Snubber

I have a system with seismic and snubber and I made load cases with next seismic condition:
Horizontal load = 0.3 g
Vertical load = 0 g

Problem is how to properly define load cases if I have snubber in the system.
I made next load cases (only those important for seismic):

1 W+D1+T1+P1+F1 (OPE)
2 W+D1+T1+P1+F1+U1 (OPE)
3 W+D1+T1+P1+F1+U2 (OPE)
4 W+D1+T1+P1+F1-U1 (OPE)
5 W+D1+T1+P1+F1-U2 (OPE)
6 W+P1+F1 (SUS)
7 LC7=LC2-LC1 (OCC) Algebraic method
8 LC8=LC3-LC1 (OCC) Algebraic method
9 LC9=LC4-LC1 (OCC) Algebraic method
10 LC10=LC5-LC1 (OCC) Algebraic method
11 LC11=LC7+LC6 (OCC) Scalar method
12 LC12=LC8+LC6 (OCC) Scalar method
13 LC13=LC9+LC6 (OCC) Scalar method
14 LC14=LC10+LC6 (OCC) Scalar method

Load cases 7,8,9 and 10 give displacement only due to seismic load.
Load cases 11,12,13 and 14 are standard Occasional load (Sustained load at NO + seismic load).

The question is. If I use load cases 2,3,4 and 5 as OPE, the snubber is not active. If I use these load cases as OCC, the snubber is active.
But!!!???
In this case load cases give me displacement on the snubber equal 0 (zero). It can not be correct, because I have other loads it the system. In the same time for load cases 7,8,9 and 10 I have displacement on the snubber different than 0 (zero). For load cases 7,8,9, and 10 displacement should be 0 equal (zero).

Does anybody have some suggestion how to solve this problem? How can I make correct model with seismic and snubber?

Best regards!

How are you Zdravo?,

Your question is a good one. The answer however is not exactly what you might want to hear! First of all the snubber in CAESAR II was never intended for use in static loading situations. I suspect that it was put in during the glory days of Nuclear power work here in America.

As such the popular way to deal with seismic loads then, was dynamic analysis. This is the way, which this works. What do I mean works? Well first we need to discuss how a true snubber works.

A snubber is intended to allow the quasi-static thermal displacements to occur freely. However, when a dynamic load is applied they hopefully become fully locked up in a couple of milliseconds. The way in which CAESAR II “locks up the snubber” is the use of (OCC) in a load case. When the OCC is used the snubber becomes a fixed boundary condition. The problem when using this in a static case when thermal displacements are present is that the snubber will be “locked” and will restrain the pipe thermally as well as for the statically applied occasional load.

So what can be done? I suggest the following approach…
1) Place a node(s) where the snubber(s) is to be attached to the system.
2) Run the required operating cases i.e., W + T + P + (F for spring can if req'd) + D (if req'd)
3) Note the displacements linear and rotation that occur at the snubber node(s).
4) Call out snubbers attaching them to the piping node(s) and then cnode them to a fictitious node.
5) Then go to the displacements input screen and input the operational displacements for the cnode.
6) Finally set up your load cases….
CASE 1 (OPE) W+D1+T1+P1

CASE 2 (SUS) W+P1

CASE 3 (OCC) W+D1+T1+P1+U1(Case for algebraic summation)

CASE 4 (EXP) L4=L2-L1

CASE 5 (OCC) L5=L3-L1 (Results in U1 loads/ stresses alone)

CASE 6 (OCC) L6=L2+L5 (Weight + Pressure + U1 loads)

When you look at your restraint summary you will find that the snubber only experiences loading during the application of the uniform load and yet allow the system to be free thermally. I hope this helps I was going to write this up for the newsletter to include sketches etc.

Hello all,

I have the same problem as Zdravko when defining the load cases in a non-linear model with snubbers and static seismic forces. Has there been any change in CAESAR during these 4 years or the only way to define it is as John says (note the displacemets, create a c-node, etc..)?

By the way, any newsletter handles this topic?

Best regards,

No changes.... what was applicable 4 years ago is still applicable....

I am working on an article which as an aside will discuss topic this but I need to finish it....

I strongly suggest you consider using gapped restraints in lieu of a snubber. The gapped restraints may very well be sufficient for your purpose.

I resume this old post... I don't understand the need to use C-Node. I tried this method with and without c-node to the snubber, I obtained the same results for all the cases which is obvious, results were satisfying too because (I tried with wind load) when the OCC load was acting snubber did its work. I didn't insert any data in snubber stiffness. Maybe I miss something....

In this application, the CNODE permits you to displace the "other end" of the snubber. If you don't do this, the snubber is assumed to be fixed to a "fixed point" in space.

If you leave the stiffness field blank, then c2 uses a "rigid stiffness value", which defaults to 1E12 lb/in.

Ok, many thanks, I think I am getting it. Why in his explanation he puts D1 in the OPE case
"CASE 1 (OPE) W+D1+T1+P1"
??
This D1 should be a consequence, an effect not an imposition. I mean that from the OPE case you get D1 and you apply it to the OCC case L3.
I think this formally uncorrect but it doesn't make any difference from a results standpoint.
Did I say right things or i am still missing something?

Ordinarily you would be correct, the D1 would be a consequence not an imposition. However, snubbers are different - you only want them active in the OCC event. So you can't setup a load case such as W+T1+P1+"occ event" with the snubber in the model, since it will restrict all movement of the node, not just that imposed by the "occ event".

The only proper way to handle snubbers in a static analysis is the multi-step (analysis) setup described by "Anonymous" above.

Zdzavko,

Are you sure the Vertical Load = 0 g, whereas HL = 0.3g at site?

Dear Richard

A couple of weeks ago I came to check "Piping Handbook" by Mohinder,if I'm right about the name, about using snubbers.I don't really remember the exact phrases but the concept prohibited the using of snubbers near equipment nozzles. As said the reason is the snubber's DEAD BAND that makes it not work in the proper time.
I thought if they cannot be used near nozzles because of their probable malfunction, they cannot be used anywhere else. This could result in disasters especially if snubbers are used near compressor suction or discharge nozzles to dampen seismic loads.

I would be thankful if you clarified that.

Thanks
Mehdi

In order to lock up, a snubber must first detect movement, usually due to seismic or similar forces. Generally, pipe local to equipment such as a compressor is far too stiff to allow this, and a snubber so located simply will not work, even though a stress analysis model may suggest different.

Thanks to MoverZ for his kind reply

I tried modeling snubber just giving the axial displacement to the cnode, since it's an axial snubber. No load changes I found. Is it ok to model like that?

No response...

If you obtain no load changes in W+P+T is normal, if you obtain in W+P+T+U than activate the snubber in load case editor.

Have you heard of using large gap guides or stops instead of snubbers? I thought it could be a good idea but one of my friends says if the gap is increased support will experience more load and software will not calculate the real loads in the field. Referring to the physics formula(X=V*t) he says if X as support gap is increased the velocity is proportionaly increased and thus using these supports is not applicable.

B31E (setting seismic design of B31 piping) has a short paragraph on gapped restraints during seismic events. In many cases (based on gap size), the restraint can be considered active for the seismic evaluation. BUT, the restraint must be designed for twice that load as this would end up as an impact.

Thank you Dave

You mentioned the restraint must be designed for twice that load. Is this mentioned in the code or it's based on some calculation or experience?
Could you please share a shot from that paragraph (gapped restraints during seismic events)or send the B31E code itself? Unfortunately I do not have access to that.

No response ...

Impact load is twice applied load. That's a basic and accepted approach.

You can purchase a copy of B31E from ASME.

W+D1+T1+P1

D1 with Cnode as described, this will add on the W+T1+P1 case
if -2 is the Disp in W+T1+P1
-4 becomes in W+D1+T1+P1

similar
W+D1+T1+P1+U1
restrains XSNB shows more loads
if Restraint X model it shows same load

how to read SNB results

Richard Ay Please Help me.