For a dynamic analysis the way a rest support may be treated/modelled can be different from a static analysis. This is what I want:

If the weight of the (horizontal) pipe is significantly higher than the axial excitation force then it may be assumed that pipe movement at the rest support is restrained in all three (XYZ directions), just due to the friction force. Of course many other effects have to be considered, but I want to keep the example simple.

I noticed that CAESAR allows me to model a support with all kinds of restraint types, even allow me to use/specify a MU factor which is intended for making use of that friction factor. So I made a little test model and tried if I could set up such a situation.

As far as I understand CAESAR (remember I am a new user) I think that the X, Y and Z restraint behaviour are treated separately in the solver. So I don't manage to couple the X and Z restraint behaviour to the Y restraint status. In simple words: I want the X and Z restraints turned off as soon as the pipe is lifted off the support. In my test model I did not succeed in creating this situation.

Just to be sure: it is sufficient to model the support as a lineair element. The behaviour described above should be checked during the static analysis and then set 'fixed'. So if there was sufficient friction the restraint mode should become XYZ, if there was unsufficient friction the restraint mode should become '+Y' if not yet lifted or '' when lifted.

The current solution (workaround) I've chosen is to make 2 models, one for static analysis and one for dynamic analysis. Maybe stress engineers think this is wrong or strange, but there is a significant difference in how to model pipe supports dynamically or statically. Dynamically spoken many supports can be considered as XYZ (fixed) points, whilst they are just rest supports (+Y) in the static analysis.

Has anyone experience in this specific problem?

To make it even more clear I include this text part from the User Guide:



This is the kind of behaviour I am looking for, EXCEPT that I want the status of the +Y support control the behaviour of X and Z.

You need to search "friction" and "dynamic analysis" on this forum. You will get a lot of good answers. To list some things you have stated or implied that I disagree with.

1. Friction is never (I cannot write "never" here in big enough or bold enough type to express my true feelings) your friend. If you are relying on friction to keep your pipe in place during a seismic event, don't park your car (or walk) underneath that part of the piping system on the day the event occurs. In fact, if you ever file a report to qualify a system where your analysis "takes advantage of" friction in any way, shape, or form, you will eventually become an "ex-stress analyst." It may take months or years, but it will happen.

2. There is no such thing as a "seismic force" - we generally simulate seismic events with lateral g-loadings, so that the force produced in any part of the piping system is proportional to the mass. Big pipes generally require much more restraint than small ones in seismic events. You need to read ASCE-7 or one of ENV-1992 or ENV-1998 (I forget which, but one of those codes has seismic design rules that are very similar to ASCE-7) and absorb the analysis methodology required.

3. I doubt that your design code will permit you to qualify a piping system with a dynamic analysis that restrains more degrees of freedom than the operating case does. Do you have a genie to switch the added restraints "on" as soon as he feels the earth tremble? If you do, you are going to be a very wealthy man!

Note that one reason CAESAR II "clamps" resting supports is because typically design codes do not require vertical accelerations of more than a fraction of a g. I suspect they do this also to avoid having to deal with the mathematical issue of trying to extract eigenvalues from a nonlinear stiffness matrix. If a support "lifts off" in the operating case, they do not want to have to cope with the additional nonlinearities that would have to be added to the model to account for the lift-off case. I'm sure Richard or Dave will chime in shortly to discuss this.

I do agree with you that it is often a good idea to have separate models for static and dynamic analyses. But you have to be careful to keep the system geometries identical, it's a lot of work.

Hello Craig,

thanks for jumping into this matter! Unfortunately I have not been clear enough about my question. With 'dynamic analysis' I meant the studying of the (dynamic) behaviour of a pipe system due to excitation by compressor flow pulses (resulting in pressure pulsations).
I fully agree with you that, with respect to seismic events, friction force is not a good and save thought.

But for checking the response on pulsation induced vibration forces (due to for instance reciprocating compressors) this is a completely different matter. Forces are often relatively low and the weight of the pipe can be large enough to supply sufficient friction force (and dynamic movement for SMALL forces is thus restrained).

Would be nice if CAESAR II could handle such situations. I see no other solution than making two different models (one for static, one for dynamic).

Can anyone of you help me in explaining the basis of deciding the
"Friction Stiffness Factor", which is used to linearise the the effect of friction in case of the Non-Linear restraints.

Following is the part of the notes appeared in c2TR(training manual,
page no 5-50)in regard with this.

I have pasted it here, to just make you aware of What i am really seeking for.


From page No.5-50 of c2TR:
"Stiffness Factor for Friction (0.0 - Not Used)
(Active for: Harmonic, Spectrum, Modal, Range, and Time History)
As noted above, all of CAESAR II’s dynamic analyses are currently linear, so non-linear effects must be linearized.
Modeling of friction in dynamic models presents a special case, since friction actually impacts the dynamic response in
two ways—static friction (prior to breakaway) affects the stiffness of the system, by providing additional restraint, while
kinetic friction (subsequent to breakaway) actually affects the damping component of dynamic response; due to
mathematical constraints, damping is ignored for all analyses except time history and harmonics (for which it is only
considered on a system-wide basis).
CAESAR II allows friction to be taken into account through the use of this Friction Stiffness Factor. CAESAR II
approximates the restraining effect of friction on the pipe by including stiffnesses transverse to the direction of the
restraint at which friction was specified. The stiffness of these “frictional” restraints is computed as:
Kfriction = (F) (S) (Fact)
Where:
Kfriction = stiffness of frictional restraint inserted by CAESAR II
F = the force at the restraint taken from the static solution
S = mu, friction coefficient at restraint, as defined in the static model
Fact = Friction Factor from the control spreadsheet
This factor should be adjusted as necessary in order to make the dynamic model simulate the system’s actual dynamic
response (note that use of this factor does not correspond to any actual dynamic parameter, but is actually a “tweak”
factor to modify system stiffness). Entering a friction factor greater than zero causes these friction stiffnesses to be
inserted into the dynamics job. Increasing this factor correspondingly increases the effect of the friction. Entering a
friction factor equal to zero ignores any frictional effect in the dynamics job."

Jan
if you wish to do what you say using only one model, you would either use a friction stiffness factor under control parameters or input some virtual 'snubbers' on the snubbers tab.

Be carefull !
What do friction stiffness in dynamic ?
The eigenvalues will be higher calculated as calculated without friction stiffness !
1. This is not conservativ for all dynmacic impacts.
Example : Earthquake spectrum has the maximum acceleration about
3Hz until 8Hz. With raise at eigenvalue you go out from the max. acceleration.
So you can calculated instead of 50 m/s*s only with 30 m/s*s !!!!

2. A you sure that friction force be constant during the time earthquake work ? NO ! And what is with the + vertikal force from earthquake ?

Conclusion : Respected friction in dynamic only you are sure it is conservativ !

Dears,

Is there any reference in any recent edition of API-618 which allows reciprocating compressor pulsation analysts to consider friction at supports in mechanical study of reiprocating compressor pipings ?

Isn't it an ad-hoc high-risk assumption of pulsation analysts?

When someone says that shaking forces are not high, how will we get assured of sanity of this assumption ?

If frequently small bore pipings connected get broken frequently, can we still consider the shaking forces in main pipings small enough to be contained by friction at every support ?

If we consider this in pulsation study, why don't we do so for seismic events for simmilar forcing frequencies ?


'In fact, if you ever file a report to qualify a system where your analysis "takes advantage of" friction in any way, shape, or form, you will eventually become an "ex-stress analyst." It may take months or years, but it will happen.' Hats off to our mentor CraigB for 'jumping into this matter' to save us.

regards,
sam

"When someone says that shaking forces are not high, how will we get assured of sanity of this assumption ?"

I wouldnt call this assumption 'insane'.Infact, you could see the recommendation of using friction clamps to resist the low dynamic vibrations in many of the mechanical response study reports.

Many stress analysts, when dealing with situations where the mechanical study requirements conflict with the thermal analysis requirements, resort to 'friction as a friend' approach after proper evaluation.

Dears,

I am sorry, if 'sanity check' term hurts.

But, how reliable is the dependance on friction in high risk service like high pressure hydrogen reciprocating compressor service under API-618 ?

In one pulsation study-mechanical report of reciprocating compressor, analyst specified minimum recommended pipe support stiffness - say 0.49 kN/mm for 2"NB sch 80 piping - is it connected with some minimum piping span, say 7.7ft corresponding to 30 Hz for a 5Hz compressor piping? The analyst is silent in the report.

In API-618 4th Ed clause 3.9.2.2.1, ptp 179.2MPa(26000psi) endurance limit for CS pipe with opr temp below 371 C is specified. We, the stress analysts, do not know the number of cycles it corresponds to as per ASME Sec-VIII Div2 Fig 5-110.1 ? For CS with UTS<=80 ksi as applicable for the CS piping material used, 10000 cycles has allowable ptp38000 ksi & 100000 cycles has ptp20000psi allowable as per that ASME Sec-VIII Div2 Fig 5-110.1. Are we considering too high ptp 179.2MPa(26000psi) endurance limit for CS pipe of UTS<=80ksi ?

Can you help us in getting the answer of the above two problems ?

We, piping stress analysts, are as concerned & sincere as the pulsation study-mechanical analysts, may be less informed/trained & so we ask for help from mentors!

regards,
sam

I hope some of our excellent veterans here would give convincing answer to you (and me as well).I dont consider myself qualified to give an authoritative answer.
But I think, your question sounds like asking if we are correct in considering a safety factor of 1.5 or whatever number it is.Perhaps that depends on how much the most brilliant minds thought they have cracked the codes of science.