The migration of fines suspended in the liquid through porous media represents a key challenge in many chemical engineering processes, yet the inherent fundamentals and roles of clusters were not well elucidated. In this work, we numerically study the formation, growth, and connection of clusters in a porous medium and unveil the clogging mechanism of fines at pore-scale, i.e., bridging and locking essentially. A saturation volume fraction of fines in the entrance occurs at the critical transition point from depth filtration to caking, where the volume fraction is calculated by the spherical cap method. An inter-particle contact pair method is developed to quantify the cluster size growth. Inherently, an exponential correlation between clogging ability and size ratios is obtained using the cluster information - growth rate of clusters. A probabilistic model of clogging ability and performance is derived and used to verify the exponential formula. This work provides an effective method to quantify clogging ability and could be applied to more complex systems.