Neutral Earthing in Solar Transformers

Question:
Three winding transformers are usually used for collecting the power from solar power inverters, popular connection being Delta/Star-star. The neutral of secondary stars is never earthed, many times it is not even brought out. In one case, it was inadvertently kept earthed and inverters started failing frequently. Why it is not earthed in service? How earthing of neutral is affecting inverters,leading them to fail?

It is a bit strange connection for a step up transformer. But this is the one used for solar projects. LV star is fed from inverter output (1-2 MVA <1kV) and delta HV is connected to 11 or 33kV collector grid. Recently an IEEE standard came out for solar transformers C57.159-2016 where under clause 5.1.4 it is said:

“The inverter operation is not affected by the inverter transformer vector group (Dy1, Dy5, or Dy11 will make no difference). No neutral connection is required on the primary (LV) side of the transformer. If a neutral point of the primary (LV) winding is available, it is recommended that this neutral point is neither grounded nor connected to other ground points. On the secondary (HV) side, the inverter transformer can have an isolated neutral point or resonant grounding or low resistance grounding.”

Hope power electronics experts can explain why this connection is used (usually converter/inverter transformers are with two LVs one delta and other star) and why star neutral should not be grounded.

Replies: @WaRoss
Assume three identical transformers connected in wye:delta.
The issue is with the delta secondary.
The delta locks in the voltage and phase relationships of the three phases.
My favorite explanation is to consider the delta as an open delta plus a single phase transformer.
The open delta forms a virtual transformer across the open side. This virtual transformer has similar characteristics to the two real transformers.
Now when we complete the delta, we may consider a single phase transformer in parallel with a single phase virtual transformer.
What happens when we parallel two transformers with unequal voltages? Circulating currents, limited by the transformer impedances.
Now take a wye:delta bank with the neutral connected on the wye side.
Assume that the impedance of the transformer bank is 3%
Consider the voltage on one phase to be 9% low.
Now we have a virtual transformer with 100% voltage in parallel with a real transformer with only 91% voltage.
This 9% voltage will cause a circulating current limited by three times the transformer impedance or 100% of rated current.
This will cause rated KVA to be drawn from the source, but supplied by only two transformers and by two phases.
If we have three 100 KVA transformer the almost 300 KVA load will now be fed by two 100 KVA transformers or a 50% overload.
In this example a 9% voltage unbalance has caused an almost 50% overload with no power drawn on the output.

Mitigation.
The delta does not have to be an equilateral triangle. The voltages may be unequal if the phase angles are allowed to shift to compensate for the uneven voltages.
If the neutral is connected, the voltages and phase angles are locked in and the trouble starts.
Now consider supply voltages of 100%, 100% and 90%.
Draw a triangle of sides 10, 10, and 9. The angles will not be 60 degrees. Not a problem. These voltages and phase angles will be proportional on the secondary of the transformer bank.

The important thing is the delta vector diagram will be closed. It is when the delta does not close that issues with circulating currents arise.
Leaving the neutral floating allows the delta triangle to shift to accommodate the error in voltages and phase angles.

Note:
We should be referring to a neutral connection not a ground connection. Grounding the wye point makes no difference UNLESS the source wye point is also grounded. Then the connection becomes a neutral connection that is incidentally grounded.
If the transformer wye point is solidly connected to the source neutral or wye point but neither point is grounded the same problems will arise.
Distribution engineers and linemen tend to refer to the neutral as the ground. Well, the neutral is always grounded so it makes little difference most of the time.

Utilization engineers and electricians make a distinction between the ground and the neutral They must as North American codes make a difference and different rules apply to utilization grounds and utilization neutrals.

In this case, the accepted distribution language may be somewhat misleading as the issue here is not grounding, the issue is connecting the neutral from the source to the transformer wye point.
Ground the neutral of the source.
Float the neutral of the wye:delta bank.