Electronic – use a TRIAC on a secondary isolated AC winding

I've never seen a seven-winding transformer but I have dealt with three-winding transformers.

The nameplate information for one particular three-winding transformer is as per below. (Graphic redrawn from photo to avoid showing the full extent of the very large nameplate.)

The impedances given are all on 60 MVA base. In this case tap no. 5 is the principal tapping, so consider the impedances as HV-LV 19.32%, HV-TV 33.08%, LV-TV 9.93%.

The impedance quoted for each pair of windings is with a voltage applied to the first winding, the second winding shorted, and all other windings open circuit. Refer IEC 60076.1:2005 Power Transformers - Part 1: General:

Therefore, assuming --

1. your transformer nameplate gives impedance for each individual winding
2. the testing was done as described above
3. your fault condition will be three phase fault on one winding, other windings effectively open circuit

-- your fault current depends only on the impedance from the HV winding to the particular winding under fault.

In this example, a three-phase fault on the 33 kV winding would produce a maximum three-phase fault current of 60 MVA ÷ 19.32% ÷ 33000 V ÷ √3 = 5.43 kA. A three phase fault on the 11kV winding would produce 60 MVA ÷ 33.08% ÷ 11000 V ÷ √3 = 9.51 kA.

For the AC component you can use a simple current transformer. You can directly wrap a few loops of wire around the current carrying wire (i.e. use air as the core material) to test the concept, or you can buy a pre-existing solution.

However, a buck converter will have some AC+DC component, and any transformer setup cannot measure the DC level.

There are only 2 solutions for measuring the DC level: hall effect sensors and current shunts. I don't understand why you think a hall effect sensor will necessarily be expensive (possibly a locality issue?): take for example the DRV5053RAQDBZR, which digikey sells for about a dollar in 1-off quantities.

As PlasmaHH meantioned, there are current shunts available which can measure these currents with ~100mV or lower burden voltage, however they are usually somewhat expensive. If you don't need to worry too much about tempco, you can just use a length of wire with a known resistance (measured before-hand using 4 terminal sensing).