that HAVE to be separate subnets.
typically you would use a special Class C address set aside for WAN addresses, like
10.254.254.0/24
and then each interace on a WAN would be 10.254.254.1/30, 10.254.254.5/30, etc
If you don't want them in the same subnet (this is really crazy IMHO) then you would bridge all of your WAN connections. At that point all sorts of trouble and broadcast storms with only one T1 being actually used would be what happens.