Incoprea Theory
Volume IV · 2026

Theory 20/Urban Systems & Cities /Traffic theory/3 min

Congestion Onset Is Variability, Not Volume

Traffic models add lanes. The queue is often made by the peaks, not the average.

The claim

The onset of urban congestion is set by the variability of demand rather than its mean, so two roads carrying the same average flow can be free-flowing or gridlocked depending on the variance, and smoothing demand is worth more than adding capacity.

01The evidence

  • Overflow queue models for signalised urban intersections treat congestion as a queueing problem in which variability of arrivals, not just mean flow, determines whether queues clear between cycles.Transportation Research Board
  • Work on congestion in networks has described the onset as a phase transition driven by load relative to capacity, which is the structure a variability-controlled threshold requires.Berkeley, congestion phase transitions

02The tension

Capacity planning is based on average daily traffic, with a peak-hour factor applied as a correction. The correction is a fixed multiplier, which assumes variability is a constant fraction of the mean. It is not, and the residual shows up as unpredicted gridlock on roads that are nominally under capacity.

03The theory

A road is a queueing system. Whether a queue clears depends on whether arrivals in a cycle exceed the capacity to discharge them. A mean flow below capacity can still produce a growing queue if the variance is high enough, because the exceedances are what accumulate.

That makes the onset a function of the ratio of variance to spare capacity rather than of the mean. Two roads with the same average can be stable or unstable depending on the variance, and the transition is sharp because queues grow without bound once arrivals persistently exceed discharge.

The prediction is that congestion onset should be predicted by demand variability better than by mean volume-to-capacity ratio, and that smoothing demand, through pricing or scheduling, should delay onset more than adding capacity of the same cost.

This changes the economics of congestion policy: capacity is a high-variance fix, while demand smoothing targets the variable that actually controls the transition.

04What this is not

Volume-to-capacity ratio planning

The ratio compares mean flow to capacity and treats variability as a peak factor. This makes variability the controlling variable and predicts that the same ratio can be stable or unstable.

05Predictions

  • P1Congestion onset should be predicted by demand variability better than by mean volume-to-capacity ratio.
  • P2Roads matched on mean flow but differing in variance should differ in breakdown probability.
  • P3Demand smoothing should delay onset more than equivalent-cost capacity additions.

06What would refute it

Onset tracking the mean ratio regardless of variance would restore the capacity account.

07What it would change

It would shift congestion policy from building to smoothing, and change how road schemes are appraised.

08Evidence base

  1. Modeling overflow queues on urban signalized intersections — TRBhttps://trid.trb.org/view/843845
  2. Congestion in networks: phase transitions — Berkeleyhttps://www.stat.berkeley.edu/~aldous/Research/OP/congestion.html