Why Two Buses Arrive Together After a Long Wait
Bus bunching creates self-reinforcing service gaps: one bus falls behind and collects more passengers while the next catches up. Real headways and rider experience matter more than scheduled vehicle counts.
One-Sentence Insight
More buses do not guarantee shorter waits if vehicles cluster together. Reliability depends on how evenly service arrives, not just scheduled frequency.
Why It Matters
A network may advertise a bus every ten minutes, yet riders wait twenty minutes and then see two arrive together.
This clustering can reinforce itself: the late bus collects more passengers and loses more time, while the following bus carries fewer riders and catches up. Extra fleet size alone need not resolve the problem.
Simple Example
Imagine six buses scheduled over an hour, one every ten minutes. If four arrive close together, a long service gap may follow.
The hourly vehicle count can remain correct while passenger waiting becomes unreasonable. Actual intervals tell the story that average frequency conceals.
How It Works
Drivers face traffic, signals, boarding time and stop delays. Dispatchers track vehicle positions and try to maintain regular headways.
Operational measures may include brief holding at suitable stops or service adjustments, depending on the network. Improving headways can inconvenience riders already on board, so trade-offs need testing.
Feedback Loops
Delay loop: leading bus slows → more passengers accumulate → longer boarding → further delay.
Catch-up loop: the following bus meets fewer waiting passengers → shorter dwell time → smaller separation.
Stabilizing loop: observed headways → safe control actions → more predictable service. This needs dispatch capacity and operational safeguards.
Evidence
ITDP's BRT planning guide explains that four buses per hour imply a 15-minute scheduled headway when arrivals are even.
The 2024 BRT Standard includes deductions for poor reliability and bus bunching. A 2021 National Academies synthesis reviews operational technologies for headway management. These sources are methodological, not evidence of a particular city's performance.
Economic Logic
Measure operating cost against user waiting and service reliability, not just the cost of extra buses.
Illustration: perfectly even ten-minute headways imply about five minutes of average wait for passengers arriving randomly. Uneven headways can increase expected waiting despite six buses still passing per hour. Real conditions depend on arrival patterns and headway distribution.
Limits
The five-minute estimate assumes random passenger arrivals and regular service. Riders on infrequent lines may instead time their arrival.
Road safety, driver duty, infrastructure and stop capacity take priority. No algorithm should direct unsafe stops or passing just to improve a metric.
What's Often Missed
Agencies often count buses and vehicle-kilometers. Riders experience uncertainty: when will the bus arrive and will they reach work on time?
A crowded vehicle followed by an empty one can indicate irregular timing rather than simply insufficient capacity. Operational control may be worth testing before fleet expansion.
Critical View
Headway control can reduce waits, but holding a vehicle for too long harms on-board passengers and connections.
Compare wait time, in-vehicle delay, accessibility and full-journey reliability. Software vendor claims require field validation, not acceptance by slogan.
Sidy's Synthesis
My synthesis: transit capacity should be measured in reliable journeys, not the number of buses purchased. Vehicle procurement is visible; passenger waiting is harder to audit.
Start with one corridor and three observations: actual headways, waiting time and passengers left behind. Identify where delays begin to compound.
Before buying more fleet, test safe, modest interval management. If it fails, the evidence can still reveal where extra capacity is genuinely required.
AI and Future Lens
Now. Vehicle location can reveal actual headways, provided timestamps and route matching are valid.
Five years. Better operations tools may anticipate boarding variation.
Ten years. Simulations may support more informed frequency and transfer planning.
Twenty years. Passenger safety and experience remain goals, not mere optimization parameters.
Operational safeguard. A dispatch optimization must show how it affects people already on board, connections, accessibility and drivers' safety. Better average headways do not justify unsafe vehicle instructions or leaving vulnerable passengers behind.
Build From This
Build a bus headway reliability dashboard. It informs dispatchers and never automatically instructs drivers.
Inputs: route, stop, planned and observed passage times, and approximate passenger demand, without individual tracking.
Outputs: service-gap map, irregular intervals, estimated waits and problematic periods, separating estimates from observations.
Pilot: one route over five days with operator authorization. Cross-check records in the field; service interventions need responsible safety review.
First Steps
Record ten consecutive buses at one stop. Calculate actual intervals, ask willing passengers about their waits and compare the observations.
Remember This
Six buses per hour do not guarantee a bus every ten minutes. Delays can reinforce themselves. Regularity matters to passengers. Safety comes before algorithms.
Primary sources
Facts, figures and quotations should be traceable to the sources below. Sidy's synthesis is labeled as synthesis and does not replace sourced facts.