The Monte Carlo method simulates the underlying processes that govern the system of interest (that is, the physical interactions between photons and their environment). A photon is selected from a photon source and tracked through the system until its weight falls below some minimum at which point it ‘dies.’ Each time the photon experiences an ‘event,’ a surface intersection, scattering or absorption for example, the physical quantities of interest are updated. This process generates a complete ‘history’ of that photon in the system. Many photon histories need to be generated to get good estimates of the physical quantities of interest in a system. Photon sources are selected (that is, ‘sampled’) on the basis of emitted radiation, each band being treated independently for non-gray models.

In CFX, the main computational overhead in generating a history is in tracking the photons across the domain. It is therefore essential to produce a balanced description of the domain to efficiently track the photons. This is done by using a coarser mesh for the radiation field than for the flow field under the assumption that the radiation field has less sharp changes than any other transport variables. When the domain material does not emit, absorb, and scatter radiation, there is no need for a mesh in the volume because the radiation transfer is between the boundary surfaces only (see Transfer Mode). For details on mesh coarsening controls, see Thermal Radiation Control.