Master of Science (MS)
Geography and Anthropology
In the face of severe disasters, some or all of the endangered residents must be evacuated to a safe place. A portion of people, due to various reasons (e.g., no available vehicle, too old to drive), will need to take public transit buses to be evacuated. However, to optimize the operation efficiency, the location of these transit pick-up stops and the allocation of the available buses to these stops should be considered seriously by the decision-makers. In the case of a large number of alternative bus stops, it is sometimes impractical to use the exhaustive (brute-force) search to solve this kind of optimization problem because the enumeration and comparison of the effectiveness of a huge number of alternative combinations would take too much model running time. A genetic algorithm (GA) is an efficient and robust method to solve the location/allocation problem. This thesis utilizes GA to discover accurately and efficiently the optimal combination of locations of the transit bus stop for a regional evacuation of the New Orleans metropolitan area, Louisiana. When considering people’s demand for transit buses in the face of disaster evacuation, this research assumes that residents of high social vulnerability should be evacuated with high priority and those with low social vulnerability can be put into low priority. Factor analysis, specifically principal components analysis, was used to identify the social vulnerability from multiple variables input over the study area. The social vulnerability was at the census block group level and the overall social vulnerability index was used to weight the travel time between the centroid of each census block to the nearest transit pick-up location. The simulation results revealed that the pick-up locations obtained from this study can greatly improve the efficiency over the ones currently used by the New Orleans government. The new solution led to a 26,397.6 (total weighted travel time for the entire system measured in hours) fitness value, which is much better than the fitness value 62,736.3 rendered from the currently used evacuation solution.
Document Availability at the Time of Submission
Release the entire work immediately for access worldwide.
Qin, Xiaojun, "A social vulnerability-based genetic algorithm to locate-allocate transit bus stops for disaster evacuation in New Orleans, Louisiana" (2009). LSU Master's Theses. 1838.