Decoupling a pipe model 2

[callmebob]

I am having doubts on the way co-workers suggest when to decouple or where to finish a model to make it small in size to analyze. The way I am doing this is to include a substancial length of piping of the analysis I am not analyzing, this will give me a very accurate feel for what that line will do to my current analysis.
I do not agree that just because at a Tee connection, if the moment of inertia is >10, that I can put an pseudo anchor (retrains lateral, axial, vertical) and apply the thermal displacements of that pipe (header)from previous analysis. The reason is that this will not accuratelly provide the displacements of the header under wind conditions.

Any comment or documentation I could read will be appreciated...

 

[NozzleTwister]

Bob,

Your right about the displacements. The dispalcements you apply to the terminal branch for the de-coupled branch analysis need to match load case wise. Thank goodness with CAESAR you can input up to 9 different displcement conditions.

There was just recently a thread on COADE's CAESAR forum on this subject. The link is below.

http://www.coade.com/cgi-local/ultimatebb.cgi?ubb=get_topic;f=1;t=001013

NozzleTwister
Houston, Texas

 

[callmebob]

Thanks NozzleTwister,
Yeah, that true, I could input all the other displacement conditions. I am still troubled thinking why should I decouple at the tee intersection, why not let the software do the analysis on the Tee as well. Is it acceptable to include part of that header which has been analyzed separately and neglect that part of the analysis? Basically I include reference piping to a point I know it limits the movements at the branch, i.e. a line stop, nozzle connection, corner with a guide, etc. This same approach will then be done to the referenced header analysis, but in a different model. AT the end I'll end up with a overlap of nodes around the branch for which both models pass the allowables.

 

[NozzleTwister]

There is no rule that says you must de-couple the branches from the header and do a separate analysis for them. It's perfectly acceptable to have they entire system in one model. That choice is yours. The bottom line is that you need to make the decision about what is the best way to go for the system in question. To overlap part of the header analysis with the branch and vice versa is accepable too.

Today the software is so much more powerful than it was some years ago. Today it seems there to be no limit to the size of a problem.

The idea is to analize your systems with reasonable accuracy and also model so the affects of assumptions, de-coupling, etc. would error to the conservative side.


NozzleTwister
Houston, Texas

 

[StressGuy]

Nozzle Twister is correct - the classic reason for breaking a system into sections was that the software could only handle so many nodes. Now, that's not really a concern.

Still, I and my associates will typically still try to identify break points in systems for several good reasons. For one, you can put more people to work on smaller packages. For another, the piping design may not be complete when you start on the analysis, so you want to isolate the unknwown areas from what you can work with. And, I think most importantly, despite the thousands of nodes that the machine can handle, there's still only so much that we can we can process when looking at a problem, especially when it comes to pouring over lists of elements and looking at big graphical plots.

Edward L. Klein
Pipe Stress Engineer
Houston, Texas

"All the world is a Spring"

All opinions expressed here are my own and not my company's.