YNyn0 Transformer with Unearthed Primary Neutral

Situation
I have recently encountered a fairly unusual installation where a YNyn0 transformer has been used to supply a low voltage pump installation incorporating a large VFD.
The transformer was initially chosen to offset harmonic distortion with an adjacent Dyn11 supplying an identical VFD.
The transformer has been installed with the primary neutral connection floating in an effort to prevent zero sequence currents from causing nuisance tripping at the supply.
My concern is that, at least in theory, the zero sequence impedance on the secondary of the transformer will be high, and will not permit enough current to drive the protection in the event of an earth fault.
The add to the complexity of the situation, the upstream supply transformer is earth fault limited to 10 amps. There is no reticulated neutral on the 22kV.

The three potential solutions I am considering are:
1) Earth the primary star point of the pump transformer (however I'm concerned this will allow some fault current to bypass the upstream NER)
2) Carry out zero sequence impedance testing to accurately model any flux linkage through the tank, to accurately determine how much current would flow
3) Replace the transformer (obviously the costly option)

As I don't often encounter this vector group in my line of work, I'm interested in the thoughts of anyone with experience in this area.

Transformer details:
22kV/433V 50Hz
750kVA
5.7%
ONAN

Answer
You should give consideration to grounding the high-voltage neutral. If the YNyn0 transformer is supplying a single, 3-phase, VFD load and there are no other loads which are connected phase-to-neutral on the low voltage side of the transformer, then blocking zero-sequence harmonics will not be required because the VFD rectifier will not produce zero-sequence harmonic currents during balanced conditions.

The upstream supply transformer (supplying the 22 kV) uses a NER to limit the ground fault level to 10 A. When the YNyn0 transformer is grounded on both sides, the NER will also limit the ground fault level on the 433 V side to approximately 508 A at 433 V. You will need to review this to determine if your ground fault protection will have selectivity.

The answer to that question can be found in the zero sequence model of an YNyn0 transformer.

When both neutrals are connected to ground, the transformer will provide a path for zero-sequence current supplied from the HV side to the LV side. When the HV neutral is disconnected and isolated from ground, the zero-sequence circuit path will be controlled by the magnetic circuit formed by the reluctance of the iron tank of the transformer and free-space which will be very high, essentially an open circuit, and the zero-sequence current which flows to a ground fault on the LV side will be very small.

In your case, when the both neutrals of the YNyn0 transformer are solidly grounded, the ground fault level on the 433 V system will be controlled mostly by the NER on the 22 kV side which will be something smaller than the 508 A value.