I recently worked a project with a designer who was adamant that pipe should NEVER lift off supports. I could never explain the following to him:

1. CAESAR II is an approximation to the real world. Among the things that it does not account for UNLESS YOU TELL IT TO are:

a. The stiffness of the supporting elements in your pipe rack.

b. The ovalization of large-bore pipe at support locations.

c. Imperfections in fit-up and assembly of piping systems.

2. CAESAR II does a marvelous job of checking the state of each support and correcting your model "on the fly" for support nonlinearities such as lift-off.

3. When Spielvogel wrote the first real textbook on how to perform pipe stress analysis, nonlinear support effects were, in practice, impossible to analyze. But the Code committees of the era were well aware of the fact that pipe does, in fact, lift off of supports in many cases (and that guides with gaps often exert no forces on the pipe). So they built conservatism into the Code to account for things they couldn't analyze. That conservatism is, for the most part, still there.

So, when you have a 1800 mm pipe that lifts off a support at the top of a riser when hot, CAESAR II recognizes that fact and, in the lift-off cases, computes stresses and deflections without accounting for any supporting force at the lifted-off support locations. The stresses and deflections that it calculates for those load cases are consistent with a model where there are no support forces at the lift-off locations. So, when CAESAR II says that your model passes the Code checks, it means what it says. You may create instabilities due to wind or seismic loads at lifted-off supports, but you will not experience sustained or thermal fatigue failures.

Still, if you want to have an analysis where all the i's are dotted and all the t's crossed, you can calculate a stiffness for each structural member supporting your pipe near the lifting supports. And you can go into Roark and calculate a stiffness for the surface of the pipe resting on each support, accounting for ovalization due to the point load. If you include these things, you will have a much more accurate model, and it's very likely that your supports will no longer be lifting. But you will have blown your budget beyond repair.

My designer eventually gave up trying to arrange a support system with no lift-offs. He wasn't happy about it, but in addition to blowing the stress budget (me), he was starting to blow his own budget. He wasn't much worried about me blowing my budget, but when his started to go he became much less fussy. I'm not sure he is convinced, though, that lift-off isn't really an issue.

It's good to see young engineers worrying about everything. But when you see something that bothers you, the first thing you should ask yourself is, "Am I the first to observe this?" If you answer "yes" to this question, you are probably wrong.

Once you convince yourself that the problem that is bothering you has been seen many times before, the next step is to consider the consequences of your worst fears. As I noted above, CAESAR II calculates stresses and forces vary accurately in lifted-off support cases. But if you allow a support at the top of the riser to lift off, you now have a situation where lateral loads may cause unacceptable deflections and other nasty things. So you might want to analyze a wind or seismic case to see what happens.

As one of the earlier posters mentioned, it's usually a good idea to support long risers so that thermal growth occurs at the bottom rather than the top. This is easy to accomplish if you place the upper support much nearer (do some hand calcs to determine the value of "much!") to the riser than the lower support.

When you're designing a support system, your only true friends are gravity and leverage. Friction is a fickle ally, ready to run away at a moment's notice when dynamic loads appear. But gravity and leverage remain true friends. If you have a problem that is nagging at you, try to think how you can use gravity and leverage to help you. You'll find a way, most likely.