In this case, the "virtual" anchor being described is just a point of zero movement along the axis of interest (in this case, along the axis of the header). Caesar will gladly give you a "perfect" calculation that says that point won't move.

In the real world, you need a directional anchor/line stop/bumper in order to insure that the real piping system behaves as the model predicts. Friction (which will also read zero at that point in Caesar) and uneven expansion from a progressive heating of the line means that your "virtual anchor" won't stay put. You can end up with one loop taking more expansion than it was designed for and the adjacent loop not taking enough.

So, definitely call for a stop in between the loops and give your structural engineer enough load to account for it having to resist a percentage of the expansion load when one side heats up before the other.

I have literally seen beams twisted over in the field at such "virtual anchor" points because they were put in without any worthwhile load being designed for.