An example of bus bunching seen on the Kings Road, London.

Bus bunching refers to two things: (1) a bus route having highly irregular service intervals, and (2) a classical theory for a causal model for irregular intervals, on the premise that a late bus tends to get later and later as it completes its run, while the bus following it tends to get earlier and earlier.

Classical Theory

The theory is that the two buses eventually form a pair, one right after another, and the service breaks down as the headway degrades from its nominal value. The buses that are stuck together are called a bus bunch or banana bus and may involve more than two buses. It is often theorized to be the primary cause of reliability problems on bus and metro systems.

Theorized Causes

Abnormal Passenger Loads

The time taken for a bus to complete its duties is related to the number of people attempting to board or alight at stops. The bus that is already late tends to attract a higher number of riders due to the longer service gap between it and the previous bus. The higher number of riders boarding the bus results in delaying it further.

Speed of Individual Drivers

Another cause is that some drivers are faster than others. This results in catching up on long or high-frequency routes.

Deliberate Acts

According to the article "Progress Has Passed Metrobus" by Lyndsey Layton (December 27, 2005) bus bunching may be deliberately caused by bus drivers, so that the bus ahead of them picks up more passengers and decreases their own workload.

The classical bus bunching theory is an example of chaos theory. The orderly procession of buses is presumed inherently unstable and buses are presumed to tend towards bunches if left unchecked. According to proponents of this theory, it is impossible to predict from the outset which buses will be bunched and which buses will proceed on schedule to the destination, because bunching is presumed to be caused by random conditions such as traffic, stoplights, and the number of passengers at a stop. The study of the actual practice of bus operations, however, has been shown to be quite different.

Practice

The existence of classical pairwise bunching has not been borne out by vehicle tracking systems data. Studies into metro operations have broadly debunked the theory of pairwise bunching as a major cause of irregular intervals on metro lines, and have tied irregularity largely to problems in other key scheduling and operational processes.

Recently research has demonstrated that simulation models of bus routes based on the classical theory of bus bunching have failed to replicate actual conditions of bus service intervals as captured in bus location tracking databases, even when random external events are incorporated into the model. One researcher attributed the classical theory's claim to the phenomenon of physics envy.

While station dwell time, as dependent on passenger volumes does influence interval variability, other explanations of bus service unreliability are the following:

• The lack of ability in resetting scheduled departure times at the start of the line. This is often the case because outer terminals in bus networks are often remote and on an isolated route rather than a convergence of routes, and it is uneconomical to position a supervisor at such a remote site for just one bus route. AVL/CAD systems have been used successfully in some surface transit systems to remotely revise terminal departure times, thus improving overall variability in service intervals from the start of the line.

• Schedules and service plans that provide very little recovery margin compared to actual running time performance will accumulate lateness. Service control actions may be taken to keep bus drivers within union-agreed contractual work parameters, in many bus systems through the use of unscheduled short-turning. Unscheduled short-turning often occurs in the post-peak period and results in many passengers off-loaded onto the following vehicle, which itself may also be crowded.

• Bus routes are subject to street closures, which may increase running times, leading to further lateness of drivers and greater levels of intervention necessary to keep drivers within work parameters.

• Bus operation is also dependent on the aggressiveness of driving. This is partly predictable driver-specific effect which has been quantified by researchers studying the Portland (Oregon) bus network.

See also

• Matthew effect
• Positive feedback