This document presents a study on developing a hierarchical approach to pathfinding to speed up crowd simulation. It proposes a method using navigation meshes partitioned into a hierarchy of graphs. The preprocessing step involves partitioning the navigation mesh using multilevel graph partitioning. This creates a hierarchy of graphs with portals between partitions. The online pathfinding computation searches the hierarchical graph from the highest level to find an optimal path between start and goal positions faster than searching the entire navigation mesh graph. Experimental results show the hierarchical approach finds paths up to 81% faster with much lower memory usage compared to searching the full navigation mesh graph.