Sewage sludge is an unavoidable by-product in the process of sewage treatment. Due to its characteristics of high pollution and difficult to treat, the efficient and harmless treatment of sludge is still facing certain challenges. A new sludge treatment technology, spray drying technology with relatively simple process and high value-added utilization after sludge atomization has a great promotion effect on sludge treatment. The gas-assisted atomizer has a good atomization effect on high-viscosity fluids, and can ensure a better atomization quality at a faster atomization rate. In the study of sludge atomization, many scholars have carried out certain researches, but there is no numerical simulation study of sludge atomization, and numerical simulation can be low-cost, more intuitive study of sludge atomization breakup, has certain advantages. In order to realize the numerical simulation of sludge atomization and breakup, computational fluid dynamics software is used to explore the influence of gas-assisted sludge atomization characteristics and operating parameters (gas velocity, gas-liquid ratio, spray angle) on the effect of sludge atomization. The results show that the density and viscosity of sludge gradually decrease with the increase of moisture content. Gas velocity, gas-liquid ratio and spray angle are the three most important operating parameters that affect sludge atomization and breakup. During the atomization process, the high-speed airflow makes the sludge vibrate and unstable at the front end of the atomizer, resulting in the tearing of the sludge and the breaking of the droplets. The density of droplets in the center area is greater than that in the edge area and there are a few large particles agglomerated. For sludge with moisture content of 87% and a density of 1.065×103 kg/m3, the atomization effect is the best when the gas velocity is 180 m/s, the gas-liquid ratio is 126.3, and the spray angle is 55°. The average particle size of the droplets is about 0.193 mm, and the experimental results are in good agreement with the simulated particle size, and the maximum relative error is 5.80%.