Sensitive detectors, as transition edge sensors, operating in X-ray instrumentation from space (or in laboratories) are susceptible to degradation due to photons at lower energies than X-rays. Although the detector does not trigger on individual low energy photons, the statistical fluctuation of the absorbed energy during the effective X-rays detection time, can introduce a degradation of the energy resolution of the detector known as Photon Shot Noise (PSN). Thermal filters are usually interposed between the optics and the detector to minimize the PSN due to out of band radiation from the instrument warm surfaces and from external light sources. The PSN strongly depends on the geometry of the system, since the photons rejected by a filter are mainly reflected, and after further reflections they can reach the filter again and again. Multiple reflections on the system surfaces have thus the effect of increasing the amount of unwanted energy that can reach the detector, raising the PSN. A careful design of the geometries and of their surface properties (reflectivity/emissivity) is necessary to guarantee the PSN levels needed for meeting the instrument requirements. This code, "ray3cer", has been developed to evaluate the PSN for a specified configuration, taking in account multiple reflections. A trivial Monte Carlo simulation is not apt, since the very high rejection given by the filters allows only for a small fraction of photons to reach the detector, and an unreasonable number of photons should be simulated to attain a passable statistic. The approach adopted is to calculate all the possible trajectories for each simulated photon, within a probability threshold limit, to evaluate the overall probability for the photon to reach the detector. This technique bypasses the limitation due to the filter rejection and allows for a simulation with a good statistic with a relatively small number of simulated photons.