Currently the random ray solver does not support true void materials, so low cross section (near void) materials are required. In many fusion models, true void is often assumed. It is easy to explicitly treat true void regions in random ray, but the flux attenuation and source update functions need to branch to follow a different set of logic to handle this. The logic is not an approximation, but simply stems from a different derivation of the characteristic equation where Sigma_t is assumed to be zero. Adding in this treatment is fairly simple and doesn't require any special user input etc.