Per the various replies you have received. Protecting the conductors between the transformer's secondary terminals and the first OCPD is difficult. Typically primary side fuses simple do not operate fast enough at the primary current a secondary arcing fault would draw.
Alternaives: 1) protect from the primary device with a contactor or circuit breaker. A contactor would need fuses to protect it from too high a fault above its own inherent withstand and interrupting rating. Try to get a 3 cycle device versus a 5 cycle device... the 2 cycle improvement may save you a couple of calories or more!
1a) Assuming you can afford and fit that device you then need the sensing to go with it. Transfomer differential with the LV CTs on the OCPD secondary side would be good if you can fit it in and have reasonable wiring requirements. This will not affect system selectivity and should give you relatively fast protection.
2) Trips that can provide Zone selective Interlocking across the transformer are available from at least two North American Manufacturers. One provides relatively fast ST capability and the other can ZSI the instantaneous functions as well as provide significant selective capability between the trips and downstream loads such as MCCs. In some cases 100% selectivity and 100% instantaneous protection may be possible from the MCC branch to the MV device ahead of the transformer.
3) Maintenance switch, RELT or alternate setting group settings in a MV 50/51 type relay on the MV device can help if using CTs on the transformer secondary is not practical. However this is not a 7x24 solution and requires rigorous implementation within maintenance procedures.
4) Shunt Energy devices are available that can reduce Incident energy below a calory on a 480V bus. However, to cover the "blind" spot, you still need the primary switchable OCPD.
5) For new installations you can consider putting the secondary main in its own dedicated equipment. A section appart. Ensure that the construction is such that the "probability" of an arcing fault is minimal and that the device can be operated remotely, including rack-out. Distance is a good thing when it comes to AF.

Some suggestions above.

Search IEEE Explore for articles on zone selectivity, maintenance settings and other related terms. There are various articles that deal with this.

6) And, if you can do none of the above do look at your fuses and see if you can optimize their sizing better. You "may" be able to get the energy close to 40 calories if the fuses are smaller or have lower current limiting thresholds. Not easy to do but if you do not need as much extended rating it may be possible.

Paul, I think you are right at the heart of the matter. If you use CL fuses, they tend to have a more vertical time current characteristic. Once a fuse is selected with its TCC falling to the right of the inrush point (8-12 times FLA), it can be pretty slow for secondary faults.

NO matter what you add at the transformer, you are left with the problem of working on the transformer. You need to test for voltage, before the terminsls are condsidered dead. Onesolution is current limiting fast acting fuses on primary if this will limit the arc-Fault on the secondary. A second solution is test for lack of voltage on primary if this is a lower category. A third solution is replace the transformer with a special higher impedance transformer. Then there is the old solution of installing a current limiting reactor in the primary leads to the transformer.

A current limiting fuse in the primary generally will not act in current limiting range for a secondary fault. Higher impedance or current limiting reactors may increase arc energy by slowing fault clearing.

You all focused on all the equipments but i would like to make in notice about the flame resistant coveralls and clothing for the body protection which is really a very integral part to be kept in mind and should be taken care of.