Bus Network Planning for Transfers and the Network Effect in Melbourne, Australia

Although passengers dislike transferring, efficient transit systems should facilitate transfers to provide auto-competitive citywide access. This paper reviews bus transfer behavior in Melbourne, Australia, to understand causal factors. It also explores network effects: high ridership associated with frequent services or simple (grid) networks or both. Half of bus users make transfers, mostly to rail. The number of tram and bus transfers is generally low; however, in inner and central business district areas where trams run, they dominate. The young, students, males, commuters, and riders with periodical or full-fare tickets have high bus transfer rates. Middle-aged and older groups, off-peak riders, riders on shopping trips, and concession ticket holders have lower rates. Weekday peak hour commuters have high transfer rates. Frequent or longer-distance bus route types, simpler or straight route alignments, commuter services, and routes that require transfers have high transfer rates. Schedule coordination with rail increases transfer rates, but only a little. A modest but significant relationship (R-2 = .25) between the volume of transfer trips and average service frequency was shown. High transfer volumes occur where at least one route has a frequency of 10 min or better. Current bus plans propose a 15-min based grid route network. These findings suggest that major routes need at least 10-min headways to generate high transfer rates. A high bus transfer subnetwork was identified with features consistent with the network effect; however, conclusive proof of the network effect remains elusive. In simple terms the network effect, though intriguing, remains an unsubstantiated theory that informs good practice but should be treated with caution when applied in the real world.