## Graduate Theses, Dissertations, and Problem Reports

Summer

2013

Thesis

MS

#### College

Statler College of Engineering and Mineral Resources

#### Department

Industrial and Managements Systems Engineering

#### Committee Chair

David R. Martinelli.

#### Abstract

In an event of a natural or man-made disaster, an evacuation is likely to be called for to move residents away from potentially hazardous areas. Road congestion and traffic stalling is a common occurrence as residents evacuate towns and cities for safe refuges. Lane reversal, or contra-flow, is a remedy to increase outbound flow capacities from disaster areas which in turn will reduce evacuation time of evacuees during emergency situations. This thesis presents a discrete-time traffic assignment system with lane reversals which incorporates multiple sources and multiple destinations to predict optimal traffic flow at various times throughout the entire planning horizon. With the realization of lane reversals, naturally the threat of potential head-on collisions emerges. To avoid the occurrence of such situations, a collision prevention constraint is introduced to limit directional flow on lanes based on departure time.;This model belongs to the class of dynamic traffic assignment (DTA) problems. Initially the model was formulated as a discrete-time system optimum dynamic traffic assignment (DTA-SO) problem, which is a mixed integer nonlinear programming problem. Through various proven theorems, a linearized upper bound was derived that is able to approximate the original problem with very high precision. The result is an upper bound mixed integer linear programming problem (DTA-UB). The discrete-time DTA model is suitable for evacuation planning because the model is able to take care of dynamic demands, and temporal ow assignment. Also, simultaneous route and departure is assumed and an appropriate travel time function is used to approximate the minimum and maximum travel time on an arc.;This thesis discusses the different attributes that relates to Dynamic Traffic Assignment. DTA model properties and formulation methodology are also expounded upon. A model analysis that breaks down each output into individual entities is provided to further understand the computational results of small networks. A no reversal DTA-UB model (NRDTA-UB) is formulated and its computational results are compared to DTA-UB. Through the extensive computational results, DTA-UB is proven to obtain much better results than NRDTA-UB despite having longer solving time. This is a step toward realizing the supremacy of having lane reversals in a real-life evacuation scenario.

COinS

#### DOI

https://doi.org/10.33915/etd.3401