A collision lasts a fraction of a second and leaves a long trail: tyre marks and rest positions on a scene plan, deformation in the repair photographs, dashcam and CCTV footage, and increasingly the data recorded by the vehicles themselves. Collision analysis is the discipline of reading that trail as an engineer, testing the accounts against it, and saying honestly how much it can and cannot prove.

The questions the analysis answers

  • Consistency. Whether each party’s account survives contact with the physical evidence: whether the damage to each vehicle matches the other in height, overlap, direction and severity, and whether the described speeds and movements could have produced it.
  • Speed and severity. What the deformation, the masses, the energy absorbed and any recorded data say about closing speed and the change in velocity each vehicle underwent. Each method carries its own tolerance, and the report states it.
  • Video and recorded data. What dashcam, CCTV and vehicle data fix about the collision, each with its tolerance, and whether the accounts are consistent with it.
  • The other expert’s work. Whether a reconstruction already in the case rests on sourced inputs, whether its calculations reproduce, and whether its conclusions are within the tolerance of its evidence.

How the work is done

Every analysis is built from the papers: the police collision report and scene plan where there is one, photographs, repair estimates, witness statements, any video, and any download from the vehicles. The first product is a schedule of what has been supplied, what is missing, and what it would be worth obtaining. Event data and dashcam footage are overwritten quickly; an early letter asking for them is often the most valuable step in the case.

Analysis then proceeds from the most reliable evidence to the least. Physical marks, deformation and recorded data come first, witness estimates of speed and distance last. Where a calculation depends on a coefficient, the source is cited and a realistic range is carried through to the answer, so the opinion is stated as a bracket rather than a false single figure. A result that fails its own momentum or energy audit is reported as a failure, not softened.

Where I stop

Vehicle and mechanical engineering, and nothing else. I do not survey scenes, map rest positions or reconstruct pre-impact speeds and paths from scene evidence, and I give no opinion on driver perception or avoidance. Momentum and energy analysis is confined to testing a reconstruction already in the case against the physical evidence. Delta-V is an engineering quantity reported for the medical expert; I give no opinion on injury causation. A question outside the boundary is declined or referred on, in writing.

The methods page shows this discipline working on published data.

Review of collision reconstruction evidence

Many instructions arrive with a reconstruction already in the bundle, from a police collision investigator or the other side’s expert. A structured review asks four things: are the inputs sourced, do the calculations reproduce, is the uncertainty carried through, and does the conclusion stay within what the evidence supports. The output is a note an advocate can use directly, and, where the review supports it, written questions under rule 35.6 or points for a joint statement.

Civil and criminal

The same engineering serves both. In civil claims the report is written under CPR Part 35 and its Practice Direction; in criminal proceedings, most often dangerous or careless driving and causing death or serious injury by driving, under CrimPR Part 19, with the additional disclosure of anything that might undermine the opinion that the criminal rules require. Single joint expert appointments are welcome. Each report ends with the declaration the relevant rules prescribe, and the opinion is the same whoever is paying.