...Good day all, first I'd like to acknowledge the fact that a lot has been said about Support friction/Sliding and Non-convergence of Ceasar II analysis. But I still need some clarification about one common & "seemingly" simple solution to Non-convergent analysis; THE USE OF A HOLD-DOWN Restraint (+Y) at the "bouncing support" node, without changing the Friction coefficient or stiffness of support.
I'd like to know if there are any significant TECHNICAL implications of this "simple" solution...
Thanx, look forward to your response.

The technical implication is simple.

If you model it as a hold-down, you must BUILD it as a hold-down. Look at your pipe support catalog for rigid struts, or use some loose U-bolts. Rod hangers are much better at this - there is always enough of a manufacturing tolerance in the bolt hole clearances, etc., to cover for a couple of mm of lift-off.

I continue to be amazed at the contortions we now go through to keep pipe from lifting off of supports in analysis. The software is aware that the support is inactive, and therefore the stress state it calculates for load cases with lift-off are correct and can be used in the Code equations.

Most small lift-offs (<1mm) are amply covered by the stiffness of the steel that you attach the support to - in the loaded state, our assumption that the support steel is "rigid" is not supported by real-world structural design practices. The best solution to most lift-off problems is to calculate a load-deflection stiffness for the steel you are attaching to and enter this as a "spring" for your "rigid" support. Obviously, this is not going to happen for small-bore piping, but on the other hand small-bore piping is usually not rigid enough to produce significant lift-off.

But anyway, even I go through this lunacy from time to time. Welcome to the madhouse!

By the way, a +Y restraint in CAESAR II is NOT a hold-down, that would be a Y restraint. Check your software documentation.

Hi CraigB

As usual practice we slightly differ than your opinion.

1) If model as Y support in Caesar still we use normal resting support (not hold down) if restraint summery loads are all -VE

2) Remove friction to avoid convergence error if support is lifting up at that point

3) Hold down is used if specifically required to pass stress, limit deflection or required for pulsating line

4) Do not recommend to use support stiffness for only one or where ever need Caesar benefit rather than to for composite analysis ( pipe + structural model).

Please pass your comments on that.

Regards


Habib

..CraigB, a zillion thanx for the quick response.
yeah sure, Y-restraint is HOLD DOWN...typographical err

Habib

3) Hold down is used if specifically required to pass stress, limit deflection or required for pulsating line

Yes that is my point. CAESAR II knows that there is no supporting force there, and calculates stresses accordingly. If these stresses, used correctly, do not violate your Code you have no need of a hold-down unless:

a) you wish to limit deflection, or

b) you have a pulsating line.

Pulsating lines are different animals, and often require extremely carefully engineered support systems in order to avoid damage. I don't think the original question concerned this issue, and my answer of course would have been different if transients were an issue.

So, we are left with limiting deflection. Yes, there are cases where excessive vertical deflection of the piping in an operating or excursion load case could cause an operating problem. But if you have a vertical deflection of a couple of mm and are concerned that this may cause you an operating problem, I submit that a much better solution to your problem would be to slope the line in that location.

There are many ways to deal with the uplift that MUST occur whenever you have an elevation change on a rack in a line with a significant thermal range.

1. You can use a spring. Most refinery or oil & gas operations prefer not to do this.

2. You can tinker with the distance between the upper support and the riser. Often, gravity will solve your problem for only a few minutes of your time.

3. You can hold the piping down and increase the stress range. As I have noted in many posts, I prefer not to do this unless I see no viable alternative. Hold-downs, after time, often corrode to the point where they become de facto anchors, which may not suit the piping layout you have. When this is necessary, I strongly prefer to use loose U-bolts, with a small vertical gap.

4. You can account for the stiffness of the support steel and/or pipe supoprt components. I am not sure why you regard this as a distasteful alternative, Habib, since the calculation usually takes only a few minutes, and no Code that I have ever used would prohibit this. After all, we use friction routinely, and the calculation of a friction load is far less precise than calculation of a force-deflection constant for a beam in bending.

5. You can allow it to happen. I have heard people I otherwise respect greatly say that they do not want any support lift-off EVER because it "doesn't look good." I respect their opinions, but I disagree. If we wanted piping systems to look good, we could cover them in lace or hire artists to paint pictures on the insulation jackets.

Finally, you mention that you can remove support friction to assure that the friction convergence error is not causing the lift-off. I have no experience with doing this, and you may be right, but I submit that if friction at the a support can cause an adjacent support to lift off, there is a major problem with the layout of the piping system.

There are a few different approaches to settle down a nonconverging model. If a +Y support is "bouncing" between on and off, I would replace that one (or just one if several are present) with a Y support. If that converges, just check to see if all loads here are -ve. If so, OK. If not, see how big the +ve loads here are. If small, I would ignore. I figure they should be small as that's why they can flip from +ve to -ve from one iteration to the next.

Hi CraigB

Thanks for your detail explanation.
I agree with Mr Dave's approach to deal with non convergence .

I like to clarify following points.

""4. You can account for the stiffness of the support steel and/or pipe supoprt components. I am not sure why you regard this as a distasteful alternative,""

My point is we should not used stiffness for only one support rather than should used stiffness for all support if we wish. If system is not highly critical stiffness for only one support may be acceptable.

""5. You can allow it to happen. I have heard people I otherwise respect greatly say that they do not want any support lift-off EVER because it "doesn't look good." ""

CraigB I fully agree with you. Once I had to report one supervisor with whom I used to have regular fight because he use to reject my support selection even if that lift by .001 mm without giving me any technical justification.

""Finally, you mention that you can remove support friction to assure that the friction convergence error is not causing the lift-off.""

My point is slightly different. Non convergence problem may happen where support have tendency to lift up. If I can allow that support lift off at that point, I remove friction to solve non convergence problem rather than putting Y support. I Agree with Mr. Dave also if +y load is not much there is no much harm to use Y support to solve non convergence.


Anyway thanks for your time to clarify issues.

Regards

Habib