Hi,
Here is an observation we have found recently which we were unaware of earlier, but could have some significant negative consequences for those that use this load case.

The MAX combination in Restraint Summary will not always give the MAX, but instead give a sum of two forces as shown below.

Take the following load case setup as an example:


L25 which is supposed to give the maximum Fy (saving the sign) between L4 to L11 does not do so in this particular case as seen below. It instead takes the sum of Fy found in L10 and L11. This only happens in some cases, but not others.




We contacted Intergraph who sent us the document in the link below:

MAX load combination in Restraint Summary

This states that is is a misunderstanding of the way caesar II calculates and reports the MAX value. What I got from this document is that we should never use MAX combination case as it is not reliable.


The CAESAR Help file states clearly that

MAX - Indicates a combination that reports the maximum displacement, the maximum force, and the maximum stress value of the cases combined. This method retains the original sign. Displacements are the displacements having the maximum absolute values of all the load cases included in the combination. Forces are the forces having the maximum absolute values of all the load cases included in the combination. Stresses are the stresses having the maximum absolute values of all the load cases included in the combination. This method is typically used to report the greatest restraint loads from among a selected set of load cases.


But since the bolded part is not always correct, it should perhaps be more clearly explained in the guide. A few companies use the MAX combination in the restraint loads to design supports.


My question now is, can this MAX combination be used reliably for stresses?

For example, using the above load case set up, is it correct if we say:

• L29: L12, L13 (OCC) (MAX)
• L30: L29+L4 (OCC) (Scalar)

in order to get the max occasional/accidental stress in the WIN1 direction (North/South)? Will this be always reliable?

I believe your trouble is limited to the Restraint Summary for those nodes having more than one restraint (vector) and friction is included on these restraints. I believe the individual restraint reports are OK since they itemize each restraint at the node.
The Restraint Summary shows only one line per node rather than one line for each restraint at the node. These, normally orthogonal, restraints each provide their unique contribution to the total in the summary. But friction at each of these individual (usually orthogonal) restraints are all summed and reported together. I believe your report is "off" because of this summed friction. In this sense, then, I agree that where different restraints at a node can contribute loads in the same global direction (X,Y,Z), the Restraint Summary may not be a dependable source for the maximum restraint load. And this is unfortunate when you have so many restraints to review.
As you can see, it's not an easy point to explain.
Conceptually, one way to avoid this is to assign individual restraints to individual nodes. In this manner, there is no summation (of friction) across different restraints at a node. But that can be too tedious to be practical.
Perhaps our restraint summary should list each restraint rather than collapsing all restraints into a single listing for the node. I think that will avoid the issue here. I guess you could think of our restraint summary as focused on everything that's going on with the node rather than listing all the restraints at the node. There is also the issue of mixing load cases in this report - the reported maximum Y load could be causes by the hydrostatic leak test while the reports maximum Z load could be thermal strain. Reporting these two numbers on the same line may cause one to believe that the support should be designed for these loads together even if these loads are not coincident.
Seems to me, the best change to CAESAR II would be to show all restraints, individually, in the Restraint Summary.
One additional point - If I was to produce such a restraint report to hand over to the support engineer, I wonder if it wouldn't be more appropriate to provide a report where friction was NOT included. (You can turn friction off using the Load Case Editor.) Use friction in your CAESAR II evaluation of pipe but the structures guy probably will consider friction using your calculated normal load.

Hi Dave,

Thanks for your reply.

We have NOT considered friction in any of the above cases. As you rightly said, it is better to give loads to structural group without friction.

We managed to replicate the issue however. The incorrect "max" load appears when there is a restraint with a +Y support with gap, and a -Y support with another gap.
In this case, the restraint had at the same point +Y(150mm gap), -Y(50mm gap), and the X and Z restraints had 25mm gaps.
This is an unorthodox support, but is an unfortunate requirement to solve for the case where there is a flow line experiencing large wellhead movements, large explosion loads, and high fatigue lifetime requirements.

The problem seems to only appear when the Y restraint has gaps (different gap in negative and positive). When the X or Z restraints are given gaps, the loads are correct.

Our solution has been to completely ignore the MAX load combination from all load cases. We instead are now using SignMax and SignMin only.
The loads for the horizontal restraints will be the absolute max of the two. Rest and Hold Down loads will be individual. This gives correct results.



-Vivek

I have had a similar problem in the past. The max command appears to look for the maximum load for each of the restraints, i.e. the gapped -Y support has a maximum load of 77445N and the +Y support has a maximum load of -49885N. Using the maximum command then combines any loads from any of the restraints at that node acting through the cases being 'maxed', max +Y load and max -Y load hence the 27560N recorded (77445+-49885=27560).
Best way of getting around it for a results summary is as you say to use the signmin and signmax combinations.