Public Transit Transfer Planner

Door-to-door journey time is dominated by the parts of a trip that are not spent moving. This model sums in-vehicle time, distance divided by average commercial speed, with access and egress walking, plus a penalty for each transfer made up of waiting, roughly two minutes of boarding and alighting, and roughly three minutes of walking between platforms or stops. Average commercial speeds, which already include dwell at stops and time at signals, typically run 15 to 25 km/h for urban buses, 25 to 35 km/h for light rail and 30 to 45 km/h for metro. Because the fixed costs are large, effective door-to-door speed on a short urban trip often lands between 10 and 15 km/h even when the vehicle itself moves much faster. Waiting time depends on headway. Where service is frequent, under roughly 10 to 12 minutes between vehicles, passengers arrive at random and the average wait is about half the headway; where service is infrequent they time their arrival to the timetable, so the observed wait is shorter but the cost of a missed connection is far higher. Travel demand models have long treated a transfer as costing more than the clock time it consumes, with transfer penalties commonly valued at 5 to 15 minutes of equivalent in-vehicle time because of the uncertainty involved. The reliability figure here compounds a per-connection probability rather than a per-leg one, so at 85 percent per connection two transfers leave about 72 percent of trips with every connection intact, and a direct trip scores 100 percent because there is no connection to miss. Emissions comparisons assume a reasonably occupied vehicle; a near-empty late-night bus can emit more per passenger-kilometre than a car.

Systematic modelling of transit journey time came out of the large American urban transportation studies of the 1950s. The Detroit Metropolitan Area Traffic Study of 1953 and the Chicago Area Transportation Study of 1955 to 1962, directed by Douglas Carroll, produced the four-step sequence of trip generation, trip distribution, mode split and network assignment that still frames travel demand forecasting, and the 1962 Federal-Aid Highway Act made a continuing planning process of that kind mandatory for urbanized areas above 50,000 people. Explicit transfer penalties entered mode-choice work during the 1960s and 1970s, when researchers repeatedly found that riders weight waiting and transferring roughly two to three times more heavily than time spent seated in a moving vehicle. On the practitioner side, the Transit Cooperative Research Program was created in 1992 under the Intermodal Surface Transportation Efficiency Act, and its Transit Capacity and Quality of Service Manual, first issued in 1999 and revised in later editions, codified the frequency, speed and reliability measures agencies use today. Passenger-facing trip planning changed with the General Transit Feed Specification, developed in 2005 by Bibiana McHugh of TriMet in Portland together with Google engineers, which turned agency timetables into a shared open format that any application could read. Real-time arrival feeds through GTFS-Realtime followed early in the 2010s, which is what allows modern planners to estimate connection risk instead of assuming the printed timetable holds.