Would modelling the boosters as if they were 'almost' tube fins give the correct CP? Three seperate fins, one at 0, and then -7deg and +7deg would emulate a tube fin. But the external tanks are not quite tube fins, as they are 'solid' and not hollow, therefore no internal plane. So, try modelling it with just two, one -5deg and one +5deg. My tiny likkle brains says that will estimate the external curvature of the tanks quite well.