Glad I found this topic. Was going to make a bug report...
Exploding a sphere divides it into 13 smaller spheres that all fit within the original sphere. The original mass is divided among the resulting spheres and the high density results because the mass is the same, but the total volume is less.
You're right that this is unrealistic. Feel free to share your thoughts on how this could be improved. I look forward to improving the explosion code in the next few months.
Thanks for pointing out this problem.
Why are you doing it like this? is there a technical limitation in US why you can't use a better approximation (there are several)?
Some simple methods that come to mind, which at least would be better approximations than the current:
First expand the original body radius and only after that divide to spheres. That's actually how it would "approximately" happen in reality (considering something blowing up due to internal pressure, for example). The math for this is simple (since you already have the part down of dividing to spheres contained in equal volume, you can even use that for derivation of the required fraction) and this would be infinitely better approximation of reality.
Or first transform the body to a cube with equal mass and density and then divide that to cubes of required number. Dunno how your engine would cope with cubes, but since there are dice in the catalog, I suspect this would be simple too.
I guess using conic sections would be pushing it.