For lowside windings of ≤25kV, phasing voltmeter sets are routinely available for tests at the open lowside tie.  Plan on making, recording and evaluating 15 voltage readings before closing the tie.  PT secondary readings may also be used provided that PT connections have been shown to accurately reflect the two sets of ø-ø and ø-grd bus voltages.

Unmatched neutral-grounding resistors or dissimilar configurations need investigation.  Be aware of incraaesed fault duties in the paralleled configuration.

Also, you might anticipate significant var circulation between the two units—potentially on the order of 1MVAR.  

Do you think that this could be a problem. Will the transformers mechanically withstand this hammering ?  How is the 1MVar circulating reactive power going to reflect on the customer's metering? What is the fault that could occur between the substation's metering and the customer's metering because of that circulating VAR's ?

France Transfo states,that connection in parallel is permissible if :
1. The power output of the largest Tfo should not be   more than twice that of the smallest unit.
2. Transformers must be supplied by the same network.
3. Connections must be the same length and have similar characteristics between the LV terminals (if you have MV/LV Tfo)or MV terminals (if you have HV/MV Tfos)and the coupling circuit breaker.
4.Couplings ( vector group) must be identical.
5.Short-circuit voltages must be identical (+-10%         
6.Differences in secondary voltages between the different
Tfos must not exceed 0.4%.
7. The switch(tap-off) shall be positioned identically on each Tfo.
8.If the power outputs of the Tfos are different,the total available combined power output will be approx. 10% less
that the sum of power outputs for each Tfo.
    

If the transformers are identical in all aspects other than the 1% impedance difference there will be no open cct voltage difference on the low sides before parallelling.  

The impedance difference will mean that one transformer will carry more load than it's higher impedance twin, so use CTs in each xfmr to provide overload protection.

On the question of the impact on customer metering - as long as the metering downstream of the transformer secondary interconnection, there should not be any impact.  

Also, I just want to emphasize busbar's point on the increased fault current.  If your customers purchased equipment based on a 40 MVA transformer upstream, then parallelling the transformers will nearly double their available fault current.  This could be a serious issue.

dpc

Derek—I goofed in suggesting caution with 1MVAR circulating current.    A VAR/VA ratio of one in 40 corresponds to 0.9996 PF.  My [unfounded] concern was that in nonidentical transformers, circulating currents can often be >10x that of nonparalleled excitation current.  For OLTC-less banks, that’s just something you may have to live with.