This is how the “max” option works for “restraints”.
The “Maximum combination” loadcase (say, L32, in your example) selects the maximum load on each of the modeled restraints from all the selected loadcases (L23, L24). If there are more than one restraint defined at a node (e.g., on the node 1360 there are 3 restraint types: rigid GUID w/gap, ridig +Z, and rigid -Z w/gap), then each restraint type is calculated and evaluated individually. CAESAR II considers each restraint component/type separately and picks the highest load among them as well as from all the specified component loadcases. Thus, the “Restraints” report for “max combination” case will have largest magnitude loads (saving the sign) for different restraint types from all specified load components at a node – I can't really quote the numbers here since I don't have the complete data set (loads on each restraint type that can be found in the "Restraints" report).
If you then use the “Restraint Summary” report for that same “max combination” loadcase L32, the loads from different restraint types for each degree of freedom simply summed together at the same node, whether they occur on the same restraint or not, and whether they belong to the same loadcase or not. The “Maximum combination” loadcase does not keep track of where its loads came from. My suspicion is that the loads from "+Z" type could be "canceling out" the loads from "-Z" type on the node 1360; so the FZ are "smaller than expected" because of this on the "Restraint Summary" report.
The SRSS loadcase L35 again acts on each restraint type separately (squared FZ from +Z restraint, summed with squared FZ from -Z restraint, summed with FZ from GUI restraint gives a different number).
Again, it's hard to speculate without working with real numbers for each restraint type on loadcases L32, L33, L34 (not the Summary report).
Therefore I would suggest that you use the “Restraints” report instead of the “Restraint Summary” report for combination methods such as max/min if you have more than one type of restraint at a single node.
If you want to make sure that loads of opposite sign don’t cancel each other on the “Restraint Summary” report, I suggest that you run the load case using the “SignMax” summation convention, and possibly add a second load case using the “SignMin” convention to cover the range.
_________________________
Nadia Strikovski
Intergraph CAS