If you can safely isolate the HV side and LV side, and have a controllable voltage source like a 110 VAC adjustable test set used for protection testing, you may be able to inject low voltage across one phase at a time.
Connect the source to r & s phases and have meters monitoring test set output current and voltage. Slowly raise the voltage and monitor current. If any cables are cross connected, there will be current flow but no voltage. If everything is clear, the voltage will increase as the test set supplies the transformer excitation current. Lower the test output to zero and try on other phases.
No need to go to a very high voltage, 5-10 Volts might be enough. (There is a possibility of false results if the cables are long enough that the V drop through the cables is 5 volts).
This test requires safety awareness by the testing personnel. High voltages could be present on the other terminals.
We have found phase to phase bus duct faults on the secondary of transformers this way. A straight ohmmeter or other insulation resistance test can't identify the short because the transformer winding is essentially shorting the phases together and you read zero ohms on a good system.
We did this after a bus to transformer connection blew up when energized. Transformer and bus duct test results were good prior to energization, but the tests were done prior to connecting the flexible braids between the bus and the transformer bushings. Improper installation of copper shielding tape on the bolted connections shorted the phases. A good NETA technician showed us how to check for phase-phase faults using this voltage injection method.
