Hello stuuk78
Is this a new installation, or is this a problem that has developed after a period of time?
I assume that this is new installation.
1. Transformers draw a very high inrush current at switch on. This inrush current is independent of the secondary circuit. The magnitude of the inrush current is dependent on a number of factors including the amount of flux in the iron. If the transformer is over fluxed, i.e. is wound for 460 volt and used on 490 volt, there will be an appreciable increase in the inrush current. The inrush current of transformers only lasts cycles, but can be in the order of 10 - 20 times the full load primary current of the transformer. If your transformer has been very conservatively rated, and is actually say a 2.5KVA capable transformer, then the inrush will be higher again.
2. Fuses fail when the energy through them is too high. You cna over load a fuse and it fails earlier the failure time tends to reduce with the square of the current, hence we talk about the i squared t of the fuse. (more common terminology with semiconductor fuses) A slow fuse will last longer with the overload, but will still fail. Sometimes, we just have to use a fuse that has an i2t greater than the switch on energy of the transformer in order to prevent failure.
3. If the transformer has a shorted turn or similar insulation fault, it will have a high primary current, will tend to be noisy and the open circuit secondary voltage will be low. It would also heat up reasonably quickly. Temporarily fit oversized fuses and run the transformer unloaded for a period of time. If the transformer primary current is reasonable, the transfomer does not make excessive noise or smells, and does not heat up excessively, then I would suggest that you just have a healthy inrush current.
4. In my part of the world, we do not operate under NEC, and I would fuse for the cable rather than the transformer. I would protect the output of the transformer, and use thermal protection in the transformer. (Usually a thermal fuse built into the primary circuit). I agree with Sanditech, about four times the rated primary current is about right in most cases.
Good luck,
Best regards,
Mark Empson