During the construction phase of a project, one of the contractors decided to install heavier repads for some of the Tee’s and even decided to replace some of these fittings with tees of a heavier schedule.
While he assumes full responsibility for the changes, by stamping the documents, I’m trying to asses what other aspects (consequences) he may overlook with the “upgrade”.
This is what I could see as potential issues:
- using heavier schedule fittings with lighter schedule pipe may increase hydraulic losses beyond what the process group factored in for their calculations;
- overall flexibility of the piping systems will decrease resulting in increasing the equipment loadings beyond acceptable limits;
- SIF’s will change, rendering the local stress evaluation done in CAESAR as erroneous;
- in fact, by modifying the wall thickness at these intersections, the contractor renders the entire analysis done in CAESAR as inaccurate for all the lines subjected to these changes.
His motivation for the modifications lies in the fact that FEA run on the original configuration resulted in failed members (tees).
I have no means of checking that but I know that the materials engineer on the project has used the code to determine wall thickness requirements for pipe and fittings, including tees, elbows, etc.
Please comment.

You seem to have most of the key issues covered.

Note that FEA of a tee provides calculated local stresses that are not necessarily related to the Code allowables. There is no "Code allowable" for a purely local stress, which is what you get out of an FEA.

This is not to say that FEA is an invalid approach to the design of piping components. It just provides more information than is necessary.

Hello gabionex,

The slightly reduced inside diameter (flow area) will not likely be a concern (there will be some design margin that will be reduced).

Your erroneous CAESAR II structural analysis will error on the conservative side. For internal pressure design, the B31 Codes accept "listed components" of the correct pipe wall thickness (schedule) without further analysis. The design of these fittings is based upon burst testing of the fittings and not based upon calculated stresses.

If B31 Pressure Piping prescribed allowable stresses were used in the FEA evaluation, I am not surprised that the FEA of B16.9 Tees indicated that the stresses in the fittings "exceeded the B31 allowable stresses". When FEA is used, close your B31 book and put it on the shelf and open you ASME B&PV Code, Section VIII, Division 2. We cannot mix analysis methodologies (nor can we mix their allowable stresses). Remember that the B31 Codes use beam theory ("patched-up with flexibility factors and stress intensification factors") and the ASME B&PV Codes use shell theory. Local membrane bending stresses are, as Craig pointed out, not a beam theory capability. Then again, if you stick strictly to the rules of the B31 Code it is not an issue.

Regarding the B31 Pressure Piping Code evaluation of Tees: at bends and elbows (with B31 rules) we use flexibility factors and stress intensification factors while at Tees we use stress intensification factors and we assume the flexibility factor to be 1.0 (e.g., as straight pipe) - this is very conservative. We know that the flexibility factor at a branch connection is less than an elbow but more that straight pipe. So really, your piping system will have MORE flexibility that the CAESAR II analysis indicates because CAESAR II always uses the rules and equations presented in the B31 Codes. CAESAR II does NOT vary from B31 rules.

Also, I wonder how your contractor modeled the B16.9 Tee in FEA. Where did he get the geometry that he modeled? There are VERY few geometry issues that are controlled by B16.9 (thicknesses at any points, shapes, crotch radii, etc. are NOT "standardized"). The only way we have ever been able to get the actual data is by physically sectioning an exact fitting as provided by the exact manufacturer and taking our own measurements - no two OEM's products will have the same dimensions. So, if your contractor modeled two perfect cylinders (uniformly exact diameter and wall thickness) intersecting at the centerline intersection (even if he went to the detail of modeling some sort of crotch radii) the FEA model will not be accurate. The real question is how much "accuracy" is really needed - that is why we use "design margins" (factors of safety are really indices of ignorance).

Regarding fabricated branch connections (not B16.9 "Tees"): you will find that if you design a fabricated branch connection according to the pressure design rules of a B31 Code and you have the correct reinforcing, an FEA of the fabricated branch connection will indicate that the pressure loading results in stresses greater than Sh as used in the B31 Code(!!??). Also, developing an accurate FEA model of such a fabrication is not very often done. Think about how complicated the model would be with the load path going through fillet welds and with there being "contact elements" (not contiguous material) between the reinforcing pad and the OD of the pipe (most FEA modelers simply thicken the pipe wall thickness at the reinforcing pad). Also, think about how the material properties would have to be altered at the heat affected areas of the weldment. Likely, the FEA model is "an approximation".

So, your compulsive-obsessive contractor performed needless analyses of these components and his analysis cannot be directly compared to your CAESAR II B31 analysis performed by your firm using any rational approach (the comparison is of academic interest only). Where does that leave you? Well, review tour firm's analysis and be sure that it was faithfully done in accordance with the contract design Code of record (i.e., B31.X). When you find this is the case you have fulfilled your contractual requirements. If there is any concern regarding the final "as built" system, simply modify your CAESAR II model (save as) to represent the "as built" system and rerun the analysis. It is not likely that you will see any problem (except for the possibility that the spring hangers are slightly "off"). But even then, this "new" CAESAR II model cannot be directly compared to the FEA model.


Regards, John.

Thank you very much; it is a privilege and an honour to receive consideration and such responses.
I personally discarded the results of the FEA (conducted by the contractor), simply based on the fact that no external loads were provided to him.
I suggested the use of the B31.3 design requirements with regards to calculation of thickness and total repad area for branch connections.
However, he run FE Pipe and used what he claimed “as reasonable values” for forces and moments.
That brings an interesting point because, reading your posts, he can claim B16.9 as the qualifying code for fittings and then his drive to conduct FEA becomes legit. That will complicate things even further because, here in Canada, these fittings require a CRN issued before installation.

One issue is not mentioned in the posts. If the line is high temperature line making the fittings thicker may cause problems with thermal stresses during heating up and cooling down. Thicker items heat up slower. Heavier pads may be even more difficult to fit than those asked for. End result is cooler pad/slower to heat up and high stresses in the connection weld. Both cases can be a real problem in creep range.

ASME VIII Div 2 is a complex and comprehensive code for FEA. Light weight add on packages give only equivalent stress (von Mises) or equivalent and maximum/minimum principal stresses. For quick checks on weld on brackets, dummy branches and similar comparison to code allowable should be OK. Any comments?

Does anybody know 2 engineers who will get same FEA result on a problem if they work independently?

"Does anybody know 2 engineers who will get same FEA result on a problem if they work independently?"

Brilliant absolutely a stellar comment... therein lies part of the problem.