Hertz Contact Stress in Pipe Support 1

[SteveMcQ]

I have an FEA model of a generic pipe support problem I face and I am looking to see if anyone has any insight as to how to possibly run calcs by hand? Using Roark's for stress on a hollow cylinder as a line will fail the pipe due to local stress but an FEA model shows a lower stress as the distribution is not a line.

 

[tbuelna]

Your situation may or may not present a problem. You need to be careful about how you interpret the results of the static hertzian analysis approach you used from Roark. Even though your hand calcs show the pipe surface locally yielding initially at the "line contact" with the support, after the initial yield of the pipe surface the contact area will increase and the contact situation will stabilize. Basically, what you'll be left with is a local pipe surface having a compressive pre-stress condition due to the mechanical cold-working, and some localized stress concentrations around the perimeter of the deformed contact area at the pipe surface.

What you need to consider is whether the local combined stress conditions existing at the pipe/support contact surface post yielding actually presents a fatigue or stress problem.

 

[SteveMcQ]

That's where we are at. The Roark's equations blow it outta the water so we went to FEA to see if there may be a chance at removing some conservatism in the Roark's equations but we're seeing high local stresses that may or may not need to be considered.

 

[tbuelna]

With a static contact analysis, your hand calcs may give results as good as your FEA. Can you provide details of your FEM/FEA approach? Also, what requirements do you have for conducting your analysis? What are the analysis factors required for things like contact bearing, misalignment/load distribution, Kt due to surface defects, etc.

 

[SteveMcQ]

Right now just looking at a 20" pipe supported with a 10" HSS with a plate in between. Load from the pipe is 140 kips and using the Roark's equations the pipe cannot withstand the bearing load. Was hoping FEA may gleana little more into this using a contact analysis in ANSYS.

 

[tbuelna]

A 20" OD steel pipe bearing on a flat steel plate with a 140Klb radial load does not sound too bad. Were the Roark equations you used based on rigid solid bodies, rather than a hollow pipe cylinder bearing against a beam supported at each end?

 

[SteveMcQ]

That they were. The problem is with FEA we're getting some high contact stresses that may be real but are struggling to determine how to handle them.

 

[tbuelna]

An FEA of a surface contact is always difficult, since it tends to be non-linear in nature. But apparently you modeled the contact with ANSYS, which should have given a fairly accurate result, right?

The only other variable that would have a significant effect on your analysis results would be the boundary conditions used. The structural stiffness characteristics of the adjacent pipes and supports could have a significant effect on the contact bearing stresses between the pipe surface and the support bracket.

 

[SteveMcQ]

The result seems to be real and the distribution matches with hand calcs and the peak is just below the surface so I would argue they are accurate.

I'm starting to believe that the supporting structure is far more critical than initially believed. It is rigidly supported currently. I may model the HSS underneath it instead of the plate to see what we get in terms of a reduction in stress.

 

[Doove117]

I'm no stress engineer in the slightest, but typically I would be looking to support the pipe with a saddle, or weld a bearing plate on the pipe to reinforce the contact area. Remember there may be other factors like thermal movement or vibration.

 

[dhengr]

SteveMcQ:
140kips sound like one hell of a pipe reaction, on a flat plate support. What’s in this pipe and what is the spacing btwn. supports? I don’t normal do much pipe analysis, but I think you might be going about this the wrong way. What’s causing that kind of gravity reaction? Are there other forces on the pipe, in play here? Show us some sketches with loads, forces, and dimensions. They’ve been supporting pipes on beams, on pipe racks, for years and haven’t had this kind of problem, what are they doing right that you are doing wrong. I haven’t even bothered to look at Roark’s formulas tonight, but I suggest you read some of the verbiage around the formula that you have applied, so you have a vague idea of what the Hertz stress is all about. The upshot is that you have to do something to alleviate that line load on the pipe. What do the pipe dia. on a flat plate have to do with the problem? Why would you use a saddle of some sort, why would you put a thick rubber pad, or some such, btwn. the pipe and the saddle?

 

[SteveMcQ]

They are emergency loads in a power plant and the loads are the loads, enveloped from a variety of conditions out of my control. A FBD would just the pipe on an HSS frame support with a plate between the pipe and HSS if need be. If a better bearing stress answer cannot be obtained through FEA, which I suspect may be the case, a more involved support such as a saddle is in store.

 

[Duwe6]

Look at an ANVIL #439 support/saddle. Easy/cheap to make, spreads the load out a lot. Might work for you. If not, use something like these:

http://www.pipingtech.com/blog/2007/06/13/42-large-bore-pipe-saddles/

 

[racookpe1978]

I'm no stress engineer in the slightest, but typically I would be looking to support the pipe with a saddle, or weld a bearing plate on the pipe to reinforce the contact area.

"Theoretically" you do have a flat plate supporting a "flat" surface (the HSS upper wall under a flat plate) supporting a round pipe, so without local deformation of the "round" pipe, you would see very high local stresses. Which is what that local deformation is "responding to" as the pipe wall yields slightly.

I agree with the several observers above who recommend a off-the-shelf simple commercial pipe saddle to spread the load out both radially (rigidly wrapped around the pipe) and axially (the saddle will have slightly longer support surface down the length of the pipe). otherwise, your only contact surface is the width of the HSS wall and the deformed "spread" of the pipe wall against the (unreinforced, itself-subject-to-bending) thin flat plate: That would be a final area of 1/8 inch, maybe 3/16 inch of contact surface? To paraphrase what one writer here uses in his signature, "When everybody does it that a way, there is usually a reason why everybody does it that a way ...".

 

[TGS4]

Interesting discussion. I'm going to take a slightly different angle on this - what is the criteria that you are using to determine if the stresses in the contact region are "acceptable" or not? What failure modes are you trying to protect against?

 

[KevinNZ]

You will not see pipe supported directly in our industry (geothermal). We always use pipe slippers/shoes.

We carry out FEA on the pipe shoes and even they have high stress in the pipe. There is no way we could support the pipe with out a shoe.

After a few thermal cycles a directly supported pipe will wear and lose wall thickness.

Kevin

 

[SteveMcQ]

Not sure if an off the shelf saddle is an option. Trying to meet ASME code for a power plant and trying to investigate FEA vs hand calcs to see if there is any conservatism that can be removed to gain a little more margin before changing support type.

 

[dhengr]

SteveMcQ:
The Hertz stresses are real, they don’t go away. As the pipe or the HSS wear, the bearing line becomes wider and the bearing stress goes down. The FEA program will show these high stresses too. I think what Kevin might be talking about as a shoe, in its crudest form might be; take a 16" length of the same 20" pipe, make two longitudinal cuts on the pipe so you end up with a 90̊ arc of pipe. Clean this up, shape the corners with nice radii, press this to the pipe and weld it to the pipe as a wear pl. Weld the other wear pl. to the HSS, and let em wear.

 

[DSB123]

Hi All,
Have you ever noticed how pipe support Companies are eager to sell their pipe shoes/supports and quote allowable loads for these items. However it is up to the piping engineer to check what the pipe supports do to the pipe and what "local" stresses are developed in the pipe wall. There are many piping engineers that do not consider this. They select the pipe support from the catalogue - check the actual loading is below the allowable in the catalogue - and hey presto specify that support. No checks of how the loading is transferred between the support and the pipe wall - no check of the local stresses in the pipe wall and combine these with other stresses developed in the pipe. End of gripe!!!

On the present subject. There is a local stress due to the contact between the two items and generally the softer of the two items yields slightly until the surface area is sufficient to sustain the contact stresses. So there is some indentation of the weaker material. This indentation is normally checked by assuming a "reasonable" bearing length (which is small). It should be remembered that this indentation can affect the "free" movement of the pipe in certain cases. Using contact stress formulae will result in unrealistic stress levels due to the actual "local" yielding.