Crossing the Accident Reconstructionist: Velocity and Delta-V Calculations

When an accident reconstructionist testifies to a collision severity, the figure that carries the injury-causation argument is delta-V, the change in a vehicle's velocity across the impact phase. Delta-V correlates with occupant loading because it is the velocity change the occupant compartment, and everything belted to it, must undergo. It is the number the biomechanist then maps to an injury threshold.

The cross-examination leverage starts here: delta-V is almost never measured. It is the output of a physics model fed with input values. The only common exception is an event data recorder (EDR), the crash-file captured by the airbag control module and read with a Bosch CDR tool, which may record a longitudinal delta-V directly. Where an EDR exists, any gap between the recorded delta-V and the expert's computed delta-V is a first-order wedge. Where it does not, every digit of the delta-V traces back to inputs.

The entire audit reduces to one discipline: force the witness to sort every input into measured (rest positions, skid and yaw marks, crush depths, EDR data, scene survey) versus assumed (stiffness coefficients, drag factor, restitution, principal direction of force, occupant and cargo mass). The assumed inputs are where the opinion is soft, and they are almost always the inputs that move the answer the most.

The momentum method applies conservation of linear momentum through the impact. In vector form, the summed momentum of both vehicles before impact equals the summed momentum after: mass one times velocity one plus mass two times velocity two, before, equals the same sum after. The reconstructionist recovers the post-impact velocities from skid and yaw analysis and rest positions, fixes the approach and departure angles from the scene, and solves for the pre-impact speeds. Delta-V is the difference between each vehicle's pre- and post-impact velocity.

The method's strength is that it is independent of vehicle stiffness. It never touches a crush coefficient. Its weakness is geometric conditioning. Near collinear or shallow-angle impacts, the solution becomes ill-conditioned: a change of a few degrees in an assumed departure angle can swing the computed closing speed substantially, because the trigonometry divides by small quantities. Rotation and angular momentum are frequently simplified away, which loads error into planar-only solutions.

• Demand the source and precision of every approach and departure angle: total-station survey, drone photogrammetry, or eyeballed from photographs.
• Ask whether departure headings were mapped from physical tire evidence or inferred from rest position and assumed trajectory.
• Have your expert re-run the solution across a plausible angle band, plus or minus a few degrees, and put the output range next to the single number offered.

The crush-energy method estimates the energy absorbed as permanent deformation, then converts it to speed. It rests on the empirical model that the crush force per unit width equals a stiffness intercept plus a slope times crush depth, written as force density equals A plus B times C, where C is the crush depth. Integrating force over crush distance gives energy density equal to A times C, plus one-half B times C squared, plus a term G, where G equals A squared over two B and represents the energy absorbed up to the threshold of visible damage.

The expert measures a crush profile at equally spaced stations across the damaged width, commonly six points labeled C1 through C6, and applies a trapezoidal integration across the damage width L. The absorbed energy converts to an equivalent barrier speed, then combines with the partner vehicle to yield delta-V. Because this method needs only the standing damage and the coefficients, it works when post-impact motion is unknown, which is exactly when momentum fails.

The cost of that convenience is that the answer now depends entirely on A and B, and those are not properties of the specific vehicle in front of you. They are category averages fit to NHTSA staged barrier tests. Selecting the category, the model-year span, and the structural type is a judgment call carrying real spread, and it is the least visible step in the whole calculation.

Stiffness coefficients are the softest data in the model, and they are usually stated with unearned confidence. They are regressed from a handful of full-frontal barrier crashes, grouped into broad vehicle categories, then applied to a subject vehicle that may not belong to the same structural class. Three failure modes recur.

• Category mismatch. Coefficients derived from passenger sedans applied to a light truck, a unibody figure applied to a body-on-frame vehicle, or a modern platform matched to a coefficient set from an earlier generation. Structure changed; the coefficient did not.
• Wrong impact region. Front-structure coefficients used to analyze a side or rear impact. Side and rear stiffness bear no fixed relationship to frontal stiffness, so borrowing frontal A and B for a side crush is a category error, not an approximation.
• Undocumented scaling. Coefficients quietly adjusted, sometimes called stiffness scaling, with no stated test basis. Ask for the arithmetic.

Because delta-V scales roughly with the square root of absorbed energy, a coefficient error does not pass through one-to-one, which experts sometimes offer as reassurance. Do not accept it. The B term multiplies crush depth squared, so coefficient error and measurement error compound inside the same quadratic. Demand the specific NHTSA test series or report numbers, the exact vehicle or vehicles used to derive the values, and confirmation that the impact region of those tests matches the impact region in your case.

Even correct coefficients produce a wrong delta-V if the crush measurements are wrong, and crush field measurement is error-prone. The energy calculation is sensitive to the geometry of the profile, not just its maximum depth, because the quadratic term rewards deep local crush.

• Reference profile. Crush is depth relative to the original undeformed body line. If the undamaged reference dimension came from a spec sheet or an assumption rather than the exemplar vehicle, that error offsets every station.
• Induced versus contact damage. Only direct-contact crush absorbs the modeled energy. Induced damage, buckling away from the contact zone, must be excluded. Including it inflates absorbed energy and delta-V.
• Damage width and station spacing. The integration depends on the measured width L and the number and spacing of stations. Sparse or unevenly spaced measurements degrade the integral.
• Estimated from photographs. Establish whether the expert physically measured the vehicle with a scale and a reference, or scaled a photograph. Photo-derived crush without a calibrated scale in frame is an assumption wearing a measurement's clothes.

Ask for the measurement sheet, scaled photographs, and the undamaged reference source. Then have your expert recompute energy with the defensible profile.

Restitution is the single most consequential assumption in minor-impact cases, and it is frequently buried. For a central impact, each vehicle's delta-V takes the form one plus e, times the mass ratio, times the closing speed, where e is the coefficient of restitution, the bounce-back fraction, running from zero for a fully plastic impact to one for a perfectly elastic one.

The mechanism matters because e enters as a multiplier. At low closing speeds, bumpers and energy absorbers behave elastically and e is materially above zero. As speed rises and structure crushes plastically, e collapses toward zero. Many reconstructions simply set e to zero, which understates low-speed delta-V, because the one-plus-e factor is dropped. An inflated e does the reverse and overstates it.

In minor-impact soft-tissue litigation, this one scalar can carry the delta-V across whichever injury-threshold line the biomechanist has drawn. The audit is direct: ask what value of e was used, what its source is, published bumper-test data for that specific vehicle pairing versus a bare assumption, and re-run the delta-V across the documented range. Note which direction the chosen value favors. When the restitution assumption consistently favors the retaining party, name that on the record.

The momentum method inherits every error in the friction inputs, because post-impact speeds come from skid analysis. The standard relation is that speed equals the square root of two, times the drag factor f, times gravitational acceleration, times skid distance. The drag factor f is where assumptions enter.

Reconstructions often use a table value for a surface type, a dry-asphalt range for example, rather than a measured value. That single choice is disturbed by ABS (which changes the mark left and the effective factor), partial rather than full braking, tire tread and inflation, road grade, superelevation, and surface contamination. Because speed varies as the square root of f, a twenty percent error in f is roughly a ten percent error in speed, which sounds tolerable until you note it enters the post-impact term for both vehicles and then propagates through the momentum solution.

• Demand the source of the drag factor: a site-specific drag-sled or decelerometer measurement, or a handbook range.
• Establish whether ABS or braking state was accounted for, and how the mark was interpreted.
• Require a sensitivity run across the plausible factor range for the actual surface, grade, and tire condition.

Mass sets the split. In any two-vehicle impact, each vehicle's delta-V scales with the other vehicle's share of the combined mass, so the lighter vehicle always experiences the larger delta-V. Small errors in the mass ratio move both numbers.

• Curb versus actual weight. Curb weight omits occupants, cargo, fuel load, and aftermarket additions. A loaded vehicle can differ materially from its curb figure, shifting the ratio and therefore both delta-Vs.
• Distribution and center of gravity. Front-to-rear weight split and CG location drive the principal direction of force, the post-impact rotation, and the yaw moment of inertia used in any spin analysis. A CG assumption that is off relocates the PDOF the crush method assumed.
• Structural region implied by loading. Where the mass sits also informs which structural stiffness is doing the absorbing.

Confirm exactly which weights were used, curb, gross vehicle weight rating, or an actual weighed figure, how occupants and cargo were accounted for, and what CG location was assumed and on what basis.

The two methods are independent: momentum ignores stiffness, crush ignores post-impact trajectory. A defensible reconstruction runs both and shows them converging, because agreement between two independent paths is strong corroboration. Running only the method that favors the retaining side, or presenting one method while the data for the other sat available, is a reliability flag worth surfacing.

Translate the input audit into the governing law rather than arguing physics in a vacuum. Under Federal Rule of Evidence 702, as restated in the 2023 amendment, the proponent must show by a preponderance that the opinion reflects a reliable application of reliable methods to the facts of the case, and 702(d) targets exactly that application step, which is where assumed inputs live. Daubert v. Merrell Dow supplies the reliability factors, and Kumho Tire v. Carmichael confirms they reach engineering and other technical experts, not only scientific ones. In Frye jurisdictions, general acceptance of the method does not immunize an unsupported application of it.

• Error rate. Was an uncertainty band or sensitivity analysis reported, or was a single point estimate presented as if precise? A point estimate with no stated range invites the known-error-rate factor.
• Testability. The model is testable. Was it tested against the EDR, and if the EDR contradicts it, why?
• Application to the facts. Inputs assumed rather than measured, coefficients borrowed from a different vehicle class or impact region, one method where two were available.

Calibrate the ask to the defect. Input-level softness usually goes to weight before a jury. Method-level failures, wrong-region coefficients, or an application untethered from the case facts are where a 702 challenge to admissibility has traction. Do not overclaim. The objective is to expose assumptions and, where the gap is structural, to challenge admissibility. Neither this analysis nor any expert can guarantee an evidentiary ruling or an outcome.