Skid Piping Boundary Conditions

[mlogeng]

Hi all, this is my first post on here. I am doing stress analysis on a piping skid we designed containing various 6"-.75" pipes ranging from ambient (70F) up to 400F as a design temp. I've got the skid piping all modeled up in CAESAR II with the appropriate conditions aside from the boundaries. The skid is meant to be verified for general use so as to be reproducible. This means seismic and wind loads are not taken into account for this analysis; though, it should be noted that the piping has significant lateral restraint.

My uncertainty comes from how to treat the piping connections/tie points to other (not yet designed) piping. Typically in a stress analysis I would model all the way to some equipment and use an anchor at the nozzle node to determine the maximum loads on the equipment and see if they are under allowable loads (vendor/code specified.) In this case, I have assumed as many anchors as possible, starting with all inlet tie points. Our current piping configuration fails to keep expansion stresses under allowables with every inlet and outlet tie point anchored. I know that there is still some room between my mega-conservative model and reality that I can adjust to get an acceptable piping layout that passes stress checks, just not exactly sure what's right. Here is my best guess:


All of the skid will be connected to other piping which should have some inherent flexibility. Much like allowing for some nozzle flexibility to make a pressure vessel connection model a bit less conservative, I have reduced the anchor on some problematic outlets to the following restraints:

-2 rotational restraints which lock the pipes alignment at the tie point node (for an outlet with +Y flow -> RX and RZ)

-Justification: The outlets are all vertical (Y-direction) with nearby guides so these reaction moments end up small.​

-1 restraint along the pipe axis (for an outlet with +Y flow -> -Y)

-Justification: I use this restraint to adjust the stiffness of the piping connection. While I do not know the direction of the force that additional piping will exert on the skid piping,
the direction which causes overstress on the skid is clear. The force is also likely to be -Y since all outlet tie points are vertical lines exiting the top of the skid.​

The axial restraint with stiffness needs some further clarification. Firstly, the outlet lines have an issue with compressing the vertical run underneath when the tie-point is anchored, since most of them have a sliding dummy leg support directly underneath at the bottom of the skid which offers no Y gap or flexibility. I have noted the free expansion by not anchoring the outlet tie-point, but see little use in this info other than that I know the actual displacement of these tie points will be somewhere in the middle of this and zero. I am treating this support stiffness for each tie-point as an independent variable and increasing its value right up until the piping fails the expansion stress check. At this point, the maximum allowable stiffness, and corresponding tie-point displacements/reactions are recorded.

It's almost as if I can see the skid connections being treated as an anchor in someone else's analysis, whenever the rest of the piping is dreamt up. In that case, the skid almost becomes a piece of equipment with its own "allowable nozzle loads", which helps to show my thought process on providing the client with some sort of allowable. That is the best I can think of, and I would be happy to hear suggestions on modeling something like this. Please do chime in on the tie-point anchor assumption itself and whether it is inherently always conservative as well. Surely a piping redesign is not out of the question, but undesirable.

 

[RVAmeche]

What exactly is the question? A sketch or screenshots would help.

Since you're designing the skid itself, you could stop your analysis at the piping boundaries and iterate on what external forces/moments are acceptable from the external piping. Hard to evaluate on the support changes within the skid just from text.

 

[mlogeng]

@RVAmeche You are right I have gone on to ramble about this system hoping everyone else has spent the hours I have. Here is two snips of the CAESAR II skid model.

The first pic shows the full skid with the temps in each line and restraints are hidden:

The second pic is a close-up of the corner of the skid having expansion stress problems when the 3 vertical pipes (shown without outlet anchor) are modeled with outlet anchor:

I am specifically wondering about what type of boundary conditions should be enforced at those 3 outlets, which is still a conservative assumption, and can lead to approval of this pipe design from a stress analysis perspective.

I am less-specifically wondering about different engineers' general experience and go-to practices regarding the boundary conditions for their stress models, as I am a newer engineer to the field.

 

[LittleInch]

Hmmmm,

My experience of buying skids, especially those produced already or not specific to my project, is that all of the connections are usually essentially anchored / restrained close to the tie in points and then usually I'm provided with a set of forces and moments which are acceptable ( usually pretty low....).

For the skid vendor, doing anything else is leaving you open to abuse from unknown future third parties, so I understand that. You really need to have a boundary that you deal with all the stress shit inside that boundary and you let the other designers deal with it outside.

So yes, fix it at the tie ins, then apply some forces x, y and Z and moments and see if the flange or the pipe fails. Then reduce that and call it the maximum forces that the skid and your anchors can take.

You have NO CONTROL or knowledge for how this skid and its connections will be used by others so you can't assume anything and stay in business. That's why you list the acceptable tie in nozzle loads and moments. So yes, your current layout may not work. You need to make it work.

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[1503-44]

I agree with Little Inch, but have just one other comment to add concerning the practicality of how to determine the loads.
You can't fix it at the tie-ins themselves, as all loads from both sides will just be adsorbed into the anchors. Do something like the vessel and pump manufacturers do. That would be to back away from the flanges to where you can put an anchor on your pipe, or make the assumption that the pipe is fixed at that point. That point might be the pump casing, or at a thick vessel wall that a nozzle is attached to or some kind of relatively quite rigid pipe support. Then you can add loads to the tie flange, increasing them up to the point when the flange or pipe from flange to your anchor point fails. Then you could state those loads as the maximum allowed.

--Einstein gave the same test to students every year. When asked why he would do something like that, "Because the answers had changed."

 

[mlogeng]

Thanks for the responses. LittleInch you are right, with no knowledge of the off-skid piping I need to provide allowables. I agree with 1503-44 that if I anchored the tie points and then applied some forces/moments to that same node, it would not affect the piping aside from the reaction at that node. moving the anchor to the closest skid support and then applying the forces/moments to the tie points makes physical sense to me, and it gives me some failure criteria since all I have to go by is allowable stress. In that case I will need to make sure my civil guys know that those supports should be more than a U-bolt, correct? Even then I should supply them with the maximum experienced loads on the anchors corresponding to the tie-point maximum loads, so that we can be sure the skid structure is okay (through FEA I imagine?) I admit, I am quickly diverging from my area of expertise.

The piping will likely need a layout change since we are to enforce an anchor at the tie points both in the model and physically. The straight 300F/400F pipes running from anchor to dummy leg just ain't it. I'm sure it wont be too hard to add Y-direction flexibility, just an off-set or something. Maybe even a sliding spring can instead of a dummy leg. Now I'm just going crazy. Need to wait for my Senior Engineer to come back from vacation.

While I have you guys around, some other quick questions:

1) For U-bolts I am modeling the support as one restraint (typically +Y) with friction (mu=0.3) representing the skid steel, and all other directions perpendicular to the pipe axis have a restraint with a 1/16" gap. Does this seem reasonable?​

2) Does anyone have experience with PCF imports to CAESAR II from Smart Plant 3D? I began this skid job with all the files sent by my designer, but only some of the lines would load in through the advanced PCF import, with the others causing the import function to crash causing me to have to model them myself. I was unable to identify a failure pattern aside from the file size which seems too obvious.​

 

[1503-44]

I'd tend to ignore u-bolts completely, unless they have a very high relative effect on the piping they support. I tend to evaluate "relative" effects. A u-bolt on 1/2" tubing is important, but a u-bolt on a 6" pipe will be deformed in any manner that pipe chooses to do it, therefore might as well not be present.

Same for the support. It does not have to be a 100% rigid structure to be an anchor. All it has to do is not move very much in relation (relative) to the piping around it. If say, if you apply a unit load to the support and it moves half(?) the distance that the local piping moves when under an equal but opposite unit load, then maybe you have achieved a reasonable anchor. A mushroom might be an effective "100%" full anchor if placed in the midsts of a flexible soda straw arrangement. So as long as the support is well stiffer than the pipes in the vicinity, you'd have an "effective full anchor". Just decide what ratio (half, 1/3, 1/4) you think is effective enough to be considered rigid. It might be a useful concept when you want to simplify an initial design.

In lieu of giving a force to the support designer, you might give them a deflection, or both. If you give them one, if they really know what they are doing, they will ask for the other.

All that keeping in mind that sometimes nothing on a skid can be an anchor, if the skid is not anchored to the pavement, or welded to the platform deck somehow, or if its just sitting on the ground with no anchor, or slotted anchor bolt holes, just waiting for any one pipe to push it around. It can get to where all is relative to skid friction on the concrete at times.

--Einstein gave the same test to students every year. When asked why he would do something like that, "Because the answers had changed."