Are you referring to circulating current in the delta or between one g-wye xfmr to another? I'm still having a tough time understanding what you're saying. You also lost me on why a low source impedance causes greater xfmr currents. Are you referring to a fault?
Look at it this way:
The three windings in the delta secondary (call them a, b and c) form a triangle. The vector equation
a =
b +
c has to be satisfied to keep current from circulating. The bold face letters indicate vector (or phasors if you prefer). So far so good?
Now a, b and c are each directly related by the turns ratio to the primary winding currents A, B and C. In the wye the lettered currents are in wye windings and each corresponds to a service conductor, in the secondary the lettered currents are in the delta windings and the currents out to the load will each come from both windings at that (un lettered) terminal.
Still with me?
Since the wye service conductors may be derived from a wye bank and may have unbalanced loads on them, we cannot assume that the magnitude of the line to neutral voltages are all equal.
In fact we cannot be sure that
A =
B +
C.
Since
a =
b +
c and
A is not equal to
B +
C, something has to give. (Given that the two sets of voltages are each related by the same turns ratio.)
The slack will be taken up in two ways by the same current flows:
Currents will run in the delta windings until the IR losses have made the resulting voltages satisfy the equation in conjunction with the corresponding currents in the primary causing IR drops to help make the second equation be satisfied. Neither mechanism will be responsible by itself for the balance.
On the primary side one contribution toward balance will be an IR based shift in the neutral point, the result of really high neutral currents, since the currents in the three windings will be far from balanced.
If you instead just open the neutral of the wye primary and let the center point float, it does not take any significant current flow to make the primary voltages obey the vector equation. In fact, at idle there will be no circulating currents and no primary currents except for the magnetizing current of the transformer.
Without drawing a picture and waving my hands, that is the best I can do.
In answer to your second question, to get the necessary IR drops on the primary side will take more current if the service contribution to the IR drop on each phase line and the neutral is is reduced.
The IR contributions on the secondary side remain unaffected by source impedance, but will require more current the lower the transformer winding resistances are.