This study investigated breakthrough curves (BTCs) from a series of column experiments, including different column lengths and flow rates, of a conservative tracer, tritium oxide (HTO), and a radionuclide, cesium, in crushed granite using a reactive transport model. Results of the short column, with length of 2 cm, showed an underestimation of the retardation factor and the corresponding HTO BTCs cannot be successfully modeled even with overestimated fluid dispersivity. Column supporting elements, including filters and rings, on both ends of packed granite were shown to be able to induce additional dispersive mixing, thus significantly affecting BTCs of short columns while those of the long column, with length of 8 cm, were less affected. By increasing flow rates from 1 mL/min to 5 mL/min, the contribution of structural dispersive mixing to the false tilting of short column BTCs still cannot be detached. To reduce the influence of structural dispersivity on BTCs, the equivalent pore volume of column supporting materials should be much smaller than that of packed porous medium. The total length of column supporting structures should be greatly shorter than that of porous medium column.