The elastic chain of cyclists

When a group of cyclists rides through a crossing, how close do the riders need to stay so that a car waits for the whole group instead of cutting through it?

Gap needed

9.0 m

Group stays whole

5%

How it splits

2 chains

The riders

10
20 km/h
6.5 m
or drag a rider along the chain

The car

20 km/h
2.0 m

The car waits at the crossing.

How the model works

Where the idea comes from

The model and the elastic analogy are vince.quant's. The idea came on a bike ride: when riders stay close to each other, drivers treat the group as one object and wait for it to pass. As the gaps grow, it feels as if the riders were tied together by elastic threads that stretch, turn red in the middle and finally snap. At that point the group is no longer one chain but two, sometimes with a free rider left in between, and a car can slip through. Everything below turns that picture into numbers.

1. The gap a car physically needs

To pass between two riders, a car needs a hole that stays open for the whole time it occupies their lane, while the rider behind keeps coming. Measured from the back wheel of one rider to the front wheel of the next, the minimum gap is therefore:

minimum gap = car width + 2 × safety margin + rider speed × time the car needs to clear the lane

For a car already rolling at speed u, the time to clear the lane is the distance it has to cover (its own length plus the width of the lane, about 6 m) divided by u. With riders at speed v, a car of width W and length L, a lane of width w and a margin m on each side:

minimum gap = W + 2m + (v / u) × (L + w)

With riders at 20 km/h and a car 2 m wide, keeping 0.5 m on each side:

The carTime to clear the laneMinimum gap
starts from a stopabout 3.0 s19.9 m
rolls at 10 km/h2.2 s15.0 m
rolls at 20 km/h1.1 s9.0 m
rolls at 40 km/h0.5 s6.0 m

The same group can therefore be unbreakable for a car waiting at a stop sign and breakable for a car that is already moving. A thread has no strength of its own: its strength depends on the car looking at it.

2. The elastic

The thread between two riders is measured by its stretch, the actual gap divided by the minimum gap. Drivers do not compute anything, they judge, so instead of a hard limit the chance that a driver goes for a gap follows an S-curve centred on that driver's own threshold:

chance that the car cuts = 1 / (1 + (threshold / stretch)6)

A cautious driver wants 40% more room than the minimum (threshold 1.4), a normal driver 15% more (1.15), and a driver in a hurry accepts slightly less than the minimum (0.9), which forces the next rider to brake. This chance is the colour of the thread: green when almost no driver would try, red in the middle when most would. In the simulation a thread breaks once the chance passes 50%.

3. From threads to chains

Riders joined by unbroken threads form a chain. When a thread breaks, the group splits into several chains, and a rider whose two threads are broken is a free rider. A car waiting at the crossing sees the gaps one after the other and needs only one of them, so the chance that it lets the whole group through is the product of (1 − chance) over every thread. The chain is only as strong as its weakest link, and a long group needs tighter links than a short one: with a 5% risk on each thread, a group of 10 riders holds together 63% of the time, a group of 20 only 38%.

Drivers also grow impatient. Traffic studies have observed that the gap a driver accepts shrinks as the wait goes on, which makes the back of a long group its weakest part. The impatience box in the simulation shows the effect: identical gaps turn redder towards the tail.

What the model says

Where it connects

Traffic engineers have studied the driver's side of this since the 1950s under the name gap acceptance: a driver waiting to cross a stream accepts or rejects each gap in turn, and the critical gap is the one accepted half of the time. The threads here are gap acceptance seen from the riders' side.

Professional cycling already defines a group as a chain with a time threshold. Riders who finish together are given the same time, and a gap of one second or more between the back wheel of one rider and the front wheel of the next starts a new group, with three seconds in some bunch sprints.

Assumptions

The numbers use a car 4.5 m long, a bike lane 1.5 m wide, riders 1.8 m long, a safety margin of 0.5 m on each side, and a car that leaves a stop after 0.4 s and accelerates at 2 m/s². The driver thresholds are assumptions; measuring them means filming a crossing and recording which gaps drivers take and which they let pass.

The next layer of the model is the interaction between the riders themselves. A rider who sees a gap opening speeds up to close it, which is a real restoring force, the spring behind the elastic.