Date of Award


Degree Type


Degree Name

Doctor of Philosophy (PhD)


Aeronautics and Astronautics

Committee Chair

Inseok Hwang

Committee Member 1

Daniel DeLaurentis

Committee Member 2

Denfen Sun

Committee Member 3

Seokcheon Lee


In the aftermath of natural disasters, the optimal evacuation planning plays a crucial role in saving people’s lives by providing plans to timely extricate victims to safe locations. For the evacuation in the aftermath, aerial vehicles (e.g., helicopters) become efficient assets when a disaster area is difficult to access by ground vehicles due to the destruction of the road network, or the large number of victims affected. In this regard, a mathematical model and efficient solution is necessary for optimal evacuation planning using a fleet of aerial vehicles, as proposed in this dissertation. The Optimal Evacuation Planning Problem (OEPP) is a task assignment problem combined with scheduling for the aerial vehicles in order to deliver the disaster victims to designated safe locations. In the mathematical model, practical characteristics of two different types of actors, vehicles and victims, are explicitly considered. Vehicles with different capabilities (e.g., cruise speed, capacity, or endurance) are distributed at multiple bases. The objective is to deliver the victims with different levels of urgency located at various locations to numerous specified safe locations (e.g., hospitals and refuges). The information of the vehicles and victims can be predetermined from the assessment phase. In the initial stage of disaster, the information may also contains time–varying uncertainties. In this dissertation, both the predetermined information and the information under uncertainty are considered. In order to provide the optimal evacuation plan based on the predetermined information, the OEPP is formulated as an Integer Linear Programming (ILP) with the goal to maximize the number of evacuees while satisfying operational constraints (i.e., the capacity and endurance of vehicles, and different urgency levels of the victims). The ILP, however, is intractable for a large–scale disaster problem. Thus, an efficient algorithm called as a Cooperative Multiple Agents based Algorithm (CMA) is proposed to solve the large–scale problem in a reasonable period of time. The computational efficiency and performance of the algorithm are demonstrated using illustrative numerical examples based on realistic scenarios. Note that the CMA does not explicitly account for the uncertainties. As a consequence, when the information contains time–varying uncertainties, the performance of evacuation based on plans computed by the CMA can be degraded. In order to overcome the limitation, the Stochastic Dynamic CMA (SDCMA) is proposed to explicitly account for time–varying uncertainties. The performance of SDCMA is demonstrated by numerical comparisons between the CMA and the SDCMA. The numerical simulations based on realistic data demonstrate that the proposed SDCMA could provide reliable and superior evacuation plans for the evacuation problem under time–varying uncertainty.