Accident Reconstruction Physics: Velocity, Momentum and Skid Marks

When a driver applies full braking on a vehicle with locked wheels, the tyres slide across the road surface rather than rolling. The kinetic friction between tyre and road converts the vehicle's kinetic energy into heat and rubber deposition, decelerating the vehicle and leaving a visible skid mark. The length of that mark encodes the vehicle's speed at the onset of skidding.

The basic formula. Energy conservation gives the fundamental relationship. The kinetic energy at the start of the skid (1/2 mv2) equals the work done by friction during the skid (friction force times distance = m g mu d, where mu is the coefficient of friction and d is the skid length). Setting these equal and solving for v:

v = sqrt(2 mu g d)

where v is the initial speed (m/s), mu is the dimensionless coefficient of kinetic friction between tyre and road, g is gravitational acceleration (9.81 m/s2), and d is the skid-mark length in metres. This formula, sometimes written as v = sqrt(30 f d) in the US convention where d is in feet and the result is in miles per hour (with f substituted for mu), is the backbone of skid-mark reconstruction.

The coefficient of friction. Mu is not a universal constant; it is a property of the tyre-road pairing under the specific conditions at the time of the crash. Typical values range from 0.55-0.70 for worn asphalt on a dry road, 0.30-0.45 for wet asphalt, and 0.10-0.25 for ice or wet painted lane markings. The examiner measures mu by performing test skids with a comparable vehicle on the same road surface under similar conditions, or by using a drag sled or the Vericom VC3000 or similar drag-coefficient meter. The SAE J1655 standard specifies the protocol for skid test measurements. In the UK, TRL published reference friction coefficients for standard road surfaces that can be used when test skids are impractical. The NCHRP Report 600 provides equivalent US reference data.

Assumptions and their limits. The formula assumes: (1) the vehicle had all four wheels fully locked throughout the skid; (2) the road surface is level; (3) the friction coefficient is uniform along the skid length; (4) aerodynamic drag is negligible (reasonable below 100 km/h); and (5) the entire skid length is preserved and measured. Each assumption introduces error when violated. ABS-equipped vehicles do not produce classic skid marks because the wheels are not locked; they produce scuff marks or faint swipe marks that require different analysis (tyre deceleration over micro-slip cycles). A grade correction factor applies to uphill or downhill skids (the effective deceleration includes the gravity component along the slope). Partial skid marks, where the vehicle ran off a hard surface before stopping, require the examiner to account for the transition in friction coefficient.

Yaw marks. When a vehicle undergoes cornering under braking or with steered front wheels during a skid, the tyres slide at an angle to the vehicle's direction of travel, producing a yaw mark (a curved, striated impression) rather than a straight longitudinal skid mark. The radius of the yaw mark provides information about the vehicle's centripetal acceleration at that point: a = v2 / r, where a is the centripetal acceleration (bounded by the friction coefficient times g), v is the speed, and r is the radius of curvature of the yaw arc. If a is set equal to mu g at the limit of adhesion, then v = sqrt(mu g r). Yaw-mark radius measurement thus provides an independent speed estimate that cross-validates the skid-mark result. TRL Technical Paper No. 17 and the SAE PT-107 anthology (Forensic Accident Investigation) contain detailed methodology for yaw-mark analysis.

When two vehicles collide, momentum is conserved in the absence of external horizontal forces. This conservation law, combined with measured post-collision trajectories and vehicle weights, allows the reconstructionist to calculate the pre-impact speeds of both vehicles.

The impulse-momentum theorem. For a system of two vehicles, the total momentum before the collision equals the total momentum after the collision, provided external impulses (road friction during the collision, which acts for a very brief time over the collision contact patch) are small. For a head-on or right-angle collision lasting 100-150 milliseconds, the external friction impulse is indeed small compared with the collision impulse, and the approximation holds well. In a direct longitudinal collision (both vehicles moving on the same line):

m1 v1 + m2 v2 = (m1 + m2) V (for a perfectly inelastic collision, where the vehicles lock together post-impact)

where m1, m2 are the masses, v1, v2 are the pre-impact speeds, and V is the common post-impact speed. The post-impact speed V is estimated from the post-collision skid distances of the joined wreckage using the skid formula. This gives V, and if one of the pre-impact speeds is known or can be estimated independently, the other can be calculated. If both are unknown, additional constraints (yaw trajectories, camera footage, witness positions) are required.

Crush-zone energy absorption. Perfectly inelastic momentum conservation is an idealisation. In reality, the vehicle structures deform and absorb a significant fraction of the kinetic energy as plastic deformation of the crush zone. The SAE J2080 standard provides the Energy-Equivalent Speed (EES or DV, delta-V) methodology that quantifies the energy absorbed in the crush zone from the measured deformation depth. The European Association for Accident Research and Analysis (EVU) guidelines and the ENFSI Road Transport Physics group have published EES tables for common European vehicle models derived from crash tests. Subtracting the energy absorbed in crush from the total kinetic energy budget provides a more accurate pre-impact speed estimate. This is sometimes called the Energy-Balance Method and is the approach preferred in the SAE J1739 reconstruction framework.

Elastic vs inelastic collisions. A perfectly elastic collision preserves both momentum and kinetic energy; a perfectly inelastic collision preserves momentum but dissipates kinetic energy as deformation, heat, and sound. Real vehicle collisions are between these extremes. Minor side-swipe collisions at low relative speed may be close to elastic (minimal crush). High-speed frontal collisions are close to perfectly inelastic (vehicles crumple and interlock). The coefficient of restitution (e = relative velocity of separation / relative velocity of approach) quantifies this spectrum; for vehicle collisions, e typically ranges from 0 (perfectly inelastic) to 0.3 (partial rebound). Most reconstruction software (PC-Crash, HVE, WinSMASH) uses empirical e values from crash-test databases.

Right-angle (T-bone) and oblique collisions. For collisions where the vehicles approach at angles other than 180 degrees, momentum conservation must be applied as a vector equation in two dimensions. The reconstructionist measures the post-impact trajectories of both vehicles (from tyre marks, final rest positions, and intermediate contact points) and works backward through the collision kinematics to derive the pre-impact velocity vectors. This two-dimensional calculation is more sensitive to measurement error than the head-on case; the uncertainty in the speed estimate is correspondingly larger and must be stated in the expert report.

Pedestrian-impact reconstruction is among the most complex areas in accident reconstruction. The pedestrian is not a rigid body: limbs, torso, and head move independently under the impact forces, and the trajectory of different body segments provides different information about the collision dynamics. The seminal reference is Searle (1993), published in the Forensic Science International special issue on pedestrian trauma; subsequent refinements by Stcherbatcheff, Han and colleagues have updated the empirical coefficients.

The body-launch model. When a vehicle's front bumper strikes a standing pedestrian at the leg or hip, the lower body is accelerated forward rapidly while the upper body initially remains behind (due to inertia). The result is a rotation: the pedestrian wraps forward over the bonnet. As the pedestrian's head or upper torso contacts the bonnet or windshield and then leaves the vehicle, the body is launched as a projectile from approximately the height of the bonnet surface. The body then follows a roughly parabolic trajectory until it lands on the road.

The head-throw distance (HTD) model by Searle uses the horizontal distance from the contact zone (typically the front bumper or bonnet leading edge) to the final rest position of the head to estimate the vehicle speed. Searle's 1993 empirical formula is:

v = sqrt(g x HTD / 0.58)

where v is vehicle speed (m/s), g is 9.81 m/s2, and HTD is head throw distance in metres. The constant 0.58 is an empirically derived factor that encapsulates the average launch geometry (height and angle) for a typical adult pedestrian struck by a typical mid-size passenger car front. Different vehicle types (SUVs, heavy trucks, sports cars with low-profile bonnets) require modified factors.

The uncertainty in HTD-based speed estimates is substantial. Searle's original validation data showed a coefficient of variation of approximately 15-20% for predicted vs actual speed across the crash-test sample. This means a head throw distance of 10 metres (which formula gives approximately 40 km/h) has a genuine uncertainty of approximately plus or minus 6-8 km/h. The expert who presents a single-point estimate without this range violates the scientific basis of the method and is vulnerable to Daubert challenge.

Pedestrian kinematics and injury pattern. The impact geometry determines the pedestrian's primary contact zones and the resulting injury distribution. A sedan or hatchback (bonnet height approximately 700-900 mm) typically strikes the adult pedestrian at thigh or hip level, producing the characteristic bumper contact mark (petechial abrasions or bruising at the anterior thigh) and bonnet contact injuries to the upper body or head. An SUV or pickup truck (bonnet height 1,000-1,200 mm) strikes at hip or torso level, with different injury distribution and different throw trajectory. The forensic pathologist's injury findings and the reconstructionist's vehicle-geometry data must be reconcilable for the combined opinion to be credible. This intersection with the autopsy findings is where the forensic physics and forensic medicine disciplines must communicate, though the autopsy interpretation remains within the forensic medicine domain (see the Module 7 mechanical injury topic for the pathological side of this analysis).

Braking effect on pedestrian trajectory. If the driver applies braking during or after the initial pedestrian contact, the vehicle decelerates while the pedestrian is still on the bonnet or in the projectile phase. Vehicle deceleration introduces a forward component to the pedestrian's throw, increasing the throw distance beyond what would occur at constant speed. Failure to account for braking-induced velocity differential is a common error in simplified HTD calculations. PC-Crash and other simulation software model this dynamic interaction by treating the pedestrian as a multibody system with deformable joints.

Every reconstruction opinion carries uncertainty. The sources are predictable, quantifiable, and must be stated in the expert report if the opinion is to survive judicial scrutiny.

Measurement uncertainty. Skid-mark length measured with a tape is subject to approximately plus or minus 0.5-1% precision. Vehicle mass measured from the vehicle registration is nominal; the actual laden mass (with occupants and cargo) may differ. Friction coefficient measured from test skids on a different day has temporal variability because road-surface condition changes with temperature, moisture, and wear. Each uncertainty propagates through the formula to produce an uncertainty range in the speed estimate. The reconstructionist should conduct a formal sensitivity analysis: holding all other variables at their measured values and varying each uncertain parameter across its plausible range, then combining the resulting speed range. This is the method recommended in the SAE PT-107 reconstruction anthology and in the ENFSI Road Transport Physics guidelines.

Vehicle-specific factors. ABS, traction control, and stability control systems alter the wheel-road force pattern and invalidate the classical skid-mark formula. For ABS-equipped vehicles, the deceleration rate must be estimated from the vehicle's ABS performance specification (typically 0.75-0.85 g on dry asphalt) or from event data recorder (EDR) output. The EDR (also called the Crash Data Recorder, CDR) records pre-impact speed, throttle position, brake application, and steering angle at 500 ms intervals (or finer in newer systems); EDR data, when available, is a more direct and more reliable speed source than the skid-formula derivation. SAE J1698 and the NHTSA Rule 49 CFR Part 563 govern EDR data retrieval protocols in the US. European regulations since 2022 mandate EDR-equivalent black boxes on new vehicles under UN Regulation 160. India does not yet have a mandatory EDR requirement, though the 2019 Motor Vehicles Act amendments created a policy pathway for fleet vehicles.

Witness evidence. Eyewitness speed estimates are notoriously unreliable; research by Loftus (1979) and more recent meta-analyses show that layperson speed estimation errors can exceed 40% in absolute magnitude. Reconstruction opinions based primarily on eyewitness speed estimates, without physical-evidence corroboration, are at significant Daubert-challenge risk. Eyewitness accounts of traffic-light colour, vehicle position, and driver behaviour are generally more reliable than speed estimates and should be integrated with the physical evidence rather than treated as alternatives to it.

Computer simulation. PC-Crash (Steffan and Moser, Austria), Human Vehicle Environment (HVE, Engineering Dynamics Corporation, USA), and WinSMASH (NHTSA, USA) are the principal simulation platforms. Their physics engines are validated against crash-test data, and courts in the US, UK, and Germany have generally accepted their outputs as admissible when the underlying input data and assumptions are disclosed. The expert must not present simulation output as a measurement but as a model prediction conditional on the assumed inputs. Sensitivity runs varying the key inputs are expected in any high-quality simulation report.

Formal reconstruction standards establish both the scientific baseline and the courtroom admissibility foundation. They also discipline the reconstructionist's scope: the standards specify what conclusions can be drawn from which evidence, and they identify the assumptions that must be stated for each conclusion.

SAE International (US and global). The Society of Automotive Engineers publishes the principal collected works underlying US accident-reconstruction practice: SAE PT-107 (the Forensic Accident Investigation anthology) assembles the key peer-reviewed reconstruction papers; SAE J2205 is the recommended practice for the equations of motion in vehicle accident reconstruction; SAE J986 covers vehicle crush analysis; SAE J1698 and SAE J3022 cover Event Data Recorder retrieval and interpretation. SAE-framework reconstruction analyses are cited in US federal and state courts as evidence of the scientific consensus underlying reconstruction testimony, and Daubert challenges to properly applied SAE-framework analyses rarely succeed.

Institute of Road Traffic Engineers (IRTE, UK). The IRTE Forensic Collision Investigation manual is the UK reference for police forensic collision investigators and expert witnesses. FCIN accreditation (Forensic Collision Investigation Network) is the professional quality benchmark; accredited experts follow IRTE methodology. Expert testimony in Crown Court proceedings on collision reconstruction must comply with the Criminal Procedure Rules 2020 expert-report requirements: clear statement of the method, the assumptions, the data relied upon, and the uncertainty range.

ENFSI Road Transport Physics Working Group (EU). The ENFSI group published its first best-practice manual for road-traffic reconstruction in 2010, revised in 2019. The manual covers all principal reconstruction methods and specifies the reporting standards adopted across ENFSI member laboratories in 24 EU and associated states. In EU criminal proceedings, ENFSI-compliant reconstruction reports provide a credibility baseline that courts in Germany, France, the Netherlands, and the Nordic states consistently accept.

India. India does not yet have a formally published reconstruction standard comparable to the SAE accident-reconstruction collected works (anchored by SAE PT-107) or the IRTE manual. Reconstruction evidence in Indian courts relies on CFSL examiner qualifications and on adherence to international standards (typically SAE or IRTE). The BNS 2023 § 106 and the Motor Vehicles Act 1988 create the charging framework; the BNSS 2023 § 196 inquest provisions trigger executive-magistrate or judicial-magistrate inquiry into deaths in custody and road-traffic deaths under defined circumstances, with the police investigation under BNSS Chapter XIII running in parallel. Indian courts have increasingly scrutinised reconstruction testimony for methodological rigour: High Court judgments in Maharashtra, Delhi, and Tamil Nadu have cited the need for friction-coefficient testing and uncertainty ranges in reconstruction evidence, citing English Crown Court practice as a persuasive precedent.

Australia and Canada. Australian state police forensic reconstruction units follow the ACPO (now NPCC equivalent) guidance via the UK FCIN framework and additionally the ANZFSS best-practice notes. The RCMP NFLS and provincial forensic laboratories in Canada follow a methodology aligned with SAE J1739, supplemented by the Canadian Association of Road Safety Professionals (CARSP) technical notes.

Accident reconstruction testimony is more frequently challenged under Daubert than most forensic science disciplines, because the reconstruction opinion is often case-determinative and because the methodology is sufficiently quantitative that judges and cross-examining attorneys can engage with the mathematics.

Daubert standard (US federal courts and majority of state courts). Under Daubert v. Merrell Dow Pharmaceuticals (1993) and Federal Rule of Evidence 702, the trial judge acts as gatekeeper: expert testimony must be based on sufficient facts, must reflect reliable principles and methods, and the expert must have reliably applied the methods to the case facts. In accident reconstruction, the key Daubert failure modes are: (1) using the skid formula without measuring the friction coefficient on the actual road surface; (2) applying the formula to ABS-equipped vehicles without adapting for the non-locked-wheel dynamics; (3) stating a single-point speed estimate without a confidence interval; and (4) relying primarily on eyewitness speed estimates without physical-evidence corroboration. Reconstruction testimony has been excluded under Daubert in multiple US district courts when the expert lacked documented friction measurements, applied a formula outside its stated assumptions, or failed to account for vehicle-specific deceleration characteristics.

Frye standard (some US states). In states that retain the Frye v. United States (1923) general-acceptance standard (California, Illinois, New York until CPLR reforms, Florida for older cases), the question is whether the reconstruction method is generally accepted in the relevant scientific community. The SAE-framework methods are generally accepted; novel methods (new formulas, software with limited peer review, untested models) require a foundation hearing. The practical distinction from Daubert is less significant for standard reconstruction methods but matters when the expert uses a custom analysis or a rarely applied technique.

UK Crown Court. UK expert evidence on reconstruction is governed by the Criminal Procedure Rules 2020, Part 19. The expert's report must state the range of opinion and the reasons for it, must identify the issues on which opinion is offered, and must include a statement of the writer's understanding of their duty to the court. Cross-examination in UK road-traffic trials typically targets: the friction-coefficient measurement protocol, the assumption that the pre-impact skid length is accurately measured, the applicability of the skid formula to the specific road-surface conditions, and the independence of the reconstruction from the prosecution's factual theory.

India. Indian courts admit reconstruction evidence under BSA 2023 § 39 (opinion of experts, replacing IEA § 45). The standard for reliability is qualification-based rather than methodology-gate-keeping (India has not adopted a Daubert-equivalent procedural rule). However, Supreme Court and High Court decisions in serious road-traffic cases have increasingly required that reconstruction opinions be cross-validated by independent methods (e.g., skid formula corroborated by momentum analysis or EDR data) before being given substantial weight.

Modern vehicles carry event data recorders (EDRs) that log pre-crash kinematics at intervals of 0.1-0.5 seconds before the airbag deployment event. The data set typically includes vehicle speed (from wheel-speed sensors), engine throttle position, brake switch status, steering angle, and seat-belt status for each occupant position. Some advanced EDRs also record lateral acceleration, yaw rate, and gear position. This data is stored in non-volatile memory and survives the crash event unless the memory module is physically destroyed.

EDR data retrieval in the US is governed by SAE J1698 and NHTSA 49 CFR Part 563. The Bosch Crash Data Retrieval (CDR) tool is the dominant hardware used by crash investigators; it reads data from the airbag control module of most US-sold vehicles. In Europe, the 2019 EU General Safety Regulation mandated Event Data Recorders on all new type-approved passenger vehicles from 2022, with the data format aligned with UN Regulation 160. In India, there is no equivalent mandatory EDR regulation for private vehicles, though the Ministry of Road Transport and Highways has proposed rules for commercial fleet vehicles under the Motor Vehicles Act amendments.

EDR speed data is generally more reliable than skid-formula derivations for two reasons: it directly measures the vehicle's speed at specific moments before impact (rather than inferring it from physical evidence), and it is not subject to the friction-coefficient uncertainty that plagues skid-formula calculations. However, EDR data has its own limitations: the sample rate (typically 0.5 seconds) means that the recorded speed at minus 0.5 seconds before airbag deployment may not be the speed at the exact moment of impact; the wheel-speed sensor reading may be inaccurate if ABS was active; and EDR memory overwrites in some older vehicles, meaning the crash event data may have been lost if the vehicle continued to be driven after the crash.

The chain of custody for EDR data is critical. SAE J1698 requires that the CDR retrieval be performed before any repair to the airbag system, that the retrieval be witnessed or documented, and that the data file be hashed and certified. In US and UK courts, EDR data presented without documented retrieval protocol and chain of custody has been successfully challenged. In the 2015 Australian case of R v. Osman (Queensland), EDR data from a vehicle involved in a fatal collision was admitted as the primary speed evidence over defence objection, with the court finding the SAE J1698 retrieval protocol sufficient to establish reliability.