Recent developments in dynamic penetration testing made it possible to measure a force-displacement response of the soil during each single blow. Mechanical properties other than the classical tip resistance could be deduced and possibly linked to properties usually measured from model tests. However, the loading process implied in penetration test is highly non homogeneous and very different from those of laboratory model tests. It is then important to find out how to link the properties obtained from both kinds of tests. As a preliminary step in this process, a numerical model was built to reproduce penetration tests conducted in static and dynamic conditions. Two-dimensional Discrete Element Method, based on molecular dynamics was used. A rod was driven in a confined sample either with a constant velocity (static conditions) or by applying a blow on it (dynamic conditions). The magnitudes of rod velocity used in both static and dynamic conditions tests were similar. The model was validated based on the qualitative comparison between classical experimental results and numerical results. The repeatability of numerical tests was also checked in terms of tip resistance and volume deformations.