2026-10-20

Can a Tesla Roadster stick to the road and fly?

The broadcast holds three stunts: an inverted hold on a curved wall that ends with a top exit, a half-roll and a jet descent; a jet-powered barrel roll over a Semi; and a standing hop with a full roll. This report keeps the sourced research distinct from the v7 model, which reads every number from one exported run log: the jump peaks at 9.97 m, clears the staged envelope by 1.97 m, and the three stunts use 289.0 kg of gas, 39.0 kg more than a single 250 kg fill.

I saw a recent video from Tesla where the Roadster sticks to the road and even flies. Can you check if this is real, research how it could actually work physically, and write simulation code that shows the wall hold, jet lift, barrel roll and flat-ground landing?
Every number on this page carries a label: MEASURED a source measured it; CLAIMED Tesla or another party asserted it without independent measurement; ESTIMATED derived here, with the basis stated.

The simulation

The broadcast contains three stunts. In the first (06:30–07:22) the car climbs a curved wall past vertical, creeps along the inverted overhang for about ten seconds, reverses back down to the flat, climbs again straight to the top, leaves the lip, half-rolls about its length axis to come upright and descends on its jets. In the second (11:11–11:16) it approaches the Semi on flat ground, rises on the corner lift jets, barrel-rolls counter-clockwise (seen from behind) and lands beside the Semi in smoke. In the third (21:53–21:57) it hops from a standstill and completes a full roll. The wall HUD holds near 170 degrees, about 21 m altitude and about 4,700 kg-equivalent downforce (CLAIMED display readings that visibly lag the picture; they are never used as a same-frame height).

The v7 model follows that sequence on the broadcast event table, with an open wall profile fitted to the HUD pitch/altitude pairs, fan suction during surface contact, nozzle force, torque and gas computed together from four corner lift nozzles, variable mass, body-axis drag and a resolved suspension at every landing. Coordinates: x along travel, y left, z up; roll is right-handed about the body x axis, so the observed rolls run negative; heights are the body reference point (CG) above its static ride height. All computed results below are ESTIMATED and read from one exported run log. They describe this scenario, not a validated Tesla engineering model.

Wall apex · ESTIMATED
21.22 m
Overhang top 21.06 m (HUD 69.1 ft); fan keeps the contact force positive.
Jet lift time · ESTIMATED
0.672 s
Jump gas including roll and landing: 109.6 kg.
Jump landing · ESTIMATED
2.32 g
1.17 m/s touchdown; reproduced from the published landing inputs.
Three stunts · ESTIMATED
289.0 kg
39.0 kg more than one 250 kg fill; maximum load 2.56 g.
Curved wall stuntCurved wall stuntHeight (m)01020in contactairborneLocal horizontal distance (m)04080Jet lift over the SemiJet lift over the SemiHeight (m)01020Semi envelopein contactairborneLocal horizontal distance (m)04080
The wall stunt on one height scale: climb, inverted creep, reverse to the flat, second climb, top exit at 21.06 m (HUD 69.1 ft), the airborne apex of 21.22 m, the half-roll and the jet descent back over the approach; the separate jet-lift arc over the Semi peaks at 9.97 m. Heights are the body reference point (CG) above its static ride height; coordinates are local to each set. The 4 m high, 20 m long, 2.6 m wide envelope is an assumed staged-row bound, not the cab. Data: physics-v7.json.

Replay the wall stunt and the jump

Both views run the report's own stage code inside this page: the wall stunt on the left (climb, inverted creep, reverse to the flat, second climb, top exit, airborne half-roll, jet descent) and the jump over the Semi on the right, re-integrated each time a parameter moves. One control plays and scrubs both; each view carries its own clock and numbers.

Wall stunt



Jump over the Semi






Computing the run.


in contactairbornestrikes the Semi


Semi-implicit Euler at 2 ms with the same modules as physics-v7.json: nozzle allocation, body-axis drag, variable mass, integrated roll and the two-mass suspension. The wall view plays its 53 s stunt at about 2.2× speed and the jump its 9 s at about 0.38×, both on one progress control; drag either view to turn it. The jet lift time and plan-area drag sliders re-run the jump; at the defaults its numbers equal the stored run exactly, and a strike on the envelope turns the trail and the envelope red and stops the car where it hits. Inputs: physics-v7.json.



Speed through the three stuntsSpeed through the three stuntsSpeed |v| (m/s)004.623.49.246.913.870.318.493.8|v| of the integrated velocity vectorTime (s)
One speed everywhere: the norm of the velocity vector that the integrator carries (on the wall the prescribed path velocity u t, whose required drive force is back-solved). Wall climb, creep (0.683 m/s (2.46 km/h)), return, second climb to the 5.00 m/s exit, jet descent, the Semi approach and the standing hop. Data: speed, physics-v7.json.


Load factor through the three stuntsLoad factor through the three stunts|a - g| / g000.723.41.446.92.170.32.893.8wall exitwall touchdownroll startinvertedtouchdownroll startinvertedtouchdownModel load factorTime (s)
Peak modeled load is 2.56 g; landing peaks are 2.34 g (wall, at 63.350 s), 2.32 g (jump, at 84.416 s) and 2.44 g (hop, at 92.628 s), each re-integrated from its published inputs (mass, k, c, k_t, zero initial compression and preload, touchdown rate, nozzle magnitudes). The load passes through 0 g only at the instants the roll torque is zero. Not an occupant acceleration trace. Data: gLoad and touchdown-impulse events, physics-v7.json.


Wall contact normal forceWall contact normal forceNormal force (kN)012.916.724.533.536.150.247.76759.3Surface contact onlyTime (s)
Minimum contact force 16,930 N on the 170 deg overhang; the second climb's peak needs a friction coefficient of 0.864 and a pitch torque 0.259 of the load-transfer limit (peak 6,780 N m). Suction 33.35 kN (2 g) (ESTIMATED); the footage's 46.1 kN (about 4,700 kg on the HUD) stays CLAIMED. Data: normalN, physics-v7.json.


Cold-gas supply, one fill per stuntCold-gas supply, one fill per stuntGas remaining (kg)0067.523.413546.9202.570.327093.8Remaining supplyTime (s)
Wall 101.2 kg, jump 109.6 kg, hop 78.2 kg: 289.0 kg in total, 39.0 kg more than one 250 kg fill holds, so the model refills at each cut (whether the real supply was topped up between stunts is unverified). Data: propellant, physics-v7.json.


Barrel roll about the body x axisBarrel roll about the body x axisRoll angle (deg); negative is counter-clockwise seen from behind-4000-3001.1-2002.2-1003.304.4Body roll, integrated from nozzle torqueTime since jet lift (s)
The roll starts 0.58 s after takeoff and lasts 1.62 s (222 deg/s mean, 444 deg/s peak, counter-clockwise seen from behind), integrated from the torque the allocated nozzle set produces with PD tracking of the sin² rate schedule; the lock at -360 deg occurs when the residual is inside the deadband. Body-normal thrust rotates with the car, so it pushes downward while inverted. Data: roll, physics-v7.json.
Events of the run


Phase results use physics-v7.json, cross-checked against PHYSICS-FLAT.md at build time; broadcast times are the measured event table that sets the model timeline.
stunt / phaseevent (broadcast time)computed result · ESTIMATED
Wall climbUp the curve past vertical, 06:30–06:42Overhang top 21.06 m; pitch 170.0 deg; maximum speed 6.82 m/s; required friction up to 0.864
Wall holdInverted creep, fan sealed, 06:42–06:52Minimum normal force 16,930 N; creep 0.683 m/s (2.46 km/h); sampled speed near vertical 0.697 m/s
Wall returnReverse down to the flat, 06:52–07:03; hold to 07:10.5Rest-to-rest quintic; fan sealed throughout
Second climb and exitStraight to the top, 07:10.5–07:16.4Peak speed 9.00 m/s; exit at 5.00 m/s, 170.0 deg
Half-roll and descentAirborne 07:16.4 to the 07:20–07:22 landing180 deg in 1.10 s, then 10.0 deg nose-down levelling in 0.70 s; airborne 4.04 s; gas 101.2 kg; touchdown 1.18 m/s, peak 2.34 g
Flat-road jumpJets 11:11.5, takeoff 11:11.6Burn 0.672 s; apex 9.97 m; airborne 4.27 s
Over the SemiRoll 11:12.2–11:13.8, counter-clockwise from behindRoll 0.58 s after takeoff for 1.62 s; minimum body clearance 1.97 m; jump gas 109.6 kg
Flat landingTouchdown before 11:15.6 (obscured by smoke)Vertical 1.17 m/s; horizontal 13.76 m/s; peak 2.32 g; bottomed: false
Finale hopJets 21:53.5, roll 21:54.1–21:55.8, touchdown 21:56.5–21:57.1Burn 0.472 s; apex 5.20 m; airborne 2.92 s; gas 78.2 kg; touchdown 1.26 m/s, peak 2.44 g
All threeGas ledgerMaximum load 2.56 g; total gas 289.0 kg; single-fill shortfall 39.0 kg


Landing peaks from published inputs
Each peak is re-integrated from the inputs below with the landing equations of the governing block (two-mass heave, zero initial compression and preload, tire force from zero, nozzles closing at 100 kN/s), and the export refuses a run whose stated peak does not reproduce.
stuntpublished landing inputspeak (sample time)
Wallm 1599.2 kg (sprung 1439.2, unsprung 160); delta_s(0) = 0.0000 m, delta_t(0) = 0.24 mm, preload 0 N; z'(0) = -1.179 m/s; nozzles at contact 3.86 / 3.86 / 3.86 / 3.86 kN closing at 100 kN/s each2.34 g at 63.350 s
Jumpm 1590.8 kg (sprung 1430.8, unsprung 160); delta_s(0) = 0.0000 m, delta_t(0) = 1.35 mm, preload 0 N; z'(0) = -1.174 m/s; nozzles at contact 3.82 / 3.82 / 3.82 / 3.82 kN closing at 100 kN/s each2.32 g at 84.416 s
Hopm 1622.3 kg (sprung 1462.3, unsprung 160); delta_s(0) = 0.0000 m, delta_t(0) = 1.38 mm, preload 0 N; z'(0) = -1.263 m/s; nozzles at contact 4.19 / 4.19 / 4.19 / 4.19 kN closing at 100 kN/s each2.44 g at 92.628 s


Three variants of the same sequence
Only the fan-plus-jets configuration has a run in the supplied v7 data. The other configurations are qualitative distinctions, not counterfactual simulation results.
variantoutcomepeak loadgas usedbody clearance
Fan + lift jetsAll three modeled stunts completed (ESTIMATED)2.56 g289.0 kg1.97 m
Lift jets onlyNot computed; no fan-off wall run in these inputsNot computedNot computedNot computed
Fan onlyNot computed; suction cannot supply free-flight lift after seal lossNot computedNot computedNot computed


Governing equations in the model


World x along travel, y left, z up; body x forward, y left, z up (roof); R is body-to-world, di = body +z for every lift nozzle, ri = (±1.45, ±0.9, −0.35) m from the CG, g = 9.81 m/s², g0 = 9.80665 m/s², m = 1,450 kg dry + gas, u is the signed speed along the wall, κ the curvature of the CG path, t and n the wall tangent and vehicle-side normal.


translation
m v˙=m g+R∑iTi di b+Faero+Fcontactm\,\dot{\mathbf v} = m\,\mathbf g + R\sum_i T_i\,\mathbf d_i^{\,b} + \mathbf F_{\mathrm{aero}} + \mathbf F_{\mathrm{contact}}


rotation
I ω˙+ω×(Iω)=∑iri b×(Ti di b)+τcontactI\,\dot{\boldsymbol\omega} + \boldsymbol\omega\times(I\boldsymbol\omega) = \sum_i \mathbf r_i^{\,b}\times\big(T_i\,\mathbf d_i^{\,b}\big) + \boldsymbol\tau_{\mathrm{contact}}


single-axis form used
Ix φ¨=[∑iri×Fi]xI_x\,\ddot\varphi = \Big[\sum_i \mathbf r_i\times\mathbf F_i\Big]_x
Iy θ¨=−[∑iri×Fi]y=∑irx,i TiI_y\,\ddot\theta = -\Big[\sum_i \mathbf r_i\times\mathbf F_i\Big]_y = \sum_i r_{x,i}\,T_i


attitude
R˙=R [ω]×\dot R = R\,[\boldsymbol\omega]_\times
R=Ryaw(ψ) Rpitch(θ) Rroll(φ)R = R_{\mathrm{yaw}}(\psi)\,R_{\mathrm{pitch}}(\theta)\,R_{\mathrm{roll}}(\varphi)
di b=ez b\mathbf d_i^{\,b} = \mathbf e_z^{\,b}


mass
m=mdry+mgas(t)m = m_{\mathrm{dry}} + m_{\mathrm{gas}}(t)
m˙gas=−∑iTiIsp g0\dot m_{\mathrm{gas}} = -\sum_i \frac{T_i}{I_{sp}\,g_0}


drag in body axes
ub=RT(v−w)\mathbf u_b = R^{\mathsf T}(\mathbf v - \mathbf w)
Db=−12ρ ∥ub∥ diag⁡(K) ub\mathbf D_b = -\tfrac12\rho\,\lVert\mathbf u_b\rVert\,\operatorname{diag}(K)\,\mathbf u_b
Dw=R Db\mathbf D_w = R\,\mathbf D_b


wall contact
N=mu2κ+mgcos⁡θ+S−T⋅n−Faero⋅n≥0N = m u^2\kappa + m g\cos\theta + S - \mathbf T\cdot\mathbf n - \mathbf F_{\mathrm{aero}}\cdot\mathbf n \ge 0
S=Δp AsealedS = \Delta p\,A_{\mathrm{sealed}}


wall tangent
m u˙=Fdrive+T⋅t+Faero⋅t−mgsin⁡θm\,\dot u = F_{\mathrm{drive}} + \mathbf T\cdot\mathbf t + \mathbf F_{\mathrm{aero}}\cdot\mathbf t - m g\sin\theta
∣Fdrive∣≤μN\lvert F_{\mathrm{drive}}\rvert \le \mu N


nozzle allocation
∣τx∣/b+∣τθ∣/a≤F≤4fmax⁡−∣τx∣/b−∣τθ∣/a\lvert\tau_x\rvert/b + \lvert\tau_\theta\rvert/a \le F \le 4 f_{\max} - \lvert\tau_x\rvert/b - \lvert\tau_\theta\rvert/a
0≤Ti≤fmax⁡0 \le T_i \le f_{\max}


landing
ms z¨s=Fs+Tz+Dz−msgm_s\,\ddot z_s = F_s + T_z + D_z - m_s g
mu z¨u=Nt−Fs−mugm_u\,\ddot z_u = N_t - F_s - m_u g
Fs=max⁡(0,kδs+cδ˙s)F_s = \max(0, k\delta_s + c\dot\delta_s)
Nt=max⁡(0,ktδt)N_t = \max(0, k_t\delta_t)


landing start
δs(0)=0\delta_s(0) = 0
Npreload=0N_{\mathrm{preload}} = 0
z˙s(0)=z˙u(0)=−vtd\dot z_s(0) = \dot z_u(0) = -v_{\mathrm{td}}
c=2ζk m0c = 2\zeta\sqrt{k\,m_0}
δs≥δmax⁡⇒bottoming\delta_s \ge \delta_{\max} \Rightarrow \text{bottoming}


load factor
n=∥a−g∥gn = \frac{\lVert\mathbf a - \mathbf g\rVert}{g}


integration
an=F(staten)mn\mathbf a_n = \frac{\mathbf F(\mathrm{state}_n)}{m_n}
vn+1=vn+anΔt\mathbf v_{n+1} = \mathbf v_n + \mathbf a_n\Delta t
xn+1=xn+vn+1Δt\mathbf x_{n+1} = \mathbf x_n + \mathbf v_{n+1}\Delta t
Δt=0.002 s\Delta t = 0.002\ \mathrm{s}


Surface motion on the wall is prescribed kinematics on the broadcast timing; its drive force, normal force, friction coefficient and pitch torque are back-solved and reported as feasibility checks. Flight and landing are integrated. The rolls are integrated from the torque the nozzle allocation produces, so the integrated rate reaches zero at the lock, and thrust pushes along body z throughout (downward while inverted). The inertia tensor is a fixed uniform-box estimate and the nozzle arms are estimates. The scalar load output does not establish occupant safety.


Parameters used by the simulation
model inputvaluestatus / basis
Initial wet mass per stunt1,700 kgESTIMATED model scenario; see PHYSICS-FLAT.md
Dry mass1,450 kgESTIMATED model scenario; see PHYSICS-FLAT.md
Gravity used in v79.81 m/s²ESTIMATED model scenario; see PHYSICS-FLAT.md
Standard gravity for Isp9.80665 m/s²Defined constant; distinct from the 9.81 m/s² weight gravity
Maximum total lift thrust40 kNESTIMATED model scenario; see PHYSICS-FLAT.md
Four corner nozzles10 kN each, independently throttledESTIMATED model scenario; see PHYSICS-FLAT.md
Nozzle positions from the CG(±1.45, ±0.9, -0.35) m, thrust along body +zESTIMATED model scenario; see PHYSICS-FLAT.md
Roll inertia889.7 kg m²ESTIMATED model scenario; see PHYSICS-FLAT.md
Pitch inertia3,444.9 kg m²ESTIMATED model scenario; see PHYSICS-FLAT.md
Jet valve slew10 kN per 0.1 s per nozzleESTIMATED model scenario; see PHYSICS-FLAT.md
Cold-gas reference Isp69 sMEASURED reference (NASA nitrogen thruster, vacuum chamber); not a Roadster test
Gas per fill250 kgCLAIMED · about 250 kg of air; usable mass and refills unverified
Full-thrust mass flow59.11 kg/sESTIMATED model scenario; see PHYSICS-FLAT.md
Estimated fan suction33.35 kN (2 g)ESTIMATED model scenario; see PHYSICS-FLAT.md
Separate footage display claim46.1 kN (about 4,700 kg on the HUD)CLAIMED · footage HUD; not substituted for estimated suction
Truck/staged-row clearance envelope4 m high, 20 m long, 2.6 m wideESTIMATED model scenario; see PHYSICS-FLAT.md
Target minimum body clearance1.8 mESTIMATED model scenario; see PHYSICS-FLAT.md
Jump run-up speed14 m/s (50.4 km/h)ESTIMATED · not frame-measured
Suspension stiffness400,000 N/mESTIMATED · whole-car heave rate, not a per-spring rate
Suspension travel0.27 mReference from another vehicle's damper; not a Roadster value
Suspension damping ratio0.6ESTIMATED model scenario; see PHYSICS-FLAT.md
Damping coefficient31,292 N s/mESTIMATED model scenario; see PHYSICS-FLAT.md
Tire vertical stiffness1,600,000 N/m (four tires, ESTIMATED)ESTIMATED model scenario; see PHYSICS-FLAT.md
Unsprung mass160 kgESTIMATED model scenario; see PHYSICS-FLAT.md
Air density1.225 kg/m³ESTIMATED model scenario; see PHYSICS-FLAT.md
Drag areas in body axes0.70 / 4.77 / 10.08 m² (x, y, z)ESTIMATED model scenario; see PHYSICS-FLAT.md


The source research below uses standard gravity 9.80665 m/s² for its historical derivations; the current run weighs the car with 9.81 m/s² and keeps 9.80665 m/s² only inside the Isp conversion. This explains the historical 33,343 N suction estimate versus the v7 estimate. Neither is the footage's 46.1 kN display claim. Clearance is measured to the rotated body outline, not merely its center of mass.
What the research found


Reduction of slice c1 (angles A-I) and slice c2 (angles J-R), 2026-10-20, no new retrieval. MEASURED: a source measured, tested, certified or formally recorded it. CLAIMED: Tesla, Elon Musk or another party asserted it without independent measurement. ESTIMATED: derived, formula and inputs shown. Refuted claims appear only under Refuted. Derivations use g0 = 9.80665 m/s^2 (a defined standard value, not a site measurement; the model's weight uses g = 9.81 m/s^2 and keeps g0 only for Isp) [physics.nist.gov].


Answer


Snopes rates the video authentic, dated 2026-10-15 (MEASURED); the frame audit behind this report checks on-screen events, times and projections and does not by itself certify the footage as real [snopes.com], and the Tesla event stream exists (MEASURED) [youtube.com], with the broadcast events now read frame by frame for the model timeline. In Tesla's description it shows a sealed-floor fan-suction system holding the inverted car against a track while separate cold-gas thrusters assist brief hops, not sustained flight (CLAIMED) [caranddriver.com]. The "Rocket" package has 10 nozzles on 600 L of gas, the "Thrust" package 4 rear nozzles on 200 L (CLAIMED) [youtube.com]; the 10 split as 4 rear, 4 corner lift, 2 yaw (CLAIMED) [topgear.com]. It does not establish a production-representative car, carries no independent instrumented dataset (n/a, MEASURED search finding) [caranddriver.com], and implies no approval: disclosed public-road approval is absent and an airspace restriction is not certification (MEASURED) [snopes.com]; CBS reports an FAA disclaimer and NHTSA oversight, no Part 107 approval (MEASURED) [cbsnews.com]. Tesla's patent US12377920B1 covers downforce only, separate from the thrusters (CLAIMED) [patentsgazette.uspto.gov].


Each stunt has an ordinary governing relation. Sticking is sealed-floor suction S = dp * A_s with inverted contact requiring S >= m * g. Tesla claims about 2 g of suction on a target mass under 1,700 kg (CLAIMED) [wired.com]; at 1,700 kg that is S0 = 2 * 1700 * 9.80665 = 33,343 N (ESTIMATED) [wired.com] against weight 16,671 N (ESTIMATED) [physics.nist.gov], from 4 x 20 kW = 80 kW of fans (CLAIMED) [wired.com]; suction can maintain an inverted static hold; the actual wall is an open overhang, not a full circular course. Flying is cold-gas thrust F = mdot * c_eff, c_eff = Isp * g0: nitrogen Isp 69 s (MEASURED) [ntrs.nasa.gov] gives 676.7 m/s (ESTIMATED) [ntrs.nasa.gov]; the claimed 250 kg charge (CLAIMED) [caranddriver.com] holds I_total = 250 * 69 * 9.80665 = 169,165 N s (ESTIMATED) [ntrs.nasa.gov], which is 10.2 s of ideal hover from t = m_g * Isp * g0 / (m * g) (ESTIMATED) [ntrs.nasa.gov] or 2.1 s (ESTIMATED) [ntrs.nasa.gov] at the claimed rear thrust of over 80 kN (CLAIMED) [caranddriver.com]. Unpowered projectile comparisons use R = v0^2 * sin(2a) / g and t_air = 2 * v0 * sin(a) / g; the observed jump instead begins with jet lift from level ground and rolls about the long axis. A separate, hypothetical circular-track comparison uses the centripetal condition v_top^2 >= g * r, v_bottom,min = sqrt(5 * g * r) for a lossless passive car; both are point-mass relations applied here as ESTIMATED bounds [hyperphysics.phy-astr.gsu.edu].


Not shown, and not possible with the claimed hardware: sustained flight, because ideal hover lasts 8.8-10.3 s (ESTIMATED, 250 kg / mdot_hover at Isp 60-70 s) [teslanorth.com]; a hypothetical end-over-end free-flight circle (not the observed longitudinal barrel roll), because r = 10 m at 15 m/s needs mean radial force m * v^2 / r = 38,250 N for 2 * pi * r / v = 4.19 s, and, because jets must supply T = m(a - g), a constant-speed vertical circle needs about 40.1 kN mean thrust, 54.9 kN peak and about 167.9 kN s of scalar jet impulse against the 169.2 kN s a 250 kg tank holds (ESTIMATED); that near-equality compares a hypothetical circle, not the observed barrel roll, and proves nothing impossible on its own [hyperphysics.phy-astr.gsu.edu]; wing-borne flight, because the maneuver is thrust-borne with no aero-flight dataset (see Refuted), and closed-wheel liftoff is configuration dependent, the Mercedes CLR lifting at 300-320 km/h (MEASURED driver account) [web.archive.org]; repeated use, because the thrusters are a single use of about 5 s with a 2-6 h refill (CLAIMED) [youtube.com]; occupant safety, because occupancy is not independently verified (the event transcript says the “Starman” occupant is “not a real human”, CLAIMED), G-LOC averages 5.2 g (MEASURED) [physoc.onlinelibrary.wiley.com] and no universal passenger impact limit exists (n/a, MEASURED search finding) [normsplash.com]; and road legality, because no approval was disclosed (MEASURED) [snopes.com].


How it sticks: downforce and fan suction


Passive downforce is negative lift: L = 0.5 * rho * v^2 * A * C_L, D = 0.5 * rho * v^2 * A * C_D, rho air density [kg/m^3], v speed [m/s], A reference area [m^2], C_L and C_D dimensionless [www1.grc.nasa.gov] [www1.grc.nasa.gov]; C_L * A = -2 * F_down / (rho * v^2) with lift positive upward (or +2 F_down / (rho v^2) for a separately defined positive downforce coefficient), scaling as v^2. Fan suction is S = dp * A_s, dp under-floor pressure deficit [Pa], A_s sealed area [m^2], zero after seal loss; fan power is P = dp * Q / eta, Q volume flow [m^3/s], eta efficiency [-] [doi.org].


No F1 downforce in newtons was carried; the F1 number is occupant load above 5 g in corners and braking (MEASURED) [mercedesamgf1.com]. Anchors: Dodge Viper ACR Extreme Aero over 1,700 lbf at 177 mph (CLAIMED) [media.stellantisnorthamerica.com], 7,563 N at 79.1 m/s, C_L * A = 2 * 7563 / (1.225 * 79.1^2) = 1.97 m^2 (ESTIMATED) [media.stellantisnorthamerica.com]; the McMurtry Speirling fan car at 2,000 kgf standstill downforce (CLAIMED) [mcmurtry.com], 19,613 N (ESTIMATED) [mcmurtry.com], a lateral figure above 3 g that is attribution only (CLAIMED) [mcmurtry.com], and a Goodwood run of 39.08 s (MEASURED) [autosport.com]. For the Roadster, S0 = 33,343 N (ESTIMATED) [wired.com] over an illustrative 2.5-3.0 m^2 (ESTIMATED) [wired.com] needs dp = 11.1-13.3 kPa (ESTIMATED) [doi.org]. A displayed 4,700 kgf, 46.1 kN, is a Tesla display value (CLAIMED) [electrek.co]; neither figure is measured, and a stationary inverted car proves only S >= 16,671 N (ESTIMATED) [physics.nist.gov].


Tire grip is load sensitive: F_y = K * F_z^n, mu_eff = K * F_z^(n-1), F_z per-tire load [N], K a fit constant, n the load exponent; from the Milliken endpoints n = 1 + ln(mu2/mu1) / ln(F_z2/F_z1) = 0.8186 (ESTIMATED) [en.wikipedia.org]. The performance-tire band is mu = 1.0-1.3 (ESTIMATED) [en.wikipedia.org]; a third-party reconstruction uses mu = 1.26 (CLAIMED) [wooten-labs.com], giving a tire-only 0-60 mph floor t = v / (mu * g) = 26.82 / (1.26 * 9.80665) = 2.17 s, an idealized tire-only floor that does not bound a fan-loaded car (ESTIMATED) [wooten-labs.com]; downforce raises N_total = m * g + F_down and a_lat,max = mu_eff * g * (1 + DF / W), DF groundward force [N], W weight [N].


Historical circular-track comparison, not the Roadster wall geometry: loop with corrected signs, r radius [m], theta angle from the bottom [rad], N >= 0 inward track normal, S trackward suction, T_in signed inward thrust: N - S + T_in - m * g * cos(theta) = m * v^2 / r. At the top N_top = m * v_top^2 / r + S - T_in - m * g, so v_top,min = sqrt(max[0, r * (g + (T_in - S) / m)]). Passive: v_top^2 >= g * r, v_bottom^2 = v_top^2 + 4 * g * r, v_bottom,min = sqrt(5 * g * r), lossless entry height h_min = 2.5 * r (ESTIMATED) [hyperphysics.phy-astr.gsu.edu]. For r = 10 m: v_top,min = 9.90 m/s, v_bottom,min = 22.1 m/s, h_min = 25 m (ESTIMATED) [hyperphysics.phy-astr.gsu.edu]; for the Jaguar-derived r = 9.54 m: 9.67 and 21.63 m/s (ESTIMATED) [guinnessworldrecords.com]. With S0 on 1,700 kg, S/m = 19.6 m/s^2 > g, so v_top,min = 0 and the inverted car rests with N_top = S - m * g = 16,671 N (ESTIMATED) [wired.com]; at S = m * g there is no margin. Inward thrust raises the threshold; only trackward thrust substitutes for suction.


How it flies: cold-gas thrusters


Thrust is F = mdot * v_e + (p_e - p_a) * A_e, mdot mass flow [kg/s], v_e exit velocity [m/s], p_e and p_a exit and ambient pressure [Pa], A_e exit area [m^2] [grc.nasa.gov]; c_eff = F / mdot, Isp = c_eff / g0 [s], mdot = F / (Isp * g0). Ideal expansion v_e = sqrt(2 * gamma * R_s * T0 / (gamma - 1) * [1 - (p_e / p0)^((gamma - 1) / gamma)]), gamma heat-capacity ratio, R_s specific gas constant [J/(kg K)], T0 and p0 stagnation state, gives 714 m/s for air at 700:1 and a nitrogen vacuum limit of 789.5 m/s (ESTIMATED) [diva-portal.org]; sound speed is not a cap. Nitrogen R_s = 296.8 J/(kg K) (MEASURED) [diva-portal.org]. Reference nitrogen Isp is 57-73 s with an experimental anchor of 69 s (MEASURED) [ntrs.nasa.gov]; a small-satellite analogue gives 65 s at 24 C (MEASURED) [jossonline.com]; neither is a Roadster test. Baseline c_eff = 69 * 9.80665 = 676.66 m/s (ESTIMATED) [ntrs.nasa.gov].


Inventory follows m_g = p * V / (Z * R_s * T), p absolute pressure [Pa], V volume [m^3], Z compressibility [-], T temperature [K]. Inputs: 10,000 psi, about 68.95 MPa (CLAIMED) [youtube.com]; 600 L (CLAIMED) [space.com]; 300 K assumed (ESTIMATED) [webbook.nist.gov]. Ideal gas (Z = 1) gives 68.95e6 * 0.6 / (296.8 * 300) = 464.6 kg (ESTIMATED) [diva-portal.org]; the NIST equation of state gives about 490 kg/m^3 at 700 bar, 300 K (MEASURED) [webbook.nist.gov], hence 294 kg (ESTIMATED) [webbook.nist.gov]. The reported charge of about 250 kg (CLAIMED) [caranddriver.com] matches Z = 1.9, 464.6 / 1.9 = 244.5 kg (ESTIMATED; Z of about 1.86 would give exactly 250 kg, and Z is not chosen to fit the claim) [diva-portal.org], or a lower fill; a display reading 5,500 psi, about 37.92 MPa (CLAIMED) [evxl.co], is a separate unresolved state.


Total impulse I_total = m_used * c_eff: 250 * 676.66 = 169,165 N s (ESTIMATED) [ntrs.nasa.gov]; the 60-70 s band gives 147,100-171,616 N s (ESTIMATED) [jossonline.com]. Hover needs F = m * g = 16,671 N (ESTIMATED) [physics.nist.gov], so mdot_hover = 16,671 / 676.66 = 24.6 kg/s and t_hover = 250 / 24.6 = 10.15 s (ESTIMATED) [ntrs.nasa.gov], 8.82-10.29 s across the band (ESTIMATED) [teslanorth.com]. At over 80 kN rear thrust (CLAIMED) [caranddriver.com] the budget lasts 169,165 / 80,000 = 2.11 s (ESTIMATED) [ntrs.nasa.gov]; no corner-lift rating was recovered. The interview's about 5 s on the 200 L Thrust tank with a 2-6 h refill (CLAIMED) [youtube.com] and the 5-6 s, up-to-6 h figures (CLAIMED) [torquenews.com] are boost durations at unknown thrust; 80 kN for 5 s from 250 kg would need c_eff = 1,600 m/s (ESTIMATED) [caranddriver.com], which the adopted room-temperature air or nitrogen model cannot deliver (an ideal 300 K helium vacuum limit is about 1,765 m/s).


Energy: w_iso = R_s * T1 * ln(p1 / p2), w_ad = c_p * (T1 - T2); the air example (R_s = 287.05, c_p = 1005 J/(kg K), T1 = 300 K, ratio 700) gives 564 kJ/kg isothermal and 255 kJ/kg adiabatic, so 141 MJ and 64 MJ for 250 kg (ESTIMATED) [diva-portal.org]; p * V is a scale, not a bound. Vessel rules recovered: hydrogen burst at no less than 2 times working pressure (MEASURED) [ecfr.gov]; CNG burst at 2.25 times service with 13,000 and 5,000 cycles (MEASURED) [ecfr.gov]; neither is a nitrogen-car rule and no Roadster vessel data exist. Comparison: the Falcon 9 first stage flips on gaseous nitrogen, 8 thrusters in 2 pods (MEASURED) [alexcortiella.com]; no verified Falcon thrust or Isp table was recovered, only secondary estimates [spacex-atlas.com]; Dragon offers no cold-gas comparator because Draco is hypergolic (see Refuted).


Jet lift, barrel roll and landing


The observed jump begins on flat asphalt: the downward jets provide vertical impulse before the car crosses the Semi. During unpowered portions, z = z0 + vz0 t − g t² / 2; the powered portions require integrating the rotated thrust vector. The longitudinal barrel roll is not an end-over-end flight circle. Historical jump records below are comparisons, not the launch geometry or timing of this stunt.


Separate historical records: Travis Pastrana's 81.99 m rally-car jump, 2009-12-31 (MEASURED) [guinnessworldrecords.com], no angle or landing trace; launch 91 mph, 40.68 m/s (CLAIMED) [worldrecordacademy.com], inverting to a = asin(81.99 * 9.80665 / 40.68^2) / 2 = 14.5 degrees, t_f = 2.08 s, dH = 5.29 m, v_z,land = 10.2 m/s (ESTIMATED) [worldrecordacademy.com] (barge height difference unknown). Loops: Guinness recorded the 2015 Jaguar F-PACE loop at 19.08 m (MEASURED) [guinnessworldrecords.com], a height, so r = 9.54 m (ESTIMATED) [guinnessworldrecords.com] assumes a circle; Fifth Gear's Aygo looped at 36 mph and 6 G (CLAIMED) [eng.cam.ac.uk]; the 2012 Hot Wheels loop ran 85 km/h and 7 G (CLAIMED) [guinnessworldrecords.com]; none has a released load trace.


Landing: a_mean = v_z^2 / (2 * s), s effective vertical stopping distance [m], n_net = a_mean / g, proper support 1 + n_net; a half-sine pulse a_pk * sin(pi * t / tau) has dv = 2 * a_pk * tau / pi, its pi / 2 peak-to-mean ratio being pulse specific. Stopped in the Toyota Rally1 damper travel of 270 mm (MEASURED) [toyotagazooracing.com], a 10.2 m/s sink gives 104 / 0.54 = 192.7 m/s^2, 19.6 g net (ESTIMATED) [toyotagazooracing.com]; in trophy-truck travel of 26 in front, 31-32 in rear (MEASURED) [masonmotorsports.com] [jimcoracing.com], 0.787-0.813 m gives 6.5-6.7 g net (ESTIMATED) [masonmotorsports.com]. Codified sink rates are aircraft rules: 3.048 m/s at landing weight (MEASURED) [ecfr.gov], 3.658 m/s reserve energy (MEASURED) [law.cornell.edu], legacy Part 23 7-10 ft/s with load floors 2.67 and 2.0 (MEASURED) [govinfo.gov]. Pack shock rules bound the machine: UN 38.3 applies 150 gn for 6 ms to small cells and 50 gn for 11 ms to large cells, 18 shocks, with mass-dependent battery conditions that do not transfer to every automotive pack (MEASURED) [unece.org]; GB 38031-2025 applies 7 g half-sine for 6 ms, 12 vertical shocks (MEASURED) [120.79.24.31]; their dv is 0.26 m/s for 7 g / 6 ms and 3.43 m/s for 50 g / 11 ms (ESTIMATED) [unece.org], not survivability.


Human tolerance has no single number. +Gz visual symptoms span 2-7 g, centered at 4 g (MEASURED) [medcoe.army.mil]; G-LOC averages 5.2 g (MEASURED) [physoc.onlinelibrary.wiley.com]; F1 drivers sustain over 5 g (MEASURED) [mercedesamgf1.com]. A reproduced HANS sled test at 45 mph and a 45 degree seat cut neck tension from 1,352 to 296 lbf (MEASURED) [thejns.org], 78.1 percent (ESTIMATED, (1352 - 296) / 1352) [thejns.org]. No universal landing limit or measured rally landing peak was recovered (n/a, MEASURED finding) [doi.org]; occupancy of the stunt car is not independently verified; the transcript's “not a real human” statement is CLAIMED.


Airborne attitude is finite but real: I_y * dq_p/dt = sum(M_y), M_y = r_z * F_x - r_x * F_z in Cartesian components (the nose-up torque used by the v7 model is sum r_x,i T_i about -y), each nozzle's thrust taken as T_i R d_i with its own body direction d_i (the cos/sin form holds only for a body-longitudinal nozzle) and summed separately; wheel momentum H_w = n_w * I_w * v / r_dyn on the ground (no slip; in flight H_w = sum I_w omega_w, independent of the car's speed) exchanges with the body as I_y * dq_p = -dH_w when no jet or aerodynamic pitch torque acts. Priors: I_y = 2,500 kg m^2 for a 2,000 kg, 4.3 m by 1.2 m example (ESTIMATED) [mdpi-res.com], I_w = 1.5 kg m^2 per wheel (ESTIMATED) [doi.org]. Simulation method: at minimum three coordinates and three rates with unilateral contact N = max(0, k * delta + c * ddelta/dt), friction sqrt(F_x^2 + F_y^2) <= mu * N, explicit release when the gap opens (ESTIMATED) [link.springer.com]; reference engines expose two contact families, NSC and SMC (MEASURED) [api.projectchrono.org], and BeamNGpy does not disclose its integrator (MEASURED) [beamngpy.readthedocs.io]. Energy, momentum and penetration checks gate each run; RK4 suits flight, not stiff contact.


Research parameter evidence
This table preserves source claims and historical reference values. It is not the current model input table; the Roadster wall profile and flat-ground jump assumptions are documented above.





keyquantitysymbolvalueunitlabelsource URLnote




vehicle.massTarget massm1700kgCLAIMEDhttps://www.wired.com/story/how-tesla-made-its-new-230-mph-car-fly/Stated as under 1,700 kg; wet/dry unknown


vehicle.motorPowerWheel powerPn/aWESTIMATEDnoneNo Roadster power figure; only wheel torque claimed


vehicle.dragCdADrag areaC_D*An/am^2ESTIMATEDnoneNo Roadster aero data


vehicle.liftClALift areaC_L*An/am^2ESTIMATEDnoneNo passive aero coefficients


vehicle.muTire frictionmu1.261CLAIMEDhttps://wooten-labs.com/roadster/Reconstruction inside the 1.0-1.3 band; n = 0.8186


vehicle.rollingResistanceRolling resistanceC_rrn/a1ESTIMATEDnoneNot sourced by any angle


vehicle.wheelbaseWheelbaseL_wbn/amESTIMATEDnoneNo geometry recovered


fan.downforceFan suctionS033343NESTIMATEDhttps://www.wired.com/story/how-tesla-made-its-new-230-mph-car-fly/2 * m * g0 from the claimed 2 g; display unmeasured


fan.powerFan powerP_fan80000WCLAIMEDhttps://www.wired.com/story/how-tesla-made-its-new-230-mph-car-fly/4 x 20 kW electrical


thrusters.thrustRear thrustF_rear> 80000NCLAIMEDhttps://www.caranddriver.com/news/a74145531/tesla-roadster-revealed/Over 80 kN; lift-nozzle rating unknown


thrusters.ispNitrogen IspIsp69sMEASUREDhttps://ntrs.nasa.gov/citations/19930066300Reference anchor, 57-73 s range; not Roadster


thrusters.propellantMassGas massm_g250kgCLAIMEDhttps://www.caranddriver.com/news/a74165059/tesla-roadster-thrusters-fan-system-explained/EOS implies more; usable fraction unknown


thrusters.tankVolumeGas volumeV_R0.6m^3CLAIMEDhttps://www.space.com/technology/spacex-thrusters-just-made-a-tesla-roadster-fly-and-elon-musks-car-is-still-out-there-in-space200 L plus 400 L


thrusters.tankPressureStorage pressurep68950000PaCLAIMEDhttps://www.youtube.com/watch?v=JdkIOhFmbKI10,000 psi interview; lower display is separate


course.obstacleHeightObstacle heighth_obsn/amESTIMATEDnoneNo sourced obstacle; record apex rise bounds it


course.obstacleWidthObstacle widthw_obsn/amESTIMATEDnoneNo sourced obstacle; record range bounds it


suspension.travelSuspension strokes0.27mMEASURED (another vehicle's damper travel, used as a reference)https://toyotagazooracing.com/wrc/cars/2025/Rally1 damper analogue; no Roadster value


suspension.stiffnessSpring ratekn/aN/mESTIMATEDnoneNot sourced


air.rhoAir densityrho1.225kg/m^3ESTIMATEDhttps://www1.grc.nasa.gov/beginners-guide-to-aeronautics/lift-equation/ISA sea level


human.sustainedGzGLOCSustained +Gz (average G-LOC onset, not a limit)n_GLOC5.2gMEASURED (2021 review re-citing a 1944 study)https://physoc.onlinelibrary.wiley.com/doi/full/10.1113/EP089712Average G-LOC; not an allowance


human.impactGLimitImpact g limitn_impn/agESTIMATEDnoneNo universal passenger limit recovered


records.longestCarJumpMLongest car jumpR_P81.99mMEASUREDhttps://www.guinnessworldrecords.com/world-records/longest-rally-car-ramp-jumpRally-car category; pickup separate


records.loopHeightRecordMLargest car loop (height)H_J19.08mMEASUREDhttps://www.guinnessworldrecords.com/news/2015/9/new-jaguar-car-and-stunt-pro-terry-grant-defy-gravity-with-largest-loop-the-loop-396097Guinness records the height; a 9.54 m radius assumes a circular path


roadster.claimed060sClaimed 0-60 mpht_601.9sCLAIMEDhttps://en.wikipedia.org/wiki/Tesla_Roadster_(second_generation)2017 target; 2026 figures unresolved


roadster.claimedTopSpeedClaimed top speedv_max250mphCLAIMEDhttps://en.wikipedia.org/wiki/Tesla_Roadster_(second_generation)Over 250 mph, 2017


roadster.thrusterCountNozzle countn_R10countCLAIMEDhttps://www.topgear.com/car-news/electric/tesla-roadster-has-finally-landed-thrusters-0-60mph-059s-and-flight4 rear, 4 lift, 2 yaw


roadster.airVolumeLGas volumeV_R600LCLAIMEDhttps://www.space.com/technology/spacex-thrusters-just-made-a-tesla-roadster-fly-and-elon-musks-car-is-still-out-there-in-spaceNitrogen per engineering





Verified claims



Partial, unresolved and refuted


Partial and unresolved:



Refuted, with corrections:



Contradictions between sources


    

  1. Mass label: slice c1 carried 1,700 kg as ESTIMATED, slice c2 as CLAIMED. The angle verifiers record "under 1,700 kg" as Tesla's stated target in several domains [wired.com] [en.wikipedia.org]; CLAIMED wins.
    2. 

  2. Jaguar 19.08 m: c1 says height, c2 says diameter. The angle I verifier reads Guinness as recording height [guinnessworldrecords.com]; c1 wins, r = 9.54 m is an assumption.
    3. 

  3. Hot Wheels geometry: c1 says height cannot fix the radius; c2 reports 60 ft diameter beside 66 ft height (MEASURED as reported) [espn.com]. Both hold: a reported diameter without a surveyed drawing; Jaguar stays the record row.
    4. 

  4. Pressure: 10,000 psi interviews (CLAIMED) [youtube.com] versus a 5,500 psi display (CLAIMED) [evxl.co] plus a lower fill rating in angle O without a resolvable URL; separate scenarios, never averaged.
    5. 

  5. Gas identity: engineering nitrogen with external fill (CLAIMED) [youtube.com] versus air and self-recharge wording; nitrogen is the preferred attribution, not proof.
    6. 

  6. Inventory: EOS 490 kg/m^3 gives 294 kg (MEASURED density, ESTIMATED mass) [webbook.nist.gov] versus claimed 250 kg (CLAIMED) [caranddriver.com] versus ideal-gas 464.6 kg; EOS wins; the claim is carried separately.
    7. 

  7. Impulse: 80 kN x 5 s / 250 kg implies 1,600 m/s (ESTIMATED) [caranddriver.com], incompatible with 676.66 m/s (ESTIMATED) [ntrs.nasa.gov]; the Isp budget wins; rear force does not rate lift jets.
    8. 

  8. Exhaust speed: angle O's 640-700 m/s conflicts with its own 60-70 s Isp, which gives 588-686 m/s (ESTIMATED) [jossonline.com]; g0 * Isp wins.
    9. 

  9. Launch timing: 0.81 s live, 0.85 s engineering, 1.0 s recorded, 0.55-0.59 s rollout-trimmed are distinct claims (CLAIMED) [electrek.co]; none elects a record.
    10. 

  10. Suction: 33.3 kN (ESTIMATED) [wired.com] versus the 46.1 kN display (CLAIMED) [electrek.co]; neither measured; 2 g is the input, the display a separate CLAIMED state, not a calibrated upper bound.
    11. 

  11. Vehicles: the 2017 AWD three-motor 620 mi concept (CLAIMED) [en.wikipedia.org] is separated from the 2026 RWD two-motor 310-320 mi car; the Nevera motor sum 2 x 226 + 2 x 450 = 1,352 kW is preferred over the claimed 1,408 kW for per-motor modeling (CLAIMED) [rimac-newsroom.com].
    12. 

  12. Loop sign: an earlier note's "inward force lowers the minimum speed" is wrong; only trackward force does, and the corrected equation wins.
    13. 



Gaps



Sources


    

  1. https://www.snopes.com/fact-check/tesla-flying-roadster-car-real/—Snopes, 2026-10-16; authenticity, approval.
    2. 

  2. https://www.youtube.com/live/qt5p-_tE8d0—Tesla, 2026-10-15; event stream.
    3. 

  3. https://www.caranddriver.com/news/a74165059/tesla-roadster-thrusters-fan-system-explained/—Car and Driver, 2026-10-17; roles, 250 kg.
    4. 

  4. https://www.caranddriver.com/news/a74145531/tesla-roadster-revealed/—Car and Driver, 2026-10-15; 80 kN.
    5. 

  5. https://www.topgear.com/car-news/electric/tesla-roadster-has-finally-landed-thrusters-0-60mph-059s-and-flight—Top Gear, 2026-10-16; nozzle split.
    6. 

  6. https://www.youtube.com/watch?v=JdkIOhFmbKI—Top Gear, 2026-10-16; interview figures.
    7. 

  7. https://www.torquenews.com/11826/tesla-says-roadsters-thrusters-are-one-time-use-last-only-few-seconds-and-take-6-hours-refill—TorqueNews, 2026-10; single use.
    8. 

  8. https://www.wired.com/story/how-tesla-made-its-new-230-mph-car-fly/—WIRED, 2026-10-16; mass, 2 g, fans.
    9. 

  9. https://www.space.com/technology/spacex-thrusters-just-made-a-tesla-roadster-fly-and-elon-musks-car-is-still-out-there-in-space—Space.com, 2026-10-16; 600 L.
    10. 

  10. https://electrek.co/2026/10/15/an-updated-technical-look-at-teslas-kick-flipping-0-60-in-1s-next-gen-roadster/—Electrek, 2026-10-15; timing, display.
    11. 

  11. https://evxl.co/2026/10/16/tesla-roadster-thrusters-compressed-air-tank/—EVXL, 2026-10-16; 5,500 psi.
    12. 

  12. https://teslanorth.com/2026/10/16/tesla-roadster-moravy-hagerty-engineering-interview/—Tesla North, 2026-10-16; hover inputs.
    13. 

  13. https://wooten-labs.com/roadster/—Wooten Labs, n.d.; friction comparison; circular geometry not adopted.
    14. 

  14. https://www.cbsnews.com/news/tesla-roadster-dot-flying-car/—CBS News, 2026-10-16; FAA, NHTSA.
    15. 

  15. https://en.wikipedia.org/wiki/Tesla_Roadster_(second_generation)—Wikipedia, live; 2017 figures.
    16. 

  16. https://web.archive.org/web/20171201060151/tesla.com 2017-12-01; 8.8 s.
    17. 

  17. https://web.archive.org/web/20221017013529/tesla.com 2022-10-17; wheel torque.
    18. 

  18. https://patentsgazette.uspto.gov/week31/OG/html/1537-1/US12377920-20250805.html—USPTO Gazette, 2025-08-05; patent.
    19. 

  19. https://physics.nist.gov/cgi-bin/cuu/Value?gn—NIST CODATA, 2022; g0.
    20. 

  20. https://www.grc.nasa.gov/www/k-12/airplane/rktthsum.html—NASA Glenn, n.d.; thrust equation.
    21. 

  21. https://www1.grc.nasa.gov/beginners-guide-to-aeronautics/lift-equation/—NASA Glenn, n.d.; lift, density.
    22. 

  22. https://www1.grc.nasa.gov/beginners-guide-to-aeronautics/drag-equation/—NASA Glenn, n.d.; drag.
    23. 

  23. https://ntrs.nasa.gov/citations/19930066300—NASA NTRS, 1993-06-01; nitrogen Isp.
    24. 

  24. https://jossonline.com/storage/2014/12/0102-Development-of-A-Modular-Cold-Gas-Propulsion-System-for-Small-Satellite-Applications.pdf—JoSS, 2014; 65 s analogue.
    25. 

  25. https://www.diva-portal.org/smash/get/diva2:75729/FULLTEXT01.pdf—KTH DiVA, n.d.; gas constants, energy.
    26. 

  26. https://webbook.nist.gov/cgi/fluid.cgi?ID=C7727379&Action=Load&Type=IsoTherm&T=300&PLow=1000&PHigh=70000&PInc=34500&Digits=5&UnitSystem=SI—NIST WebBook, 2000 EOS; density.
    27. 

  27. http://alexcortiella.com/aerospace/2020/05/03/F9-booster-recovery/—Cortiella, 2020-05-03; Falcon GN2.
    28. 

  28. https://www.spacex-atlas.com/?model=falcon-cold-gas&series=propulsion—SpaceX Atlas, n.d.; secondary only.
    29. 

  29. https://web.archive.org/web/20151124010749/http:/www.spacex.com/news/2013/04/04/draco-thrusters—SpaceX, 2013-04-04; Draco.
    30. 

  30. https://www.ecfr.gov/current/title-49/subtitle-B/chapter-V/part-571/subpart-B/section-571.308—eCFR, 2025-01-17; hydrogen burst.
    31. 

  31. https://www.ecfr.gov/current/title-49/subtitle-B/chapter-V/part-571/subpart-F/section-571.304—eCFR, current; CNG burst.
    32. 

  32. https://media.stellantisnorthamerica.com/newsrelease.do?id=16546—Stellantis, 2015-07-17; Viper.
    33. 

  33. https://www.fiawec.com/en/news/le-mans-hypercar-explained/6974—FIA WEC, 2020; LMH.
    34. 

  34. https://mcmurtry.com/speirling-pure/—McMurtry, n.d.; 2,000 kgf.
    35. 

  35. https://www.autosport.com/historics/news/how-a-little-electric-storm-thundered-to-goodwoods-hillclimb-record/10329632/—Autosport, 2022; 39.08 s.
    36. 

  36. https://doi.org/10.3390/en15155549—Energies, 2022; suction model.
    37. 

  37. https://en.wikipedia.org/wiki/Tire_load_sensitivity—Wikipedia, Milliken 1995; mu, exponent.
    38. 

  38. https://www.eng.cam.ac.uk/news/dr-hugh-hunt-gives-his-expertise-tv-programme-fifth-gear-stunt-special-show—Cambridge, 2009-05-20; Aygo.
    39. 

  39. https://www.guinnessworldrecords.com/news/2012/7/loop-the-loop-record-set-on-incredible-life-sized-hot-wheels-track-at-x-games-43323—Guinness, 2012-07; Hot Wheels.
    40. 

  40. https://www.espn.com/action/xgames/summer/2012/story/_/id/8107982/tanner-foust-greg-tracy-pull-world-record-hot-wheels-double-loop-dare—ESPN, 2012-06-30; loop geometry.
    41. 

  41. https://www.guinnessworldrecords.com/news/2015/9/new-jaguar-car-and-stunt-pro-terry-grant-defy-gravity-with-largest-loop-the-loop-396097—Guinness, 2015-09; 19.08 m.
    42. 

  42. http://hyperphysics.phy-astr.gsu.edu/hbase/Mechanics/cirvert.html—HyperPhysics, n.d.; loop minima.
    43. 

  43. https://www.guinnessworldrecords.com/world-records/longest-rally-car-ramp-jump—Guinness, 2009-12-31; 81.99 m.
    44. 

  44. https://www.guinnessworldrecords.com/world-records/longest-pickup-truck-ramp-jump—Guinness, 2016-08-25; 115.64 m.
    45. 

  45. https://www.worldrecordacademy.com/stunts/longest_jump_in_a_rally_car_Travis_Pastrana_sets_world_record_101482.htm—World Record Academy, 2010-01-04; 91 mph.
    46. 

  46. https://www.ecfr.gov/current/title-14/chapter-I/subchapter-C/part-25/subpart-C/subject-group-ECFRabacbe77efb6607/section-25.473—eCFR, current; 3.048 m/s.
    47. 

  47. https://www.law.cornell.edu/cfr/text/14/25.723—Cornell LII, current; 3.658 m/s.
    48. 

  48. https://www.govinfo.gov/content/pkg/CFR-2014-title14-vol1/pdf/CFR-2014-title14-vol1-sec23-473.pdf—GovInfo, 2014; Part 23.
    49. 

  49. https://masonmotorsports.com/2wd-trophy-truck/—Mason Motorsports, n.d.; wheel travel.
    50. 

  50. https://www.jimcoracing.com/pages/spec-trophy-truck-class-6100—Jimco Racing, n.d.; wheel travel.
    51. 

  51. https://toyotagazooracing.com/wrc/cars/2025/—Toyota Gazoo Racing, 2025; 270 mm.
    52. 

  52. https://unece.org/fileadmin/DAM/trans/danger/publi/manual/Manual%20Rev5%20Section%2038-3.pdf—UNECE, Rev.5; shock pulses.
    53. 

  53. https://www.chinaautoregs.com/archives/6981—China Auto Regs, 2025; GB criteria.
    54. 

  54. http://120.79.24.31/mydata/law/202504/25/17455515805353.htm—GB 38031-2025 text, 2025-03-28; 7 g pulse.
    55. 

  55. https://www.normsplash.com/Samples/UL/155112932/UL-2580-2021-en.pdf—UL 2580 sample, 2021; no limit.
    56. 

  56. https://doi.org/10.18226/23185279.v6iss1p29—vertical-acceleration paper, n.d.; no rally peak.
    57. 

  57. https://doi.org/10.4271/900760—SAE, 1990; wheel inertia.
    58. 

  58. https://mdpi-res.com/d_attachment/wevj/wevj-15-00266/article_deploy/wevj-15-00266.pdf—MDPI WEVJ, 2024-06-17; pitch inertia.
    59. 

  59. https://image-ppubs.uspto.gov/dirsearch-public/print/downloadPdf/10668959—USPTO, n.d.; attitude channel.
    60. 

  60. https://link.springer.com/article/10.1186/s40323-019-0126-y—Springer AMSES, 2019; contact model.
    61. 

  61. https://api.projectchrono.org/development/simulation_system.html—Project Chrono, 2016-2026; contact families.
    62. 

  62. https://beamngpy.readthedocs.io/en/v1.35.1/beamngpy.html—BeamNG, v1.35.1; integrator.
    63. 

  63. https://www.media.stellantis.com/em-en/opel/press/90-years-ago-opel-sounds-in-the-era-of-rockets—Stellantis, 2018; Opel RAK.
    64. 

  64. https://www.rimac-newsroom.com/press-releases/rimac-automobili/rimac-nevera-sets-23-performance-records-in-a-single-day—Rimac, 2023-05-17; Nevera.
    65. 

  65. https://web.archive.org/web/20141023191016/24h-lemans.com 2012-05-25; CLR liftoff.
    66. 

  66. https://saemobilus.sae.org/articles/closed-wheel-race-vehicle-aerodynamic-lift-off-2017-01-1516—SAE, 2017-03-28; liftoff.
    67. 

  67. https://web.archive.org/web/20010223123951/lemans.org 2000-02-18; rule change.
    68. 

  68. https://www.mercedesamgf1.com/news/g-force-and-formula-one-explained—Mercedes-AMG F1, 2024; over 5 g.
    69. 

  69. https://thejns.org/spine/view/journals/j-neurosurg-spine/25/6/article-p756.xml—J Neurosurg Spine, 2016; HANS.
    70. 

  70. https://medcoe.army.mil/pfw-images/borden/harshenv2/HE2ch33.pdf—U.S. Army, 2002; +Gz.
    71. 

  71. https://physoc.onlinelibrary.wiley.com/doi/full/10.1113/EP089712—Physiology review, 2022; G-LOC.
    72. 

  72. https://api.fia.com/news/fia-concludes-investigation-romain-grosjeans-accident-2020-bahrain-formula-1-grand-prix-and—FIA, 2021-03-05; 67 g.
    73.