Not if you're talking about the transformer's primary neutral - it should be left unterminated.
This is a _grounding transformer_, not a wye:delta transformer being used to change voltage or create a separate system.
The specific purpose of a grounding transformer is to create a neutral point that can be grounded in an otherwise ungrounded 'delta' system.
I believe (though this really should be confirmed by someone who regularly engineers such systems) that the derived neutral should be treated (and grounded) as if it were the _supply_ neutral of a 'wye' system.
In other words the OP is trying to convert a delta system into a wye system without actually changing the supply transformer (2000kVA) but instead by adding a grounding transformer (significantly smaller).
(Side note: if you have a normal grounded supply and are using a transformer with a wye primary, the reason that you don't connect the neutral on that wye primary is related to the reason that you _do_ ground the wye in a grounding transformer. In both cases the transformer is creating a low impedance derived neutral. If you have a grounded system, and you connect a wye primary and ground it, then your system will have two 'neutrals' fighting each other and you will see large circulating currents if there is any phase imbalance. If you start with an ungrounded system then that wye primary becomes the system neutral that you do ground.)
Back to the OP: I know just enough about grounding transformers to get into trouble. I know the basic theory. In my _lab_ I've wired and used a _small_ zig-zag transformer setup (to make a 30 kVA transformer work). I have not been involved professionally in systems of the size you are talking about. I am happy to help with your initial survey of what you want to do, but I believe that you need significant engineering of the change you are proposing, specifically to determine what other changes to the over-all system would be required when you add grounding.
Additionally, I believe you have a misunderstanding. If you have a ground fault in an ungrounded system, the second ground fault is not what causes transient over-voltage. The transient overvoltage can be caused by a _single_ ground fault of a specific type. A so called 'restriking' ground fault is one that is vibrating open and closed. For example a ground fault in a motor that is synchronized with the motor rotation. This sort of ground fault will interact with the capacitance to ground of the system and 'pump' the system-ground voltage up to above the normal line-line voltage.
As Larry identified, you have other options to solve the problem of an ungrounded system. Rather than using a grounding transformer, you could 'corner ground' the system. This makes one of the phase conductors a grounded conductor, and makes the voltage of the other conductors 480V to ground. To do this you would need to evaluate that all the insulation and breakers are suitable for operation at 480V to ground. A corner grounded system prevents the transient overvoltage situation, and is perfectly fine if your loads are all transformers and directly connected motors. It is less desirable if you have any connected VFDs.
-Jon