I have a 36 inch diameter DR 17 HDPE pipe that is 15,300 meters long (yes 15.3 km) and it is approximately 80% buried. Line is treated effluent at 98 degrees F. The buried pipe modeler puts in "soil restraints" as required. The model ends up with 5000 plus non-linear restraints and the solver just can't seem to converge. Should I chop my model up into smaller segments. Chop locations at virtual anchors.

Regards,

Pat LaPointe P. Eng.

Pat,
You are probably correct - something to do with those nonlinear conditions but it may be that you are reaching an internal limit on certain types data (e.g., total number of restraints). I would think the program would provide an appropriate message but you may want to check this. There is a statistics table in the Auxiliary Data window that will show how much "room" is allowed for each data group. If you are maxed out, increase your allocated memory.

15 km of HDPE? Do you actual tees & elbows in this system? I thought most HDPE systems are long runs of straight pipe that may "rope" through a trench. No much of a "piping system" that would require a flexibility analysis. You may want to see what NPPS NM.1 says about this. NPPS (Nonmetallic Pressure Piping Systems) is the new ASME code for thermoplastics and thermoset resins.

Yes, one option would be to break the system into subsets. But don't just add anchors to represent those virtual anchors. Instead, run those long lines on either end of the run of interest and let the program develop the friction restraint (and those virtual anchors) along those long runs.

just a quick question on this post: is any pipeline with the above attributes considered "buried" at all?
my point is how calculations of upward/lateral soil resistance would add up here considering that the pipe is not entirely buried?

I think OP meant that 80% of the length is buried - not 80% of the height.

In the case of 80% of the height, your pipe will likely buckle out of the soil due to sunlight if the pipe is long enough. As such, this is not really a CAESAR analysis, anymore.