Lesson 4.2Lesson 4.2 · Simulation & Analytics
Mobility & Traffic
Movement is the question cities ask their twins most often - will this road relieve the jam, will this metro line carry the load, will this pedestrian street actually work - and the twin can test every one of them before a single barrier is moved, if you respect what the model can and cannot know
Before you close a street, widen a road, or commit a billion to a metro line, you would love to know whether it will actually work. A mobility twin lets you try it a hundred times in the model - as long as you remember the model is not the street.
Of all the questions a city puts to its twin, the ones about movement come first and loudest. Traffic is the daily pain everyone feels; transport is the most expensive thing a city builds; and mobility decisions - a road widening, a new junction, a metro alignment, a pedestrianised high street - are exactly the kind of costly, hard-to-reverse commitments where testing before building pays for itself many times over. Little wonder that traffic and transport simulation is the oldest, most developed and most heavily used branch of urban modelling, and usually the first thing a city twin is asked to do.
The appeal is obvious: instead of arguing about whether a new road will relieve congestion or simply fill up again, you model it, run it, and look. Instead of guessing whether a pedestrianised street will thrive or die, you simulate the footfall and the displaced traffic first. Instead of betting a metro line's business case on a single ridership guess, you test it across scenarios. This lesson is about how those models work - the movement of vehicles, people and public transport through a twin - what they can genuinely tell a designer, and the assumptions that, if forgotten, turn a powerful planning tool into a confident way to be wrong. Because nowhere is the gap between a clean model and a messy street wider than in how real people actually move.
Test the street before you close it. But the model is a behavioural theory - ask what people it assumes, and whether the new road just fills up.
How a city models movement
Transport modelling has a long pedigree, and most city mobility work still rests on one of two broad approaches, increasingly combined. The classic is the four-step model: estimate how many trips each area generates (trip generation), where they go (distribution), by what mode - car, bus, metro, walk (mode choice), and along which routes (assignment). Feed in land use, population and the network, and the model produces flows on every link and ridership on every line. It is a macroscopic, aggregate picture - rivers of demand over a whole city - and it remains the backbone of strategic transport planning because it scales to a region and connects directly to land-use decisions.
The newer approach is microsimulation and agent-based modelling, which drops down to the individual: each vehicle or traveller is an agent, accelerating, braking, changing lanes, choosing a route and a departure time, interacting with every other agent. This is what produces the vivid animations of a junction filling and clearing, and it captures things the aggregate model cannot - queue spillback, signal timing, merging behaviour, the fine-grained mechanics of why a specific interchange fails. The price is data hunger and computation: a microsimulation needs detailed network geometry, signal plans and calibrated driver behaviour, and it covers a junction or corridor, not easily a whole region.
A serious mobility twin uses both, plus layers for public transport (timetables, capacity, dwell times, transfers, crowding) and for pedestrians (walking speeds, crowd density, desire lines, the comfort and safety of being on foot). Increasingly these are fed by live data - loop detectors, GPS traces from phones and fleets, ticketing taps, camera counts - so the twin can be calibrated against how the city actually moves today and then run forward to test a change. The essential thing for a designer to grasp is not the mathematics but the hierarchy: strategic models answer 'should we build a metro here?', microsimulations answer 'will this junction work?', and using the wrong grain for the question produces an answer that looks authoritative and means little.
Four-step = rivers of demand over a whole city. Microsimulation = one junction, car by car. Wrong grain = confident wrong answer.
Testing a road, a metro or a pedestrianisation before building
The payoff of a mobility twin is the ability to run a before-and-after test on an intervention that does not yet exist. You calibrate the model against today's measured flows until it convincingly reproduces the current city, then you change one thing - add the road, reroute the buses, close the street to cars, insert the metro line - and run it again. The difference between the two runs is the model's estimate of the intervention's effect: the congestion relieved or created, the ridership gained, the footfall shifted, the journey times changed, the emissions moved.
This is genuinely powerful and genuinely used. A pedestrianisation can be tested for where its displaced traffic will go and whether neighbouring streets can absorb it, before the bollards go in. A new junction layout can be microsimulated through a peak hour to see if it locks up. A metro line's alignment and station placement can be tested against modelled ridership to shape the business case. A bus-priority lane can be trialled in the model across a dozen configurations to find the one that helps buses without strangling everything else. The twin lets a city fail cheaply and often in the model so it can succeed expensively and once in concrete.
But a before-and-after result is a difference between two model runs, and it inherits every assumption in the model twice over. The single most important assumption in transport modelling is also the most frequently forgotten: induced demand. Building or widening a road does not just serve existing traffic - it makes driving more attractive, so more people drive, trips lengthen, and the new capacity fills up, sometimes leaving congestion no better than before. A model that treats demand as fixed - as many crude or politically convenient models do - will systematically overstate how much a new road helps, because it never lets the extra traffic the road itself creates appear. The same logic runs in reverse for road removals, which often shed traffic rather than gridlocking as feared. A designer reading a mobility result must always ask whether demand was held fixed or allowed to respond, because that single choice can flip the conclusion.
Where the model and the street part company
Every transport model is a behavioural theory in disguise: it assumes people move according to certain rules - minimising time or cost, responding predictably to prices and signals, queuing and merging in orderly ways. Those assumptions are reasonable approximations in some cities and badly wrong in others, and nowhere is the gap more consequential than in the dense, mixed, improvisational traffic of many Indian streets. A model built on lane discipline, car dominance and orderly flow can seriously misrepresent a road where two-wheelers filter through gaps, autos and cycle-rickshaws and hand-carts and pedestrians share the carriageway, and right of way is negotiated moment to moment rather than signalled. Run such a model unadjusted and it produces a precise answer to the wrong city.
The data behind the model carries its own distortions. Movement data is easiest to collect about the formal, motorised, smartphone-carrying, fare-paying city - and thinnest about exactly the people a humane mobility policy should centre: pedestrians, cyclists, informal transport users, the poor who walk because they must. If walking and cycling trips are under-counted, the model will under-value the infrastructure that serves them and over-build for cars, entrenching inequity under a veneer of objectivity. A mobility twin optimised for vehicle flow can quietly become a machine for designing cities around cars and against people on foot, which is the opposite of what most cities now say they want.
And transport models struggle with the genuinely novel and the long term. They are calibrated on how people move now; they are weakest at predicting responses to things that have never existed - a first metro in a city of two-wheeler commuters, a congestion charge, a pandemic that rewrites commuting overnight, the slow arrival of electric and shared and autonomous vehicles. The further ahead and the more transformative the change, the softer the forecast, and the more it should be read as a comparison between scenarios rather than a prediction of ridership to the nearest thousand. The discipline, again: trust the relative result more than the absolute, demand to know the behavioural assumptions, and keep the binding traffic-engineering and safety decisions - capacities, signal timings, junction geometry to code - with the qualified transport engineers and the governing standards, never with a designer's read of the animation.
Easy to count: cars, cards, smartphones. Hard to count: the pedestrian, the cyclist, the informal rider. The model sees what it's fed.
Reading a mobility result like a professional
Put the pieces together and a short discipline emerges for anyone handed a traffic or transport simulation - which, as a designer, you will be, often as the basis for a decision that affects your project. First, match grain to question. A strategic four-step result cannot tell you whether a specific junction will lock up; a microsimulation of one junction cannot tell you whether the metro is worth building. Check that the model's resolution fits the claim being made on it.
Second, find the demand assumption. Was travel demand held fixed, or allowed to respond to the intervention (induced and suppressed demand)? This one question, more than any other, determines whether a road-building or road-removal result is credible. A fixed-demand model showing a new road solving congestion should be treated with deep suspicion.
Third, ask what the model cannot see. Which modes and which people are well represented in the data, and which - pedestrians, cyclists, informal transport, the poor - are thin or missing? A result that is silent about walking and cycling in a city where most short trips are on foot is not a complete picture of mobility, whatever its apparent precision. Fourth, trust comparisons over absolutes. The model is far more reliable telling you that option A relieves more congestion than option B than telling you either one's exact future flow; use it to choose between scenarios, not to promise a number. And fifth, keep the binding decisions where they belong: the twin and its mobility simulation inform the design and the debate, but statutory transport approvals, safety-critical junction and capacity engineering, and the final political choice between competing mobility futures rest with the accountable engineers, authorities and elected decision-makers - not with the model, however vivid its animation. A designer who reads mobility results this way turns the twin from a source of false certainty into a genuine instrument of better, fairer movement.
Induced / suppressed demand
Whether a road or capacity result is credible at all
The decisive question: was demand held fixed or allowed to respond to the intervention? A fixed-demand model overstates road benefits and understates the effect of removing road space. Ask before believing.
Model grain (strategic four-step vs microsimulation)
Matching the model's resolution to the claim made on it
Strategic models answer region-scale 'should we build this?'; microsimulations answer 'will this junction work?'. Using the wrong grain gives an authoritative-looking answer to the wrong question. Module 4.1.
Mode & equity coverage
Whether pedestrians, cyclists and informal transport are represented
Movement data is richest about cars, cards and smartphones, thinnest about walking, cycling and informal modes. A twin blind to them can design cities against the people who most need other options. Module 8.3.
Binding transport engineering & approvals
Capacities, junction safety, signal design, statutory sign-off
Safety-critical and statutory transport decisions stay with qualified transport engineers and the governing standards; the twin informs the design and the debate, it does not certify the road. Module 6.4.
Workshop - stress-test a mobility claim for its hidden assumptions
The skill here is taking a confident mobility proposal - a road widening, a new junction, a pedestrianisation, a metro case - and exposing the modelling assumptions that decide whether to believe it. You will do this on a real or proposed scheme.
Just a transport scheme you can read about and a notebook. No modelling software - this is about reading a mobility claim critically, not building a traffic model.
Goal: judge a mobility proposal by interrogating the model behind it Inputs: a real or proposed transport intervention you can read about (a road, junction, metro line, bus lane or pedestrianisation in any city) + this lesson + a notebook Time: ~45 minutes
- 1Pick a scheme: choose a transport intervention with a modelled justification you can read about. Write down the headline claim (for example 'cuts congestion by X' or 'carries Y passengers').
- 2Identify the grain: is the supporting model strategic (four-step, region-scale) or a microsimulation of a specific junction or corridor? Does its resolution actually fit the claim being made?
- 3Hunt the demand assumption: was travel demand held fixed, or allowed to respond to the change (induced / suppressed demand)? If the source does not say, note that the result's credibility is unestablished.
- 4Check who is counted: which modes and users does the model represent well - and are pedestrians, cyclists and informal transport users, who may be most of the trips, present or missing?
- 5Write a verdict: is the claim credible, over-stated, or built on a fixed-demand artefact? State the one assumption that matters most and the binding engineering you would still leave to the transport engineers - framed as critical reading, not a re-modelling.
You’ll walk away with
A one-page critical read of a mobility proposal: its claim, the model grain, the demand assumption, the modes and people counted and missed, and a verdict on how much to trust it. Keep it as a checklist for the next traffic result you are handed.
Three altitudes on the same idea
Read the band that fits you — or all three.
A mobility twin decides much of the context your project lives in - the access, the footfall, the congestion, the street life - and increasingly it is where those effects are argued out. When your development generates trips, a transport model will estimate its impact on the surrounding network, and that estimate can shape approvals, required contributions and design conditions. Learn to read whether the model is strategic or a microsimulation, whether it allowed demand to respond (induced demand is the assumption most often quietly omitted), and whether it represents the pedestrians and cyclists your project should serve - not only the cars. Use mobility simulation to argue for walkable, transit-connected, humane access and to catch access problems early. But defer binding traffic-capacity engineering, junction safety and signal design to qualified transport engineers and the governing standards, and never let a smooth animation substitute for their accountable analysis.
Mobility simulation mostly lives outside your envelope, but it reaches inside it at the threshold - how people arrive, enter, queue and flow. The same modelling logic that simulates a street scales down to pedestrian flow through a lobby, a concourse, a retail floor or an evacuation route, and a building-scale flow model nests inside the city twin that supplies its arriving crowds. Understand that an interior circulation or evacuation simulation rests on assumed walking speeds and behaviours that may not match real, diverse users - children, the elderly, people with disabilities, people in a hurry or a panic - and that a result calibrated for one crowd can mislead for another. Treat such simulation as a way to reason about comfort, accessibility and safety of movement, not a guarantee; keep binding life-safety and egress engineering with the fire and safety professionals and the building code.
Learn the one question that cuts through most traffic-model claims: was demand held fixed, or allowed to respond? A huge share of bad transport decisions - roads built to cure congestion that only moved or worsened it - trace to models that treated the number of trips as fixed and so never showed the new traffic a new road induces. Beyond that, learn the grain distinction (strategic four-step versus junction microsimulation), and learn to ask who the model can and cannot see - because movement data is richest about cars, cards and smartphones and thinnest about the pedestrians, cyclists and informal riders who often make up most of an Indian city's trips. You are not expected to build a transport model; you are expected to read its results critically, know that it is a behavioural theory in disguise, and ask whose mobility it was built to serve.
“The traffic model proves that widening the road from four lanes to six will cut congestion and journey times. It is standard transport modelling, so we can be confident the wider road will fix the jams - more lanes, more capacity, less traffic, simple.”
Do it yourself
No tools needed - reason it through.
- 1What is the difference between a strategic four-step model and a microsimulation, and which question does each answer?
- 2Explain induced demand and why a fixed-demand traffic model overstates the benefit of building a new road.
- 3How does a before-and-after test in a mobility twin work, and why does it inherit the model's assumptions twice?
- 4Why is movement data typically rich about cars and thin about pedestrians, cyclists and informal transport - and why does that matter for equity?
- 5Why should a designer trust a mobility model's scenario comparison more than its absolute flow or ridership numbers?
The one line to carry out
Peer-reviewed journals & authoritative standards
- 01Traffic simulation — Wikipedia - Traffic simulation, 2026.
- 02Transport modelling — Wikipedia - Transport modelling, 2026.
- 03Agent-based model — Wikipedia - Agent-based model, 2026.
- 04Urban planning — Wikipedia - Urban planning, 2026.
Movement is one great family of urban simulation; the other is the environment the city sits in and shapes - its energy, its heat, its wind, its air and its water. Next we turn the twin on energy, environment and climate, where the stakes and the need for engineering humility are higher still.
The author
Amogh N P
Architect, interior designer, and creative polymath. Studio Matrx began in his notebooks — his vision of design made honest, useful, and open to everyone. Its Academy is written and taught in his memory, and free, forever.
More about Amogh →