In this research, an optimization framework is proposed to find the system-optimal solution for the emergency evacuation planning problem. The model simultaneously solves for the choice of destination (shelter), evacuation route, and departure time for all evacuees from given origins (endangered zones). Heuristic algorithms are proposed to find the system-optimal dynamic traffic assignment during an iterative procedure. A mesoscopic traffic simulator is embedded in the optimization modules to find travel time marginals and obtain required performance measures in every iteration. Implementing the proposed algorithm brings about dramatic improvements in network performance measures such as total travel time and network clearance time in a case study of Ocean City, Maryland.