She switches off the car engine again. The car has not moved in what feels like forever, trapped in a dense, unmoving line that stretches far beyond what she can see. Around her, drivers step out, some leaning against their vehicles, others peering ahead in irritation. A biker removes his helmet and wipes his face. An auto driver mutters under his breath, the words lost in the thick heat and noise.
She gets out of the car and looks down the road, searching for a reason that is not there. There has to be something, an accident, a barricade, roadwork, anything. But the road ahead is clear. No collision. No obstruction. No explanation. Just vehicles, stalled for no reason anyone can see.
This is the moment commuters recognise but cannot explain, a traffic jam that seems to appear out of nowhere. A phantom traffic zone.
Traffic congestion is far more complex than the simple explanation of ‘too many vehicles’. It is a dynamic system in which small actions ripple outward, sometimes powerful enough to bring an entire road to a halt. At its core, it blends human behaviour, mathematics, physics and urban design.
Mir Samreen, a mathematician and AI research analyst specialising in high-dimensional data training and algorithmic optimisation, explains that traffic behaves like a flowing system where, at a larger scale, it moves like a liquid. At the same time, at the micro level, it is shaped by individual driver reactions, making congestion a chain reaction of human behaviour.
“Both, but in different contexts. Fluid dynamics (macro models) treats traffic like water in a pipe, using the Lighthill-Whitham-Richards (LWR) model to track density and flow. However, Follow-the-leader (micro models) uses ordinary differential equations to treat cars as individual particles. The bridge between the two is kinetic theory, which explains how individual interactions (micro) suddenly transition into a collective solid state (the jam),” she said.
Density, or how closely vehicles are packed, is a critical factor in traffic behaviour. When cars are tightly packed, the system becomes fragile and vulnerable to even minor disruptions. At lower densities, traffic can absorb disturbances, but as density rises, reaction time delays compound, causing the system to break and slow or stop abruptly.
This is described in physics as a shock wave, a disturbance that travels backwards through traffic even as vehicles move forward. Once formed, it can persist long after the trigger disappears.
While these principles explain how traffic jams form, a research paper published in the Journal of the American Society of Civil Engineers, titled “Traffic congestion assessment tool for urban roads based on traffic and geometric characteristics: A case of Hyderabad, India,” examines why they persist. Dr Naveed Marazi, a post-doctoral fellow at King Fahad University of Petroleum and Minerals, Saudi Arabia, and a co-author of the study, emphasised that congestion is not just about vehicle numbers.
“Long queues at intersections, sudden slowdowns, and unpredictable delays have become part of urban life. While the usual explanation points to the growing number of vehicles, a recent study offers a deeper insight: congestion is not just about volume, but about how traffic interacts with road design,” he said.
Roads with similar traffic volumes can perform very differently depending on design, with narrow, interruption-heavy stretches struggling even under moderate traffic while well-designed corridors handle volumes efficiently.
Solutions focus on managing density rather than volume. Maintaining safe spacing, better lane discipline and controlled merging help stabilise flow, while improved road design, reduced roadside friction and coordinated signals minimise disruptions and delays.
“The most counterintuitive mathematical finding is Braess’s Paradox: adding a new road to a congested network can actually increase total travel time because everyone tries to take the ‘best’ route. That overcrowds certain roads, and the whole system becomes less efficient. So the problem isn’t always lack of roads, it’s how people use them. Mathematically, it proves that “more infrastructure” isn’t a solution; better “flow control” is,” Samreen explained.
AI is now being used to address congestion in real time. “Modern AI uses Reinforcement Learning (RL) to ensure the newer systems adjust in real time. They look at live traffic, predict where congestion is building, and change signals or routes accordingly. Some systems such as Graph Neural Networks (GNNs), can even predict a jam before it fully forms and redirect cars early. So, instead of reacting to traffic, it tries to stay ahead of it,” Samreen said.
A few metres ahead, engines start again. The jam fades without reason. If we cannot see the cause, can we ever solve it?