Ok thanks so a number 2# alu would be the grounding conductor run with the secondary conductors to the load . Just interested in why they used that table and not the branch table 250.122 because table 250 .102(c) (1) it clearly states grounded conductor and not grounding conductor in addition iam also using multi paralleled on some transformers so I see taking the total conductors adding them up and using 12.5 % to calculate the grounding conductor in that raceway or is that the bonding jumper only . This is one code rule that to me is confusing to calculate by jumping around with many notes or articles .
Disregard the reference to 450.6 as it was the wrong starting point
I was using the 2008 but the only thing that changed in the 2011 is the names they are calling the bonding jumpers, it seems they are now calling the bonding jumper that if you make the system bonding at the first disconnect, the the conductor that runs back to the transformer to bond the transformer case to the neutral or ground bar in the first disconnect is now called a supplied side bonding jumper, 250.30(A)(2) still points you to 250.102(C)(2) which still points you to table 250.66, but just the names have changed, you will be running your ungrounded phase conductors, and neutral, to the first disconnect in this you install your system bonding jumper between the neutral bar and the grounding bar which could be a screw and land your GEC's in this panel and you size the supply side bonding jumper that runs back to the transformer to bond its case, but the neutral is kept isolated from the case because the bond is made at the first disconnect, just like the main bonding jumper in a service, or you make the system bond in the transformer and run a supply side bonding jumper to the first disconnect and then you have to keep the neutrals and grounding separate , which if the first disconnect is a panel I would do the system bonding at it so the grounds and neutral don't have to be kept separate, and the GEC can land there, if you do it at the transformer then the GEC's have to land in the transformer.
But the sizing of both the system bonding jumper and the supplied side bonding jumper both have to be sized off table 250.66, so yes you need a #4 copper or a #2 aluminum run with your run to the first disconnect, some determine where to do the bonding from which is closest to the point of the grounding electrode, but I try to avoid doing it in the transformer if I have a multi-circuit panel for my first disconnect, the other reason is most panels come with a bonding screw that makes it easier to make the bond, but if you use this then you are limited to only landing the GEC on the neutral bar only because of 250.24.
Also the supply side bonding jumper can be a non-flexible metal raceway as stated in 250.30(A)(2) so if you are running a non-flexible metal raceway you don't need to run a supply side bonding jumper between the transformer and disconnect but you will be required to use bonding bushings or locknut's on each end of the raceway, could save some money.
Also the system bonding jumper requirements is in 250.30(A)(1) but also refers to 250.28 for sizing it but basically its the same as the other one.
Also 250.30(A)(1) exception 2 allows you to bond at both the transformer and the disconnect/panel if you use a non-conductive raceway such as PVC, this is just like a service.
Hope this clears it up a little, I don't know why I use the 450.6 reference as that it for a whole different install?