When the manufactuer determines the transformers impedence, the transformer KVA rating is used in the calculation.
So if you are only given one transformer impedence, you should use the transformer KVA rating that was used to determine that impedence. I don't know what the standard is, but if I had to guess, I probably guess that the standard OA KVA rating is used to determine the %Z.
If that's true, there wouldn't be any need to used the forced air transformer KVA rating.
But are the AIC ratings of your OCP devices really so close to the available fault current that it would make a difference which KVA rating you used??
I'm with Jim on this, with a small caveat: you have to match the KVA rating used for calculating fault current with the KVA rating used as the basis for the impedance rating.
Fan cooling does not change the construction of the transformer, and thus will not change its short circuit current. (This is ignoring second order effects; for example temperature changes will result in changes to short circuit current, and fans will certainly change machine temperature.)
However the impedance rating of a transformer is made on the basis of a specific KVA rating. If you take the _same_ transformer and give it two different KVA ratings (say, because of different operating conditions), then you will also have two different impedance ratings. I do not know if manufacturers regularly report the 'fan cooled' impedance; make sure you know _which_ KVA rating was used to provide the impedance value.
I think I agree with Charlie. The per unit impedance of the transformer is based on it physical construction. Overloading a transformer while fan cooling it to have it operate within the temperature rise threshold does not impact the physical construction and therefore the per unit impedance is not effected. While I admittedly need to and will research the subject further I think it is not incorrect to use the per unit impedance at a value higher than rated kVA to determine the available fault level.
I think I side with Jon/Jim on this. I think but am not certain that transformer impedance is usually expressed in terms of percent impedance referred to its ONAN rating. Some transformers have listed on the nameplate another %Z that applies to another cooling method. You obviously need to use the correct pair but I think if you pair them up, you arrive at the same conclusion, the fault current (very short term) is not really impacted by the cooling method of hundreds of pounds of steel.
Here is another forum with similar discussion and a nameplate to look at...
By luck, just this morning, I have had reason to peruse several resources (IEEE, Westinghouse, and GE) on short circuit sources.
Never once was transformer 'fan cooling' addressed, actually, the only adjustment I saw was for taps. While it is possible that a transformer might have several different %Z values, the final rule is: use the kVA at which the %Z was reported. %Z is actually a per unit value and so it must be used with its base kVA.