Crumple zones

Why is a car built to destroy itself in a crash instead of holding firm?

Every instinct says a strong car is a safe car. Yet manufacturers spend real money making the front of a vehicle deliberately weak, so that in a serious impact the rails buckle in a designed pattern and a repairable car becomes scrap. When the Insurance Institute for Highway Safety staged a 1959 Chevrolet Bel Air against a 2009 Chevrolet Malibu in an offset frontal collision, the old car's body held up far better and its dummy recorded injuries that would have been fatal. The new car destroyed itself, and its dummy would probably have walked away.

So the question is not why cars crumple. It is why crumpling protects the person, when it plainly does not protect the car.

Start with what cannot be negotiated. A car moving at 50 km/h has a certain momentum and a certain kinetic energy, and when it hits a wall both must go somewhere. Momentum must be removed by a force acting over time; energy must be absorbed by something deforming. No design choice makes those quantities smaller. The car will stop.

So if the totals are fixed, the only variable left is how the stopping is spread out. That is the whole of it, and it is worth stating precisely. The impulse — force multiplied by the time it acts — must equal the momentum removed. Stretch the time and the force falls in proportion. Equivalently, in energy terms: the work done stopping the car is force times distance, so lengthen the crush distance and the force falls.

Put numbers on it, because the numbers are startling. At 50 km/h, roughly 13.9 metres per second, a body brought to rest over half a metre of crush decelerates at about 190 metres per second squared — some twenty times gravity — and the whole event lasts about seventy milliseconds. Compress that same stop into five centimetres, which is about what a genuinely rigid structure gives you, and the deceleration is ten times worse: nearly two hundred times gravity, over seven milliseconds. The speed was identical. The only difference was the distance available.

Half a metre of crush is not free. The only place it can come from is the length of car ahead of the occupant. Hence the engineering: the front rails fold in a controlled accordion rather than buckling sideways, the engine is mounted to drop under the cabin instead of into it, and the assembly is tuned to resist with a force high enough to absorb the energy in the space available — but no higher.

Which brings the second, less obvious half. Slowing the car gently is not the same as slowing the person. An unrestrained occupant continues at 50 km/h through a cabin that has already stopped, and meets the steering column over a few centimetres — the rigid crash we just calculated, with a windscreen instead of a bumper. Injury researchers call this the second collision, and it is the one that does the harm. There is a third: the organs continuing inside a chest that has stopped.

So the crumple zone buys crush distance, and the belt and airbag exist to make sure the occupant uses it. The belt engages early, ties the body to the decelerating structure, and lets it ride the whole seventy milliseconds down. Neither restraint is a cushion in the ordinary sense; both are couplings.

Does this predict anything checkable? It predicts the cabin must do the opposite of the nose. If the passenger cell also collapsed, the crush distance would be consumed by the space the occupant occupies, and the event would end early against an intruding dashboard. And indeed modern bodies are a stiff high-strength-steel safety cell with deliberately soft structures fore and aft. Soft ends, hard middle — and the Bel Air had it exactly backwards.

Three refinements matter.

The first: crumple zones do not "absorb the impact" in the sense of making the collision gentler overall. The energy is genuinely absorbed, by metal being permanently deformed, but the reason that helps is the distance the deformation takes place over, not the absorption itself. A structure that soaked up the same energy in five centimetres would kill you just as thoroughly.

The second: a car's stiffness is a compromise, not a maximum. Too soft and the crush is used up before the energy is gone, after which the cabin arrives at the wall anyway, so the ideal front end resists with a nearly constant force across its whole travel. Designs must also work across a range of speeds, angles and impact partners, which is why the structure is tuned rather than simply weakened.

The third is genuinely contested. In a collision between two vehicles, a stiff, heavy front end pushes the deceleration into the other car. Crash compatibility is an active argument in regulation, and the fleet-wide benefit of individual vehicles being made stronger is not settled.

• Why crash pulses are tuned differently for full-width, offset and small-overlap impacts.

Every term the collection defines is gathered in the glossary.