Icing can adversely influence electric power system security. Two main issues are caused by icing: the overload of transmission lines, and the reduction in the insulation ability of the insulators. Most previous research has focused on the flashover characteristics of ice-covered insulators, but research on the icing process of the insulator is seriously lacking. Considering the effect of icing shape, the outer airflow field of an insulator was calculated and the local collision efficiencies of water droplets (β₁) were investigated according to the Lagrange algorithm. The simulation showed that the values of β₁ on the insulator edge and rod are much higher than on the insulator surface, and both were significantly influenced by the wind speed and median volume diameter (MVD) of the water droplets. Based on thermal balance equations, a dynamic dry-growth icing model was established. Using the natural icing conditions of Xuefeng Mountain (China) as an example, validation experiments were conducted on a composite insulator and the climate parameters measured by multi-cylinders were used to model the icing shape and mass. The results indicate that high wind speed and low temperature increase icing rate; the icing was mainly concentrated on the windward side and the greatest horizontal thickness was generally on the insulator edge. The dry-growth model had an average error lower than 25% for icing thickness and an average error lower than 20% for icing mass, which were affected by icing roughness.