For selecting the load case for NON LINEAR RESTRAINT status which will then be "linearized" by the program, it is , in my opinion,best to take the operating case.Design case may make it very conservative. This is valid if you have +Y ( and other unidirectional supports). However if your supports are bi-directional like Y , excluding friction, linear and non linear status will be same.Remember, CAESAR II dynamic module is based on linear analysis.

Buried pipe will not be subjected to seismic inertia forces as inertia forces are based on absolute accelearation.The effect of seismic activities on buried pipes is something you cannot simulate in CAESAR II.To properly understand what you need to consider , refer an American Lifeline alliance document on the subject.

I don't hany any problem with your load cases.

Regards