Deep soil mixing to construct stiff columns is one of the methods used today to improve performance of loose ground and remediate liquefaction problems. This research adopts a numerical approach to study seismic performance of soil-cement columnar reinforcements in loose sandy profiles. Different constitutive models were investigated in order to find a model that can properly predict soil behavior during seismic excitations. These models included NorSand, Dafalias-Manzari, Plasticity Model for Sands (PM4Sand) and Pressure-Dependent-Multi-Yield-02 (PDMY02) model. They were employed to predict behavior of soils with different relative densities and under different confining pressures during monotonic and cyclic loading. PDMY02 was identified as the most suitable model to represent soil seismic behavior for the system studied herein. The numerical aspects of the finite element approach were investigated to minimize the unintended numerical miscalculations. The focus was put on convergence tolerance, solver time-step, constraint definition, and, integration, material and Rayleigh damping. This resulted in forming a robust numerical configuration for 3-D nonlinear models that were later used for studying behavior of the reinforced grounds. Nonlinear finite element models were developed to capture the seismic response of columnar reinforced ground during dynamic centrifuge testing. The models were calibrated with results from tests with unreinforced profiles. Thereafter, they were implemented to predict the response of two reinforced profiles during seismic excitations with different intensities and liquefaction triggering. Model predictions were compared with recordings and the possible effects from the reinforcements were discussed. Finally, parametric studies were performed to further evaluate the efficiency of the reinforcements with different extension depths and area replacement ratios. The results collectively showed that the stiff elements, if constructed appropriately, can withstand seismic excitations with different intensities, and provide a firm base for overlying structures. However, the presence of the stiff elements within the loose ground resulted in stronger seismic intensities on the soil surface. The columns were not able to considerably reduce pore water pressure generation, nor prevent liquefaction triggering. The reinforced profiles, comparing to the free-field profiles, had larger settlements on the soil surface but smaller settlements on the columns. The results concluded that utilization of the columnar reinforcements requires great attention as these reinforcements may result in larger seismic intensities at the ground surface, while not considerably reducing the ground deformations. / Ph. D.