The soil arching effect is one of the most important load transfer mechanisms for piled embankments. This study performed three-dimensional discrete element method-based trapdoor tests to investigate the evolution of soil arching in unreinforced piled embankments with different fill heights and pile coverage ratios. This study measured vertical and horizontal earth pressures at different locations and calculated the lateral earth pressure coefficients. Based on the variations of vertical pressures and lateral earth pressure coefficients, this study determined the soil arch heights and proposed an empirical formula to calculate soil arch heights. The distributions of contact force chains and principal stress vectors were evaluated to reveal the microscopic behavior of soil arching. Results showed that piled embankments with a higher relative fill height had a higher soil arching effect and higher resistance to the soil arching degradation, while piled embankments with a lower pile coverage ratio had lower resistance to the soil arching degradation. The soil arching model in DEM-simulated tests with high pile coverage ratios was composed of a spherical dome arch, four plane arches, and four pyramid rigid cores above the piles. The DEM-simulated tests with low pile coverage ratios had same soil arching model as those with high pile coverage ratios except that the dome arch and the plane arch have supports in the middle due to the reduced pile zone.