Dynamic compaction of soils is an efficient and cost effective ground improvement technique, allowing for the use of sites previously considered unsuitable for construction. The technique consists of densifying loose granular materials by applying high energy impacts to a soil's surface by dropping a heavy weight carried by a crane at a given height.  Field compaction predictions based upon laboratory Proctor test results generally result in great discrepancies with the obtained field results. A Proctor test sample is restrained both laterally and at its bottom, whereas a field sample is free to move in three dimensions. Therefore, the boundary conditions of the Proctor test are incompatible with those of field compaction.  This thesis presents a numerical model capable of examining both the field and laboratory boundary conditions of a soil sample undergoing dynamic compaction. It was found that the boundary conditions of the Proctor test are incompatible with those of dynamic field compaction and that the stiffness of the underlying layer plays a role in determining the level of compaction experienced by the overlying layer. This relationship was further explored by accounting for varying thicknesses of the upper layer with a range of stiffness values for the underlying layer. A trend of decreasing compaction with increasing upper layer thickness was observed when the underlying layer's modulus of elasticity exceeded that of the upper layer. Also, compaction of the upper layer increased as the elasticity modulus of the lower layer increased for upper layer thicknesses of 1 and 2 m