STUDIES IN LARGE-SCALE EVACUATION NETWORK FLOW STOCHASTIC OPTIMIZATION UNDER SOCIAL INFLUENCE
Open Access
- Author:
- Na, Hyeong Suk
- Graduate Program:
- Industrial Engineering
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- May 03, 2019
- Committee Members:
- Soundar Kumara, Dissertation Advisor/Co-Advisor
Necdet Serhat Aybat, Committee Chair/Co-Chair
Tao Yao, Committee Member
S. Ilgin Guler, Committee Member
S. Ilgin Guler, Outside Member - Keywords:
- Cell transmission model
departure time choice model
social influence
large-scale hurricane evacuation
joint chance constrained program
sample average approximation
Large-scale evacuation network flow model
Cell Transmission Model
Joint chance constrained program
Sample average approximation
Departure time choice model
Social influence
Capacity drop
Hurricane evacuation strategy - Abstract:
- Evacuation is indispensable for protecting the land and lives in the target areas that are likely to be hit by big hurricanes. In the aftermath of recent hurricanes such as Irma, Maria and Michael, the importance of a tactical evacuation plan has been emphasized once more in order to avoid heavy casualties and substantial social costs. However, the inherent uncertainty in departure time choice of the evacuees causes numerous challenges to emergency management agencies. On the other hand, social media is transforming the communication paths to share the evacuation-related information in real-time; therefore, one of the main factors influencing the evacuation departure time of the evacuees is definitely social media. That said, in spite of the significant effect of the evacuation departure uncertainties on the evacuation traffic flows, there are only very few evacuation studies considering the effect of social influence. Furthermore, during an evacuation, serious traffic congestion can arise due to the sudden increase of traffic flow on the hurricane evacuation routes. In general, traffic congestion occurs when demand exceeds the capacity of road networks, which is followed by forming queues upstream of bottlenecks. One of the puzzling phenomena in traffic is that the capacity can suddenly drop at active bottlenecks; as a result, such capacity drop phenomenon can cause more serious gridlock. Many empirical experiments indicate that the magnitude of capacity drop differs under the influence of several factors and ranges from 3% to 20%. In particular, during the hurricane evacuation, traffic congestion may lead to not only evacuation impediments but also lower compliance rates. As such, the uncertainty in road capacity caused by the capacity drop phenomenon makes it harder to facilitate the evacuation efforts. In this dissertation, we first propose a stochastic network flow model to generate a hurricane evacuation plan considering the effect of social networks in the affected areas. We describe human evacuation behavior using a time inhomogeneous, discrete-time Markov chain, which corresponds to adopting a stochastic S-shaped response curve to model the departure trend of evacuees. The proposed model is developed based on the Cell Transmission Model with joint chance constraints and it is handled by the sample average approximation method and Monte Carlo simulation techniques. Subsequently, we contemplate the efficient evacuation strategy to cope with complex traffic flow under uncertainty in road capacity. This is because Cell Transmission Model cannot fully consider more complex traffic phenomena such as capacity drop. Hence, we define the new fundamental relation between traffic flow and density to describe the capacity drop phenomenon. Based on the fundamental relation, two alternative models are proposed to incorporate the capacity drop phenomenon into the traffic flow model. The objective of these models is to maximize the utilization of the traffic network during the evacuation period before a hurricane landfall. Several numerical case studies are provided to examine the applicability of the proposed models and to understand the effect of the evacuation traffic flows on various evacuation scenarios with different road capacities and human evacuation behavior. We believe that these evacuation studies can serve as a useful tool for decision-makers such as emergency management agencies to model a large-scale hurricane evacuation plan. In addition, our large-scale evacuation models can be usefully leveraged when a smart city is designed with autonomous and connected vehicles considering the regulation of increased traffic flow and cybersecurity. This would involve interdisciplinary collaboration with emergency management agencies and various urban scientists and planners to develop a system to meet their needs.