I've been running a pipe stress analysis on a system connecting to an API 661 air cooler and have not been having any luck. I think what it comes down to is that I don't have the correct boundary conditions modeled for the nozzles. I attended a Webinar yesterday that talked a bit about modeling boundary conditions in Caesar and this topic was touched on, but I was hoping I could find someone that might be able to describe it to me in more detail. I've tried a number of things from calculating the stiffness of the nozzles as if they were cantilevered beams to various combinations of rigid elements made to simulate the cooler itself. Any thoughts would be appreciated. Thank you.

1. Model flange and nozzle neck as piping elements for each nozzle. Why: Nozzle flexibility.
2. Model a weightless rigid from the nozzle neck(s) to CL of the headerbox.
Why: To account for thermal growth of the header box.
3. Model rigid elements that span from edge of headerbox to edge of headerbox connecting the multiple nozzle necks, whose combined weight equals that of half the air cooler, and maintains a consistent mass per unit length.
Why: So that friction effects are accounted for accurately.
4. Note vendor drawings and determine how the header box is and is not permitted to slide and how far. In a floating box design, it is generally permitted to slide in the horizontal plane with a gap of some amount in a border surrounding the header.
Why: Not all air coolers are 100% the same. Note that the drawings may not have a smoking gun to tell you this information.
5. Apply supports onto the header box to match vendor drawings as indicated in step 4.
Why: It's not going to support itself.
6. Add a weightless rigid from the center of header box at CL to the center of the air cooler. This represents the tube bundle. If a floating box design, you may consider the center of the tube bundle to be an anchor point whence everything else slides.
Why: Applying weight on the tube bundle might cause the headerbox to rotate in a manner that is inconsistent with reality. You can alternatively apply weight to the tube bundle, but apply rotational stops on the headerbox.
7. If it still fails, model the headerbox nozzle connections in FE software to obtain nozzle flexibilities and apply these flexibilities at the nozzle neck base and rerun.
8. If it still fails, consider springs or additional flexibility of the piping.

Wow! That has got to be the most comprehensive answer I've seen from anybody. Thank you!

I will do my best to implement those into my model and see what happens. Thanks again!

And most important is about the temperatures you assign on the elements.