A proper setting of boundary conditions is a standard upon which simulation results are justified. This study is specifically designed to simulate airflow over a repeating unit of simplified urban models with the application of periodic boundary condition. Similar setting of boundary conditions is used for all models which are of square layout with 25% packing density. The models are constructed with such that the initial velocity field is uniform throughout its internal domain. The results show that different domain heights of 4h and 5h (h as the building height) do not affect the spatial averaging of velocity profiles. In terms of the number of grids per building height, a finer meshing of 32 grids produce more accurate results of velocities and turbulence intensities compared with those of 25 grids when validated against the previous direct numerical simulation (DNS) data. Nevertheless, these criteria depend upon longer averaging period for better estimation of flow statistics. The boundary condition setting used in this preliminary study is nevertheless capable of producing current results comparable to past data although future works should focus on optimizing the important criteria in a simulation such as domain height, grid numbers, and averaging time.