Abstract: To ensure a continuous and stable operation, a certain amount of catalysts is discharged in increments, whereas several fresh catalysts are added to retain the activity of the previous catalysts in the equipment. The discharged catalysts are a mixture of various activities catalysts and this discharge wastes resources, thus leading to potential environmental hazards and causing difficulties in waste treatment, so all of those need comprehensive disposal for resourceful recovery and harmless treatment. According to theory of air acceleration classification, calculation was made with the Fluent software and EDEM software, based on the standard k-ε turbulence model and the Eulerian model. The coupled CFD-EDM simulation of fluidized-bed catalyst separation was performed to analyze the state of particle and column. The statistics of particle in the classification column obtained from simulations can help to study further of classification mechanism and particle movement in column. Different active catalysts obtained different acceleration behaviors because of different densities, and the acceleration is lower in lowly active catalysts than in highly active catalysts. Therefore, the positive velocity is lower and the negative velocity is higher in lowly active particles than in highly active particles. So highly active catalysts raised with air to the discharge outlet at the top of the separation column, while the lowly active catalysts gradually dropped to the discharge outlet at the bottom of the separation column. Under the conditions, the fluid field and particle state all expressed periodic changing and the fluidized-bed catalysts of 50% and 100% activities achieved separated after 9 cycles in column. This study also conducted the experiments on the discharged fluidized-bed catalysts and efficiency of air acceleration classification generally increased then declined with the increase in air velocity, and peaked at the optimal separation efficiency rate of 89.0%. The separation efficiency of simulate materials was 80.1%. The results showed that catalysts in larger activity difference were easier to classified. The simulation results are consistent with the experimental results, which verifying the feasibility of airflow acceleration classification applied in catalyst activity classification.