A statistical study of displacement cascades induced by silicon Primary Knock-on Atoms (PKA) in bulk silicon is performed by running a large number of molecular dynamics (MD) simulations. The choice of the PKA species and energy varying from 1 to 100 keV comes from a previous particle-matter simulation [1]. The electronic stopping power missing in standard MD simulations is here taken into account using the Two Temperature Model (TTM). This prevents from overestimating the number of created defects. The damaged atomic structures obtained after one nanosecond of MD simulation are not representative of what is observed in image sensors for example after several minutes. For this reason, the kinetic Activation Relaxation Technique (k-ART) is used in a second step, allowing to access longer simulation times of up to second. The obtained damaged structures can then be compared with experimental observations. Analyses reveal two possible links between the simulated structures and the measurements in solid-state image sensors. First, the cluster size distribution exhibits a shape similar to the measured exponential distribution of Dark Current (DC). Second, the temporal evolution of metastable atomic configurations resembles experimental DC-Random-Telegraph-Signals.