Hi I am a long time user of Caesar pipe stress software and a new joiner in Coade discussion forum. Have a simple question as follows:-

Caesar, like other pipe stress analysis software, uses the centerline of the piping system to model the geometry and as such reports the stresses at nodes located on the imaginary line. However, tre stresses will be located on the pipe wall which may not be a constant across a given pipe section.

So how to know the stress pattern across a pipe section, apart from the stress at the node point? Is FEA the only option >>

All you know from a "3D Beam Element" analysis is the stress at the node point. If you need other information, you need other software, such (as you said) FEA.

I just have another doubt as follows :-

Because (Caesar II and) any pipe stress analysis software work on the basis of the pipe centerline concept, and the stresses reported are at node points that lie on the imaginary centreline, is it possible that as a result of stress variations (across a pipe cross section at any node point) the max. stress can be greater than the stress reported at a node point?

If so, then in case I have a stress model with 99.9 % of code allowable at a node point, it could so happen that the true stress on the pipe could actually be greater than the code allowable stress ??
Is this an acceptable situation ??

For me stress analysis program like CAESAR II is just one of the tools pipe designer is using. Definitely not the first tool to use. There are 2 issues. One is code compliance and second is good engineering design. Code compliance doesn't mean that the design is good. If you have stress levels of 99.9% then check if the design is good. If not then change even if it is within the code.

Airplanes fly with known cracks and with safety factor under 1. In piping small indication in NDT where design safety factors are 2 plus everybody gets exited. I have never figured this out.

In my humble opinion, without knowing the system it is indeed very difficult to know for sure. In general the factor of safety in-built in all the considerations while doing a theoretical analysis is enough to make the system work in actuality for long enough before it fails in real life, unless offcourse there are gross mistakes in design. If this kind of stress (99.9%) occurs in a sustained case we must attempt to reduce the stresses (play safe always). In an expansion case I might allow the results to pass depending upon the temperature and material. In occasional cases, I would hesitatingly pass. There also, it depends on the conditions of simulation and what is the type of occasional load.
Markl, I am sure must have encountered piping components which must have failed much earlier than governed by the rule he finally had deduced from their behaviour. Surprisingly, those results have been applied to a span much wider than his experiments.
So the best solution in a case of 99.9% may be to verify using FEA packages like FE-Pipe, etc.