Modeling anchor points in pipe racks 2

[garfio]

I would like to know what is the common practice followed in modeling anchors located on pipe racks: Are the anchors normally model with a stiffness less than "rigid" to represent the deflection that the structural steel will suffer under the loads or they are modeled as "rigid"? What values of stiffness could be typically used? Is there a maximum acceptable displacement in the pipe rack that is given to the structural engineer in order to be consistent with the stress modeling?

It seems to me that if there is not an assumption made during the piping stress analysis, once the structural steel is designed, another stress run will have to be made using the deformations obtained by the structural engineer. This issue turns more relevant when equipment loads can be affected by the steel deflection.

 

[DSB123]

Garfio,
You may well have to perform some iteration in arriving at a sufficiently accurate analysis. Initially you could assume that the anchor is "rigid" but as you are no doubt aware nothing is ever "rigid" If the steelwork is already there ask the structural engineer for some stiffness values for the anchor point (and also for other supports/guides on the pipe rack - if they are significant) The more accurate your model the more accurate are the results. After assuming the anchor is "rigid" give the loads to the structural dept and ask for the deflections/stiffnesses and if significant add to your model. Anyway why are you using anchors in a pipe rack? What about a combined support/linestop and guide which will control the translational deflections of the pipe?

On another point the structure/pipe rack could be modelled using Caesar which is probably the most accurate approach.

 

[garfio]

Thanks for the prompt answer.

I wrote "anchor" to define my general question (worst case). Linestops and guides are also possible, but in any case there will always be forces transferred to the structure.

In this particular project, the output from the stress analysis will produce the information to be given to the structural guys, which are a third party, so the iteration is not a a desired situation.

I was wondering if an acceptable practice could be any of the following:

- Consider the structure rigid at the support points

- Assume a specific rigidity for the structure (based on experience)and provide that input to the structural group


As this interaction between structure and piping obviously always happen, there should be an accepted procedure out there about how to deal with it when the piping group does not have to model the structure. Is this iteration always required?

 

[SNORGY]

You can model the structure with the piping via EDIM (CAESAR II), etc. - that works quite well.

I find that I get good results treating the "anchors" as "line stops & guides" with appropriate gaps inserted, since it will be unrealistic to fabricate to a "zero gap" tolerance. That sometimes gives rise to non-convergence in the model due to the non-linear restraints, which I find can often be solved by specifying "zero gap" in the "Y" (vertical) direction or by slightly changing the friction coefficient at each support rather than having a blanket constant throughout.

You could apply elastic beam theory to compute what beam deflection would be considered acceptable under a given applied point load, and then take the load divided by that deflection to compute a spring constant, and then feed that back into your restraint field.

But...

I'm thinking that, if your forces and stresses are too high assuming rigidity of the restraints after gapping, then you probably don't want to compensate by taking credit for the elasticity inherent in the structure. You usually want to get more flexibility into the piping somehow.

Regards,

SNORGY.

 

[steve1]

I have engineered many pipe racks for power plant projects where the rack carried both main and aux steam piping. I also have experience with doing pipe stress work using Ceasar, Algor, Autopipe and other software.

My understanding is that if you model a support as "rigid" then the piping software will assume a support stiffness of 1 x 10E6 pounds per inch (a million pounds will cause 1" of support deflection). This is unrealistic for most, if not all, pipe rack steel frames. In addition the use of "rigid" supports will also produce unrealistic piping reactions that will force the pipe rack into being much larger than necessary.

If I needed a value of pipe rack stiffness I would use the following (assuming a rack not greater than say 15' tall):

For lateral direction on pipe rack frame assuming 2 legs on the rack I would use 1x10E5. For a single post support I would use 1x10E4. For the longitudinal direction for two legs 2x10E4 and for one leg 1x10E4.

 

[Crusader911]

As has been pointed out, what you typically find in a pipe rack are directional anchors (also called limit stops) rather than actualy 6-degree-of-freedom anchors. I usually don't included the pipe rack deflection because to really know what the rack is doing one would have to know what lateral loads are imposed by all the other pipes on that rack, whether that's from friction or limit stops. The flexibility can be added as described in the other posts, but in practice I have only done so when I don't like the answer that the rigid support gives me.