Impact of road network summarization on facility location decisions: A case study of the flow-capturing location-allocation problem

Typical facility location problems for traffic flow in road networks can be formulated as mixed-integer programming problems. For such problems, the number of components of a road network significantly affect the computational costs of searching for an optimal solution. Therefore, obtaining exact solutions for large-scale road networks incurs high costs. In practice, although simplifying the network to remove minor roads alleviates this problem, this simplification cannot be applied to scenarios in which minor roads are essential. Hence, to reduce the computational costs of facility location problems for road networks that include minor roads, using summarized networks that retain information on minor roads via methods such as node aggregation is necessary. However, it is unclear to what extent the optimal solutions found in these summarized networks can reproduce the solutions of the original networks. Therefore, this study confirmed the validity of the summarized road networks by comparing the optimal solutions and computational costs of the summarized networks with those of the original and simplified networks. In the experiment, the flow-capturing location-allocation problem was used as a representative facility location problem. This evaluation elucidates the impact of road network summarization on decision-making regarding facility location.