Crash simulator

Rebuild the collision from what you remember, watch it happen, and read the figures a claim file asks for — each one traceable to the source it stands on.

What this is for

The hardest part of a car insurance claim is usually not the damage. It is the gap between what a driver remembers and what a claim file needs. A driver remembers a noise, a direction and an impression of speed. An adjuster works with closing speeds, impact angles and change in velocity. This tool sits in that gap: you describe the collision in the terms you actually have — who was going which way, roughly how fast, what the road was like, when you saw the other car — and it produces the terms the file uses, along with a replay you can scrub through and a diagram you can attach.

It is not an estimating tool and it will not tell you what your claim is worth. Nothing here is converted into money, and the damage shading is deliberately a shading rather than a figure. What it produces is a physically consistent account of the collision, stated precisely enough to be argued with, which is more useful than a confident number nobody can check.

Why change in velocity, and not speed

Ask anyone how fast a crash was and they will tell you the speedometer reading. But a car that hits a parked lorry at fifty is in a very different collision from one that is rear-ended by a car travelling fifty faster than it, and the speedometer says the same thing in both cases. What separates them is ΔV: how much the vehicle’s own velocity changed in the fraction of a second the structures were in contact. That is the figure that tracks occupant outcome, and it is the figure insurers and biomechanics researchers work in.

The simulator computes ΔV directly from the impulse it resolves, splits it into the component along the vehicle and the component across it, and reports the principal direction of force as a clock position — twelve for a blow arriving on the nose, three from the right, nine from the left. From that it estimates a peak deceleration, and it labels that estimate as one, because it depends on an assumed crash-pulse duration taken from barrier tests rather than from your collision.

What it refuses to do

Two refusals shape the whole tool. The first is that it never apportions fault. It will happily record that one vehicle entered on red while the other entered on green, that a driver had been braking for a fifth of a second when contact occurred, or that a marked crossing was occupied. It will not turn that into a percentage, because the rules that turn facts into liability are written jurisdiction by jurisdiction, and a simulator has no standing to apply them. Where you are matters, and the jurisdiction pages are where that lives.

The second is that it will not print a number it cannot stand behind. Every constant the engine uses — the grip of the road surface, the time a driver takes to react, the springiness of a bumper, the mass and size of each vehicle — carries the document it was read from and the page it was read on. Where no such document could be opened, the constant still drives the replay, because the cars have to move somehow, but every figure downstream of it is marked as withheld, with the reason, and it is kept out of the written account and off the diagram. That is why the preset scenarios use particular reference vehicles rather than a generic saloon: those are the ones whose kerb mass and dimensions a manufacturer or a crash-test programme actually publishes.

Pedestrians and cyclists

A collision with a person is modelled differently on purpose. There is no impulse exchange, no ragdoll, and no figure invented from a mass and a velocity. What the tool records is what the literature relates to outcome: the vehicle’s speed at the moment of contact, whether braking force had begun and for how long, the time-to-collision in the seconds beforehand, and the sight line the driver had. It reports the risk associated with that impact speed from published curves, labelled as a population statistic, and it says which study each figure comes from — because the two most-cited pedestrian studies measure different outcomes, in different units, on different populations, and treating them as interchangeable is how a real difference gets averaged away.

For cyclists it reports the impact speed and stops there. No risk curve against impact speed could be found that was open enough to read and check, so none is shown. An absence stated plainly is worth more than a number nobody can trace.

Reading the replay

The simulation runs at a fixed step of one hundred and twentieth of a second and contains no randomness, so it produces the same result every time on every device. That is what makes the controls meaningful: the scrub bar moves through a recording rather than re-rolling a dice, the single-frame steps land on the same frames for everyone, and a shared link reopens exactly the collision the sender was looking at. You can watch from above, orbit freely, or ride in either driver’s seat, which is often the quickest way to see why someone did not see what they did not see.

When you have what you need, the written account can be copied in English or Spanish and rephrased for whoever will read it — a driver, a fleet manager, an adjuster, a broker or a lawyer each get a different framing of the same facts. The top-down diagram downloads as a PNG in the shape a report expects: north arrow, scale bar, each vehicle at first contact and again at rest. The timeline exports as a spreadsheet if someone wants to check the arithmetic themselves, which they should be able to.

What to do after a car accident · How fault is determined · Rules where you are