I cannot find any references in either the Technical Reference Manual or the Application Guide on how to model and code straight pipe crosses nor how to model and code a pipe section with two weld-o-lets and attached pipe with coinciding centerlines (180 degrees apart). This is equivalent to four pipes framing into an intersection. Can anyone help with coding or SIF's. Thank you

A "cross" is not a Code fitting. You won't find it listed in Appendix D of B31.1 or B31.3. There are no Code guidelines (that I am aware of) for the computation of the SIF value for a "cross". Therefore CAESAR II which follows the piping codes as close as possible, can not compute an SIF for this fitting, and this is why there is nothing in the documentation.

The only suggestions I can offer are:
- contact the vendor of the fitting and see if they can offer any SIF data
- perhaps employ a finite element analysis and determine the value yourself.


------------------
Regards,
Richard Ay (COADE, Inc.)

Hello Mr. Patterson,

As Rich has pointed out, the B31 Codes do not prescribe SIF’s for crosses but SIF’s can be developed by employing FEA techniques. More trouble than it is worth really. When Markl et. al. began fatigue testing fittings, they did not have crosses in hand – so no test were done. I don’t believe that Glynn Woods and Ev Rodabaugh intend to do any testing of crosses in the current fatigue tests either.

It is a very rare occurrence in design when using a pipe cross cannot be avoided. Using B16.9 welding tees – fitting-to-fitting – is usually the “work-around” in piping layout. Some good reasons for avoiding the use of crosses in the design include: 1. System flow characteristics are compromised at the cross; 2. Since there is more “opening” at the cross than there is metal, it is structurally weak; 3. In the case of fabricated crosses, the “area replacement rules” for pressure design are inadequate when the design pressure is significant (this can be shown by FEA); 4. The body of design knowledge and experience is limited (e.g., there are no developed SIF’s); and 5. For sizes greater than NPS 6, B16.9 crosses are usually a special order item – not readily available.

The limited availability is usually addressed by fabricating crosses. It would be wise to take a long look at a B16.9 welding cross before fabrication (or pressure design) is attempted. The manufacturers of B16.9 crosses provide generous amounts of reinforcing metal in the “crotch area” and usually provide the largest crotch radii that are consistent with the standardized “center-to-end” dimension (neither crotch thickness or crotch radius is standardized by the B16.9 Standard). After looking at the B16.9 welding cross fitting (especially the crotch radii), consider the stress intensification that occurs at the intersection of (nearly) cylinders. Consider the difference between the SIF for a B16.9 welding tee and the SIF for an unreinforced fabricated branch connection. Not a pleasant or comforting thought.

For fabrication, the layout, cutting and pre-weld fit-up must be near perfect. It might be better to start with a wall thickness of at least one additional (to adjacent pipe) schedule even before considering the pressure design “area replacement” requirements. Extra metal in the wall would also help reduce the local heat warping, and facilitate the welding. Carefully write the welding specification. For fabrication of crosses of diameters greater that 14 inch OD, alternate (to area replacement) reinforcing schemes should be investigated. If the operating temperature is not significant, some of the designs shown in AWWA M-11 might be considered. NASA has successfully employed “C” type external reinforcing plates at all 4 intersections in some pipe cross designs for compressed air systems (some of these designs have small diameter out-of-plane reinforcing pipes through the intersection to hold the shape). In other government facilities large spherical intersection are used at pressures up to 600 psig (under pressure the intersection “wants” to form a sphere). I would not consider the design finished until an FEA was done to confirm the adequacy (it will be necessary to sub-model the weld areas). Remember there is nothing in the Code (and therefore it is not addressed by CAESAR II) to guide you in pressure design of crosses. So pressure design is up to the design engineer and I strongly urge you to refer to B31.3, Paragraph 304.7.2.

Modeling techniques will vary as a function of the OD of the pipe involved. For sizes up to NPS 10, I usually model the cross just like it was a branch connection with an extra branch. For sizes larger than NPS 10, I model the straight run (or “through pipe”) as I would a branch connection but I use “rigid links” (twice)out to the OD of the run and pick up the branch outlet pipe size from those points (applying the SIF at the beginning of the “outlet” branch pipes). The moment at these points will be quite different than that calculated at the branch intersection (of centerlines) point and this technique allows the calculation of different moments for each “branch”. Regarding SIF’s, I think that regardless of the type of construction, I would use (as a minimum) the SIF associated with an unreinforced fabricated branch (UFB) connection. If fabricated construction is used, SIF’s up to twice that of the UFB connection should be considered (your FEA should guide you here). I should caution the reader that I cannot quantify these observations concerning SIF’s and I offer them just to reinforce my comments regarding conservatism. Perhaps Rich can share some other modeling experiences.

In my opinion, it is worth a design compromise or maybe even a complete redesign to avoid using a cross. If this is unavoidable, be conservative.

Of course all the above is only my opinion and it is not necessarily the opinion of ASME International or any ASME B31 Code Committee.

Thank you for bringing up the topic and good luck with your project.

Best regards, John.


[This message has been edited by John Breen (edited February 07, 2001).]

[This message has been edited by John Breen (edited February 07, 2001).]

Another method for modeling branches on large diameter pipe is to use the "element offset" option. This method puts the SIF value on the proper nodes, yet utilizes the proper branch length when computing the element stiffness.

------------------
Regards,
Richard Ay (COADE, Inc.)

Dear Mr. Richard,

Could you please share more detail of the "element offset method" that you mentioned regarding modelling cross tee. Kindly share the reference pages in Caesar II Manual for this method.

Thank you.

To find the documentation on "offsets", click on the offset checkbox (just below the Delta X, Y, Z fields), then use [F1] on any input field.

Assume your 24" header runs from 110-120-130 with 120 the tee (cross) node. When you model the two branches, say 120-600 and 120-700 you want to use an axial offset of 12" (half the diameter of header).

I have evaluated crosses using FEPipe but only for FABRICATED crosses not tee type fittings with nice smooth crotches. You basically model a cylinder and then add two nozzles 180 degrees apart. You have to use slightly smaller OD branches for a full size Cross (usually use the ID as the OD of the branch) to get everything to mesh properly with the use of some user defined angles and rings in the optional input.

This typically is conservative for fittings and is better than assuming Code tee type branch connection SIFs. I always would take the "nozzle" forces/moments from CII and put them in the FEA model for the "final" stress evaluation since the FEA model is built.

I'd argue that you could use the proposed method to approximate stresses in smooth forged crosses conservatively, but would be nervous about using this on an actual fabricated cross made with 4 pipes cut at 90° angles without a safety factor.

Instead of marrying the results to ID, I would use a static value, such as subtract 0.01"-0.05" from the radius. This is similar to attaching trunnions to elbow extradoses in FEBend. If you push it as far as the software permits, it will often diverge to unrealistic numbers. Make it too small, though, and stresses will be under-reported.