Evacuation Planning: A Capacity Constrained Routing Approach

Approa h y Qingsong Lu, Yan Huang, Shashi Shekhar Department of Computer S ien e, University of Minnesota 200 Union St SE, Minneapolis, MN 55455 [lqingson; huangyan; shekhar℄ s.umn.edu February 10, 2003 Abstra t Eva uation planning is riti al for appli ations su h as disaster management and homeland defense preparation. EÆ ient tools are needed to produ e eva uation plans to eva uate populations to safety in the event of atastrophes, natural disasters, and terrorist atta ks. Current optimal methods su er from omputational omplexity and may not s ale up to large transportation networks. Current naive heuristi methods do not onsider the apa ity onstraints of the eva uation network and may not produ e feasible eva uation plans. In this paper, we model apa ity as a time series and use a apa ity onstrained heuristi routing approa h to solve the eva uation planning problem. We propose two heuristi algorithms, namely Single-Route Capa ity Constrained Planner and Multiple-Route Capa ity Constrained Planner to in orporate apa ity onstraints of the routes. Experiments on a real building dataset show that our proposed algorithms an produ e lose-to-optimal solution, whi h has total eva uation time within 10 perent longer than optimal solution, and also redu e the omputational ost to only half of the optimal algorithm. The experiments also show that our algorithms are s alable with respe t to the number of eva uees.