce the rumor on this board that I cannot say anything in just a few words. Sorry about that, please bear with me.

SIF is “shorthand” for Stress Intensification Factor. SIF’s were developed by the B31 Code Committee in the late 1940’s and early 1950’s. The purpose of the SIF is to allow some additional design margin for the fact that some manufactured and some fabricated piping components have shorter fatigue life (as measured in terms of loading cycles) than straight runs of pipe. The B31 “Committee on Flexibility” undertook a program of testing various piping components by repeatedly subjecting them to cyclic displacements thereby inducing alternating bending stresses in the components. From this testing, the Code SIF’s were developed and theses SIF’s can be found (for various pipe components) in appendices “D” of B31.1 and B31.3.

It must be remembered that the analysis methodology prescribed by the B31 Pressure Piping Codes is based upon beam theory. Using this methodology, bending stresses are simply the bending moment divided by the pipe section modulus. However, fatigue testing shows that, for a given moment, in some component configurations the stresses that actually occur are greater than would be predicted by classic beam calculations on straight lengths of pipe. Code SIF's are simply "Band-Aids" on beam theory to help connect the stresses calculated for piping components by beam theory with the real world. A B31 SIF simply reflects the test results which compared the number of cycles-to-failure of a straight piece of pipe with a girth butt weld in it, to the number of cycles-to-failure of a piping component (like the branch connection). You will note from a perusal of Appendices “D”, that the B31 Pressure Piping Codes provide SIF's that are appropriate for specific types of piping geometries (e.g., fabricated components, fittings, etc.); those which testing derived data is available. However, the B31 Codes also tell the piping engineer that if he/she has more appropriate SIF's for specific fittings they are to have the latitude to use these data. . It would be prudent for the piping engineer to peruse B31.1 and/or B31.3, Appendix "D" to determine the details of the specific component geometries for which SIF's are provided. For example, the piping engineer may specify that B16.9 Welding Tees (fittings) be used, but the "crotch radii" and general wall thickness of these fittings varies greatly from manufacturer to manufacturer. If the piping engineer determines that the component to be used in the design is significantly different from those described in Appendices "D", it may be appropriate to develop alternate SIF data.

So summarizing, the “equivalent bending stresses” that are calculated using B31, Appendix D SIF's came from (very limited) cyclic load testing of piping components and really were intended to address fatigue. These beam theory bending stresses are (for a beam that is a straight pipe) simply the bending moment about a given axis (in-plane) divided by the gross section modulus. For other piping components (e.g., the branch connections at issue here), the bending stress is calculated as the INTENSIFIED bending moment divided by the gross section modulus. This B31 calculation actually yields about 1/2 the true elastic bending stress (this is the result of assigning the girth butt weld test specimen a 1.0 SIF; for example, just for interest compare the SIF's from B31 to the stress intensifiers used in Section III of the ASME B&PV Code,(ND-3600) beam stress calculations), but these B31 calculated stresses are to be compared to the allowable stress that is also provided by the B31 Code and when used in this context will result in a reasonably conservative design for most applications.

You will note from a perusal of Appendices “D”, that the B31 Pressure Piping Codes provide SIF's that are appropriate for specific types of piping geometries (e.g., fabricated components, fittings, etc.); those which testing derived data is available. However, the B31 Codes also tell the piping engineer that IF he/she has more appropriate SIF's for specific fittings they are to have the latitude to use these data. Good stress analysis software provides the piping engineer with the capability to override the Code prescribed SIF's (that are written into the software) by specifying alternate data. It would be prudent for the piping engineer to peruse B31.1 and/or B31.3, Appendix "D" to determine the details of the specific component geometries for which SIF's are provided. For example, the piping engineer may specify that B16.9 Welding Tees (fittings) be used, but the "crotch radii" and general wall thickness of these fittings, although they DO comply with the Standard, vary greatly from manufacturer to manufacturer. When the B31 “Committee on Flexibility” performed the original fatigue tests, they used only ONE manufacturer’s fittings. If the piping engineer determines that the component to be used in the design is significantly different from those described in Appendices "D", it may be appropriate to develop alternate SIF data. If you are a piping designer/analyst and you have not read all the footnotes associated with Appendices “D”, perhaps you should take the time and read them again.

There is information available in the literature (and in some of the Codes) regarding the development if SIF's. The Welding Research Council has published the following:
Developing Stress Intensification Factors: (1) Standardized Method for Developing Stress Intensification Factors for Piping Components, by E.C. Rodabaugh and (2) Effects of Weld Metal Profile on the Fatigue Life of Integrally Reinforced Weld-on Fittings, by G.E. Woods and E.C. Rodabaugh, June 1994 (26 pp) (ISBN #1-58145-391-4) ? $64.00

To obtain this document, go here:

http://www.forengineers.org/pvrc/wrcpvrcbulletin.htm

Regards, John.

 

[JohnBreen]

And..................

Having said ALL that, pehaps (should someone be interested) something more should be said about the "conservative" and "liberal" allowable stresses and under what circumstances the designer/analyst would chose which is appropriate. Note that the B31 Pressure Piping Codes do not use these ("conservative" and "liberal") terms.

But, only if y'all insist.

Regards, John

 

[JohnBreen]

And.............

I guess what I am trying to point out regarding the original question:

"My real question is, if one calculates the beam stress forces/moments and subsiquent stresses using Caesar II, does one have to check the branch components for failure due to excessive thermally induced loads as well? Or does ASME B16.9 cover this type of loading as long as the code compliance stresses(incl. lib. case) are not exceeded?"

....is this:

When you do your (beam theory) Caesar II analysis your ARE checking the branch components (connections) for cyclic fatigue (alternating thermal loading) life because Caesar applies an SIF in accordance with the Code (this of course assumes you analyze the correct loading cases). Note that B16.9 does not really involve any stress analysis - it is based upon pressure proof testing. There is very little standardization in B16.9 regarding such important issues (to fatigue) as crotch radii or the wall thicness of the component. It (e.g., the TEE) simply must be able to withstand internal pressure that would fail an attached straight piece of pipe of the same schedule to satisfy B16.9.

So, if you choose an appropriate (for pressure) B16.9 fitting and you model it correctly with Caesar II (Caesar II will include the appropriate SIF's), the sustained and expansion (displacement) stresses calculated by Caesar II (in accordance with the Code) would be compared to the appropriate Code allowable stress (and allowable stress range) and if calculated stresses are less than allowable stresses it will be a safe design. BUT, read what the Code has to say about "High Cycle Service" and be sure that you are using a B16.9 fitting that is close in its geometric configuration to the fittings that were tested by the "Flexibility Committtee" when they determined the SIF's (and yes, also remember that the SIF's came from carbon steel fittings).

Also, you MUST understand the difference between calculated maximum primary stresses (e.g., "sustained stresses of pressure and weight", AKA, "additive stresses") and the calculated secondary stress ranges (e.g., thermal expansion stresses, AKA "displacement stress ranges"). The Code addresses thes stresses (stress ranges) differently.

Regards, John.

 

[JohnBreen]

..........and finally (I promise):

Do not compare stresses and stress ranges that are calculated using B31 (modified beam theory) rules to anything else. And do not mix and match allowable stresses.

Using Caesar II, focus on B31 rules, SIF's and B31 allowable stresses (and stress ranges). Do not try to compare this with the results of 3D FEA models (e.g., FE Pipe or Ansys). If you are going to do a 3D FEA analysis use the allowables from ASME Section VIII, Division 2.

If you are going to go into 3D FEA modeling consider what you are telling the FEA program when you develop your model. Think about how an FEA branch connection model typically is described - two perfectly round cylinders intersecting with perfectly monolithic construction (same perfect material with no welds or metallurgical discontinuities). It ain't so!

If you have a fabricated branch connection, you will have to make a sub-model of the intersection of the cylinders to include the geometric discontinuity of the weld bead. Also, to get it right you will have to modify the properties of the metal at the heat affected zone. But manufactured B16.9 Tees are NOT intersecting cylinders either (B16.9 has NO controll on this geopmetry). Many manufacturers B16.9 Tees have spherical "cheeks". And what crotch rdius will you use? Again, there is no Standard crotch radii in B16.9.

Also, you have to do your Caesar II analysis to calculate what forces and moments to apply to the three legs of your FEA modeled Tee. But (unlike B31 bends) the Code tells us (and Caesar II obeys) that the flexibility factor at the TEE is 1.0. It ain't so. Caesar II does what the Code dictates but there will actually be more gross section flexibility in the TEE as it will locally ovalize under loading, similarly to the way a bend responds to loading. And as is the case with bends, additional flexibility is paid for by additional stresses.

There is good reason why the B31 Codes include factors of safety (they are not "fat") - there are many uncertainties in the real world of construction. Factors of safety, to use another term, are indices of ignorance.

Regards, John

 

[TGS4]

JohnBreen - I completely agree with you on every point that you made.

I wish to make one "tiny" modification to your terminology. I believe that the term "factor of safety" or the other variant that you use, give an inappropriate perspective to our fellow engineers and the non-engineering public. Yes, there are a number of features that our designs do not incorporate, and we place adequate margins on these.

Therefore, I prefer to use the term "Design Margin". This implies that there are factors in our design that may not be directly calculated, however, we are applying our "sound engineering judgement" to these matters, and have determined adequate (and hence safe) margins against these matters.

Semantics...

TGS4

 

[rjstephens]

Thanks to all who responded.

I personally prefer to be more conservative in my designs. I suppose it's purely to sleep better at night, having worked on a site for a few years and even with all the response to this thread. Those who prefer to take it to the 'max' - enjoy!

After reading the SIF fatigue test paper by Markl 1952, I don't think people give enough credit to Beskin, who? you say...

Cheers

Rob

 

[JohnBreen]

Hi Rob,

No question about it Beskin's work was hugely important and well know to the four gents from Tube Turns (led by A.R.C. Markl) who performed the testing. In fact, Beskins approximations for the "factor" to be used to "intensify" the calculated moment would result in calculated "effective stresses" much closer to the true elastic stress than those provaded in B31 (but, the SIF's in the Code are appropriate when you use the Allowable stresses in the Code).

There really has been much work reported through the years and that is why we have to be very careful to make sure that we are comparing the calculated stresses to the correct allowable maximum stresses (ranges). The stress must be evaluated within the context of the rules we are applying. There are so many terms to keep straight (e.g., stress intensifiers, stress intesification factors, stress intensities.......).

It may be of some interest to know that the B31 Mechanical Design Committee has an active program for additional fatigue testing of components. Some of this activity is mentioned in Glynn Woods' book on piping design (E.C. Rodabaugh, who was one of the ORIGINAL group at TT, is assisting). So, it is an ongoing learning process and B31 will keep updating the Code to reflect what is learned.

Regards, John.