The Impact of Road Networks on Facility Location Modeling: An Interpretation from a Network Topological Perspective

This research investigates the impact of the road network topological structure on facility location modeling. We create four types of road networks, i.e., the radial, the grid, the ring, and the ring-radial networks, and characterize them using centralization indices from social network analysis and connectivity indices from planar network analysis. We perform the p-median location-allocation model on each road network. To control for confounding factors, we incorporate the facility capacity, the road type, the demand locations, and the demand quantity in the model. On each road network and with every combination of confounding factor levels, we perform the p-median location-allocation model 100 times and use the large result samples to conduct spatial and statistical analysis. The radial and ring networks tend to have long facility-demand connecting paths along the radial or ring roads. The grid network has mostly short and localized connecting paths. The ring-radial network has a mix of long and short connecting paths. The optimal models come from the grid network, followed by the ring-radial network. The worst result is from the ring network, followed by the radial network. The analysis of variance confirms this result under all combinations of the four control factors. We attribute the spatial pattern of connecting paths and the model result to the topological characteristics of the networks. The grid network has the highest connectivity and distributed centrality. Both the ring and radial networks have low connectivity. It is interesting that blending these two poor-performing networks into the ring-radial network improves connectivity and generates the second-best result. Understanding the relationship between the road network topological structure and facility location modeling implies that transportation development should be focused on improving connectivity rather than just building more roads.