Quick proof of concept, and thus, pretty rough radiosity.
After weeks of using Lightwave 11's instancing and buffer outputs, I realised I hadn't even touched fracture or the bullet engine. Figured it was a good chance to test everything out, especially with more complicated particle compositing in After Effects.
Lightwave 11: modelling (Fracture), dynamics (Bullet engine), lighting (again, this is just draft quality), camera animation, and rendering (with the latest buffer export options).
After Effects CS5.5: explosive hits and dust (Trapcode Particular), DOF blur (native Lens Blur), lens effects (VideoCopilot OpticalFlares), volumetric lighting (Trapcode Shine), noise reduction (Magic Bullet Denoiser II) and grading (Magic Bullet Looks).
The trickiest bit was getting Particular to work well with Lightwave's depth map output...which happens to be inverted and linear. Once that was adjusted to Particular's liking, the particles actually intersected with the geometry pretty reliably, albeit with a lot of unfortunate banding artefacts. Though the depth maps were rendered at 32bpc, more stringent depth limitations (instead of 20m, maybe 10m - the wall is only 9m in length) and a 16bpc format might work better (I have a hard time trusting Adobe with FP renders, for some reason).
While fully volumetric 3D dust would be significantly more realistic, this was incomparably faster to render! Total Lightwave render time was around 20 minutes, and After Effects was around 10. Like I said, super rough. It's just a proof of concept.