In other words, you should allow the pipe enough flexibility to grow away from the valve instead of into it.
If you want to estimate strain using CAESAR, you have to be able to estimate the valve's rigidity, which can only be done in CAESAR as a regular piping element.
However, the accuracy of that strain will only be as accurate as the assumption you make. Dave suggests lowering the allowable - either by manually putting your eyes on the stresses at that location, or by swapping it for a material with a lower allowable stress, or a user-defined material.
If you want actual strain, you'd need to perform an FEA on the valve, with forces as reported from CAESAR, but with the resistances as calculated from the FEA.