In simulating flow over bluff bodies, is there an empirical relation between Mach number and distance between the BCs (inlet, outlet, symmetry & wall) and bluff body to get better convergence.

In the supersonic case it's quite easy because the regions of influence and dependence are determined by the Mach number. However, in the case of subsonic flow (elliptic behavior instead of hyperbolic), you theoretically have to extend your domain to infinity. I say "theoretically", because it actually depends on (a) which boundary conditions you apply, (b) which numerical scheme you use, (c) how strongly disturbances are smoothed by numerical dissipation.... and other things. In short: I would think that everyone who has done this problem has his own "empirical relation" from experience, but those experiences are not necessarily valid for your code and test case.

One thing for your bluff body: It will be most important to resolve the wake, and of course the flow around the body in order to get separation and vortex shedding right. Maybe you can describe your problem in more detail.