Thanks for the answers, im glad i wasnt just missing something, makes the design a bit easier knowing im starting with the problems i thought i had!

My mill is a large Hardinge fanuc vmc and I had considered facing all the box off, but the envelope is only 710x400x500, so would take a long long time for 3000mm, plus i have to take the side door off and due to the stupidity of the designer the door is higher than the table so i would have to use couple of large angle plates and make brackets, and id have to move the receiver for my probe, and remove my tool height setter, all of which is a pain as all my tool offsets would need to be updated and i'm bound to forget and bugger something up, plus if the beam is out by 3mm i could be machining a lot of metal off, thinning the wall and stressing it into being more twisted, then flipping it and doing it all again until ive got a whole beam of chips in my hopper and no beam!

My other plan was to build a jig in the mill and drill the box section ream some dowel pin holes to align it and locally face it, then weld up some corner brackets from plate or GFS and machine them to be square, bolt it all together and then I could use some sort of resin as suggested to make the surface for mounting the rails, i wondered if anyone has used self levelling machine grout for this? Ive used it to make foot plates for my mill, its 3 tons so needs good footings and the grout leaves a very good surface, its good for 80N/mm2 so should have no problem with a reasonable light gantry(180kg) is my guess.

Cheers for confirming my fears.