Access-based optimization and planning: An application of genetic algorithms to facility and network planning in Ferkessedougou Departement, Cote d'Ivoire.

摘要

Conventional transportation project planning makes use of cost-benefit analysis to determine which projects will be implemented and which will not. However, ex-post analyses of road project impact indicate that the most consistent benefits derived from roads are social, not economic. Access-based optimization and planning focuses on the social impacts of transportation planning and proposes an integrated framework for combining transportation and social facility planning. When treated as a network optimization problem, access can be improved either by adding more links in a network or by adding more social facilities to existing nodes.; This research uses software based on modified genetic algorithms to allow for the rapid solution of this type of network optimization problem. Using field data from Ferkéssédougou département, Côte d'Ivoire, for the years 1975–1988, this research examines solution sets derived from this technique using four different optimization objective functions. These functions optimize access in the following terms: (1) maximizing the number of clinic visits (termed here the “maximize” method), (2) minimizing the total absolute deviation between village usage rates without regard to population size (“equity” method), (3) maximizing usage rates without regard to population size (“equimax” method), and (4) maximizing usage rates without regard to population size and without regard to villages with above average usage rates (“lower half” method).; The results indicate that apparently simple objective functions can yield counter-intuitive results. When using the “maximize” method, the clinics are all clustered around Ferkéssédougou town, while the needs of under-served villages in the hinterland are ignored. Since the location of a clinic produces an accessibility “spike” in the accessibility surface of a region, the “equity” method pushes new clinic projects to the periphery in order to minimize the impact of these spikes. The “equimax” method results in clustering similar to the “maximize” configuration, because even without population weightings, smaller villages tend to cluster around large towns like Ferkéssédougou. The “lower half” method produced the most intuitive configurations, with clinics being built in previously under-served areas. As an operational technique, access-based planning is thus best served by the use of this “lower half” objective function.