Impact speed can be approached two ways that share almost nothing. Momentum analysis works from where the vehicles went after contact. Crush analysis works from the damage they carry. Because the input sets barely overlap, agreement between them is genuine corroboration rather than one method flattering the other. The corollary is less often stated: when they disagree, the disagreement is itself a finding. It usually points at an assumption that does not fit the collision, and running only one method conceals that signal.
Two methods, two input sets
Momentum analysis conserves the system's total momentum across the impact. Its inputs are vehicle masses including occupants and cargo, the directions of travel entering and leaving contact, and the post-impact paths and distances that establish separation velocities. Crush analysis works from energy: the depth and distribution of permanent deformation, applied to the struck structure's stiffness properties, yields the energy absorbed and the closing speed it implies.
The only meaningful overlap is mass. Everything else — geometry and rest positions on one side, deformation and stiffness on the other — is independent.
What momentum analysis actually needs
Impact and departure headings taken from scene geometry rather than assumed; a defensible point of impact, usually located from gouges, fluid deposits and debris distribution; and drag factors for the surfaces each vehicle crossed on its way to rest, which depend on the surface, whether wheels were rolling, locked or damaged, and whether the vehicle rotated as it slid.
The angles are the sensitive part. Small heading errors propagate strongly, which is why total-station or laser-scanned scene documentation is worth more than a scaled sketch.
What crush analysis actually needs
A measured crush profile — a set of depths taken along the damaged face at recorded intervals from a defined reference, on both vehicles where possible — and stiffness coefficients appropriate to that vehicle and that face. SAE J224 provides the collision deformation classification used to describe the damage consistently, with SAE J2433 serving the heavy-vehicle equivalent, so that a profile means the same thing to a second engineer.
The measurement discipline matters as much as the arithmetic. Crush measured from photographs, or from a vehicle already partially disassembled or moved by recovery equipment, carries an error band that should be stated rather than absorbed.
Where stiffness coefficients come from
Stiffness values derive largely from instrumented barrier crash testing, much of it conducted or published under NHTSA's compliance and consumer-information programmes. That origin defines their domain: frontal impacts into a flat rigid barrier, at test speeds and directions chosen for a regulatory purpose.
Real collisions frequently sit outside that domain — narrow-object and pole impacts, underride, override, oblique and corner engagements, and impacts into structures that deform themselves. Applying a frontal coefficient set to a side or rear structure, or to a vehicle represented only by a class average, is a modelling choice that belongs on the page.
The energy that never reached the crush
Crush measures the energy that stayed in permanent deformation. It does not capture the elastic energy returned as the structures rebound, which restitution accounts for, and it does not capture energy dissipated in tyre scrub, in ground engagement, in rotation, or in bodies the vehicle struck that absorbed energy of their own.
That is one structural reason a crush estimate can sit low relative to momentum, and why the coefficient of restitution assumed at low closing speeds is rarely negligible.
Sensitivity is the real deliverable
A single number states less than the range it sits in. Competent work runs each input across its plausible interval — mass, drag factor, departure angle, crush depth, stiffness set — and reports which ones actually move the answer and by how much.
The result is usually that two or three inputs dominate and the rest are noise. Knowing which is which tells everyone where the remaining investigative effort should go, and where argument is not worth having.
Reading the disagreement
Divergence between the two methods has recognisable causes. A crush figure well below the momentum figure may indicate an unrecorded secondary impact, an override or underride geometry in which one structure passed through rather than into the other, energy absorbed by roadside furniture, or an inapplicable stiffness set.
A momentum figure that looks low may indicate a misplaced impact point, a drag factor set for a surface the vehicle did not actually cross, or a rotation that was not accounted for. In each case the reconciliation, not the average, is the analysis.
Scene evidence as the third constraint
Tyre marks constrain the answer independently again. Skid marks bound speed through a drag factor; critical-speed yaw marks bound it through the arc radius the vehicle carved; scuff and gouge marks locate the sequence in space. Rest positions and debris distribution bound the departure conditions momentum analysis depends on.
Dash-camera, surveillance or bystander video adds timing, and photogrammetry can recover geometry from scene photographs after the marks themselves have weathered away — which they often do within days.
What a defensible result looks like
A stated range rather than a point value; each method presented with its own inputs and uncertainty; the sensitivity of the conclusion to the inputs that dominate; and, where the methods disagree, an explanation of the disagreement rather than a quiet preference for the more convenient number.
That is also what makes an opposing analysis testable. Two reconstructions that disclose their inputs can be compared; two that report only conclusions can only be asserted at each other.
This article is general technical orientation, not a failure analysis, an engineering opinion, or advice on any specific matter. Determining the cause of a particular incident requires hands-on examination by a credentialed expert.