Abstract

In most urban cities, the correct design of the bus routes network is a critical task to design a successful transportation system. Thus, network designers rely on solving the Urban Transit Routing Problem (UTRP) to find a set of bus routes that minimises users travelling time and system operator costs. Most UTRP instances lack real-life demand data information, are very small w.r.t. typical scenarios or standards, or are randomly generated. Moreover, state-of-the-art dimensional reduction techniques use inherent features of urban transportation systems; therefore, they cannot significantly reduce the order of magnitude in complex cases. This article proposes a framework for generating UTRP instances that shrink the complexity of substantial real-life scenarios through well-known clustering algorithms and geographic database applications. The generated instances preserve the original demand behaviour and road structure by connecting central and periphery locations. We discuss and validate this proposal by applying it in a complex study case, the RED system from Santiago of Chile.