Large facilities in urban areas, such as storage facilities, distribution centers, schools, department stores, or public service centers, typically generate high volumes of accessing traffic, causing congestion and becoming major sources of greenhouse gas (GHG) emission. In conventional facility-location models, only facility construction costs and fixed transportation costs connecting customers and facilities are included, without consideration of traffic congestion and the subsequent GHG emission costs. This study proposes methods to find high-demand facility locations with incorporation of the traffic congestion and GHG emission costs incurred by both existing roadway traffic and facility users into the total cost. Tabu search and memetic algorithms were developed and tested with a conventional genetic algorithm in a variety of networks to solve the proposed mathematical model. A case study to determine the total number and locations of community service centers under multiple scenarios in Incheon City is then presented. The results demonstrate that the proposed approach can significantly reduce both the transportation and GHG emission costs compared to the conventional facility-location model. This effort will be useful for decision makers and transportation planners in the analysis of network-wise impacts of traffic congestion and vehicle emission when deciding the locations of high demand facilities in urban areas.
Improved Predictive Ability of KPLS Regression with Memetic Algorithms
