This paper reviews existing Monte Carlo techniques for performing neutron transport simulations in binary random heterogeneous fissile fuels and presents a new approach offering superior efficiency at little cost in fidelity for problems involving densely packed, optically thick absorbers. The accuracy of the chord-length sampling technique is demonstrated to be a function of the total optical thicknesses and optical scattering thickness of the constituent materials as well as the packing density of the fissile kernels. The results of this parameter assessment provide a foundation for an original hybrid algorithm that combines homogeneous and explicit geometry models within a single Monte Carlo simulation. The geometry model utilized is selected according to the energy-dependent optical thickness. By partitioning the geometry representation within a single Monte Carlo simulation into homogenous and heterogeneous energy-dependent models, acceptable ensemble average results are obtained in a fraction of the run time of the detailed explicit geometry benchmark method.