I think I could compute Harm with DOW an DOI
Of course you can -
IU = W(H-H25)/C +K
can be rearranged to give
H = C(IU-K)/W + H25
I noticed, for the same AirCraft A320 I can have DOI totaly different
Not at all unusual. This just reflects a range of empty weights and CGs across a similar fleet (presuming similar configuration and the same IU formula, of course). For the example you cite, CGs of 18.33m and 18.74m possibly are a little too far apart for comfort and need investigation.
It is important to realise that each aircraft is an individual machine and will have small differences from its apparently identical fleet colleagues.
Some operators use fleet average data for empty weight and IU. If this is your case, then part of the confusion may relate to having apparently identical data in the fleet average declared data .. it is important to keep in mind that the average comes from a small range of data which is used to calculate the average..
However, getting back to your original concerns, can you indicate the specific trimsheets from which you obtained your MAC values ? If the trimsheet is consistent with the Airbus calculation models then the reason for the apparent MAC difference should stand out.