When you say "I want to apply realistic analysis between CAESAR II and field" and you are still using such support, you must keep in mind… it's not possible.
You try to guess something based on possible scenario by ignoring many others. The above posed answers are very clear. And you are going to take the responsibility for something you are not sure.
The mechanical engineering is mainly focused on "how to avoid something" (there are exceptions, of course). It's also true Caesar II and FEA are exceeding this target, but basically "allowable loads on nozzles" method remains dedicated to "how to avoid". And the answer is … by rethinking the configuration or by using spring supports... for sure not by mathematics.