Settings should really come out of the basic design - co-ordination and discrimination with downstream (and potentially upstream) devices and meeting required disconnection times being the main drivers. Zs/earth-fault current levels are a part of that - e.g. to achieve R_A ≤ 50V/I_A etc.but measured values should really only be used to verify the design assumptions as it were - since (especially for TT systems) they can vary considerably over time - so calculations should really be based on design worst cases rather than what they happen to be today.

   - Andy.

Clearly if the supply impedance is so high that the EFR will not operate, then it is set too high ;-) Between that , and as alluded considerations of voltage rise during the largest non tripping fault, as an upper bound, and not wanting to trip on day to day operations (so slower and coarser than any protection down stream )lies a wide band of acceptable settings.  Between 0.1 and 1 percent of the full load current is a good place to start if you need a trial value.
Mike.

What reading did you obtain? There are a plethora of options when it comes to setting the differential relay. Which regulation are you aiming to adhere to?