Because of their favorable thermal conductivity and commendable fuel performance, dispersed fuel elements have been widely adopted in research reactors. However, the geometric complexity resulting from the high packing density of dispersed fuel poses challenges to current geometric modeling methods. Various implicit and explicit modeling techniques have been implemented in Monte Carlo codes. Implicit methods struggle to achieve the correct packing fraction and, therefore, lack accuracy. In this paper, we introduce the Optimized Dropping and Rolling and Virtual Surface method for precise and efficient geometric modeling of such structures. Simultaneously, we apply this novel method to the packing process of an annular container. Test results across various packing fraction cases demonstrate that our modified Optimized Dropping and Rolling method significantly enhances packing efficiency, surpassing the Distinct Element Method hundreds of times.