Tyre model calculator: Pacejka Magic Formula and brush model
Enter Magic Formula coefficients to get the tyre force at any slip, the peak and the slip where it happens, the cornering or slip stiffness BCD and the sliding force. You can also work out B, C, D and E from four numbers read off a test curve, or use the brush model to get the lateral force, aligning moment, pneumatic trail and how much of the contact patch is sliding. Free, runs in your browser.
Pacejka Magic Formula · MathWorks road-surface coefficients · brush model checked by numerical integration · Updated 22 Nov 2026
Units follow x: with x in degrees, B is per degree and BCD is N/deg. MathWorks writes F = Fz·D·sin(…) with D as a friction coefficient; here D is in newtons, so enter μ × Fz.
Read the four numbers off a measured curve: the initial slope, the peak, where it peaks and the level the force settles to. The example is the problem-set lateral curve measured numerically; the fit returns its B = 0.27, C = 1.2, E = −1.6.
If you have the bristle stiffness per unit length k (N/m², "ky" in the problem sets) instead, Cα = k l² / 2.
- Calculating…
MATLAB / Octave code
B = 10; C = 1.9; D = 4000; E = 0.97;
What is the Pacejka Magic Formula?
The Magic Formula is a curve fit for tyre force against slip, introduced by Bakker, Nyborg and Pacejka in 1987. It has no physics inside. It is a sine of an arctangent, chosen because that shape matches measured tyre curves: a straight start, a rounded peak and a slow fall towards sliding.
Here x is the slip: the slip ratio κ for braking and driving force Fx, or the slip angle α for cornering force Fy. Two shifts move the curve off the origin, to account for things like ply steer and rolling resistance:
Each coefficient controls one feature you can see on the curve:
- D, the peak factor, is the highest force the tyre makes. Divided by the vertical load it is the peak friction coefficient.
- C, the shape factor, sets how far the force falls after the peak. C = 1 never peaks; C = 2 falls all the way to zero (sin π = 0). Real tyres sit in between.
- B, the stiffness factor, scales the slip axis. The product BCD is the slope at zero slip: the cornering stiffness for Fy, the longitudinal slip stiffness for Fx.
- E, the curvature factor, moves the peak and changes how sharp it is. It must not exceed 1. Above 1 the curve turns back on itself at large slip.
The figure keeps B = 10, C = 1.9 and D = 4,000 N and changes only E. All four curves leave the origin with the same slope (BCD = 76,000 N per unit slip) and would settle to the same sliding value far out. E moves the peak from about 9% slip (E = −1) through 11% (E = 0) to 18% (E = 0.97), and decides how quickly the force drops after it.
What are typical Magic Formula coefficients?
MathWorks' Simscape Driveline documentation (Tire-Road Interaction (Magic Formula) block) lists typical constant coefficients for four road surfaces. In its form F = Fz·D·sin(…) the D column is the peak friction coefficient. The right-hand columns are what those coefficients give for a 4,000 N wheel load, worked out by the calculator:
| Surface | B | C | D | E | Peak at slip | Peak Fx | Locked wheel (κ = 1) |
|---|---|---|---|---|---|---|---|
| Dry tarmac | 10 | 1.9 | 1 | 0.97 | 18% | 4000 N | 3658 N (91%) |
| Wet tarmac | 12 | 2.3 | 0.82 | 1 | 8.82% | 3280 N | 2549 N (78%) |
| Snow | 5 | 2 | 0.3 | 1 | 31.1% | 1200 N | 1142 N (95%) |
| Ice | 4 | 2 | 0.1 | 1 | 38.9% | 400 N | 384.6 N (96%) |
Two things stand out. The peak on wet tarmac comes early, at 8.82% slip, while on snow and ice it comes late, past 30%. And a locked wheel gives 91% of the peak force on dry tarmac and 78% on wet. ABS keeps the slip near the peak, which recovers that gap and keeps the tyre able to steer.
Watch the units when you move between sources. MathWorks multiplies by Fz, so its D has no units. Most textbooks and course problem sets put the newtons in D. The problem set used in Skill-Lync's vehicle dynamics challenges gives lateral coefficients with the slip angle in degrees (B = 0.27, C = 1.2, D = 2,900 N, E = −1.6) and longitudinal ones against slip ratio (B = 25, C = 1.15, D = 3,200 N, E = −0.4). The lateral curve peaks at α = 7.9° with a cornering stiffness of 939.6 N/deg. The longitudinal one peaks at 15.3% slip. Load either one from the Example menu.
How do I find B, C, D and E from test data?
Four features of a measured curve fix the four coefficients: the initial slope K, the peak D, the slip xm where the peak happens and the force y∞ the curve settles to at large slip. Rearranging the formula gives:
The formulas assume the curve has a peak (C > 1) and that E < 1, so the curve settles to D sin(πC/2). A sliding value below zero means C > 2. The verifier tested the round trip on six curves: it measured each curve's slope, peak and sliding value numerically, rebuilt B, C and E from them, and recovered the originals to 6 significant figures. For the problem-set lateral curve the inputs are K = 939.6 N/deg, D = 2,900 N at xm = 7.897° and y∞ = 2758 N.
On real test data this gives a starting point, not the final fit. Measured curves are noisy and rarely reach a clean y∞. A least-squares fit (MATLAB's lsqcurvefit, or Python's scipy.optimize.curve_fit) started from these four-point values converges far more reliably than one started from guesses. One learner's report fitted the aligning moment this way to an RMS error of 1.86 N·m, under the problem set's 2 N·m target.
What is the brush tyre model?
The brush model is the physical model behind the curve. Think of the tread as a row of bristles. A bristle touches the road at the front of the contact patch with no deflection. At a slip angle α the wheel moves sideways relative to the road, so as the bristle travels back through the patch it is dragged further sideways, by tan α for every metre it travels. Its force per unit length grows in a straight line from the leading edge, until it reaches the friction limit μq(x) set by the local contact pressure q(x). From there it slides along the limit.
The shaded area is the lateral force. The bristle stiffness per unit length c and the contact half-length a set the cornering stiffness:
With a parabolic pressure distribution (zero at both edges, highest in the middle) the integral gives the Fiala formula. It reaches full sliding, Fy = μFz, at tan α = 3μFz/Cα:
With uniform pressure the rear of the patch starts sliding as soon as Cα tan α passes half of μFz. After that the force creeps towards μFz but never quite reaches it:
The verifier does not use any of these formulas. It integrates the bristle forces over 400,001 points along the patch and compares the result with the formulas above for 14 cases. Fy and Mz agree to 5 significant figures.
Worked example: the problem-set brush tyre
Problem set 2b in Skill-Lync's vehicle dynamics challenge gives Fz = 3,000 N, μ = 0.8, a contact length of 0.2 m and a lateral bristle stiffness of 2.275 × 10⁶ N/m², and asks for the effective cornering stiffness at a 3° slip angle. The course model uses uniform pressure. The cornering stiffness is Cα = k l²/2 = 45,500 N/rad. Cα tan 3° = 2384.6 N is more than half of μFz (1,200 N), so the rear of the patch is already sliding:
That is the 598.7 N/deg one learner's report gives. The linear cornering stiffness is 794.1 N/deg, so at only 3° the tyre has already lost 25% of its stiffness. Another report on the same problem got negative values because it passed 3 (degrees) straight into tan(). With parabolic pressure the same tyre gives 560.7 N/deg:
Which pressure shape should you use? Parabolic pressure is closer to a real tyre, whose pressure drops to zero at the edges of the patch, and it gives a clean full-sliding angle. Uniform pressure is simpler and is what the course problem sets use. For the lateral brush report's tyre (Fz = 5,000 N, μ = 0.9, Cα = 50,000 N/rad, parabolic) full sliding comes at α = 15.1°. At 4° it makes 2669 N with 74% of the patch still gripping; at 10° it makes 4312 N with 35% gripping.
Aligning moment and pneumatic trail
The bristle forces are biggest towards the back of the patch, so the lateral force acts behind the centre of the contact patch. That distance is the pneumatic trail, and Fy times the trail is the self-aligning moment Mz, which tries to turn the wheel back towards zero slip. At small slip angles the trail is a/3 for both pressure shapes, so 33 mm for a 0.2 m patch. As the back of the patch starts sliding, the force there is capped and the resultant moves forward. The trail shrinks, Mz peaks and then falls, while Fy is still rising.
For the report tyre, Mz peaks at λ = 1/4 (where d/dλ of λ(1 − λ)³ is zero), at α = 3.86°. By then Fy has only reached 58% of μFz. That early drop in steering torque is how drivers feel a car approaching the limit. Switch the brush tab to see the Mz curve next to Fy.
Braking and cornering together: the friction ellipse
A tyre that is braking hard has less grip left for cornering. The simplest way to model this caps the combined force inside an ellipse:
A tyre with 4,000 N of peak grip in each direction that is using 3,000 N to brake has at most 2646 N left for cornering. Enter the force used in the other direction in the Magic Formula tab to apply this. It is an approximation. Full combined-slip models (Pacejka's weighting functions, or the combined brush model with σx = κ/(1+κ) and σy = tan α/(1+κ)) predict how the force builds up along the way, not only the limit.
Tyre load sensitivity
Tyre friction is not constant. The peak friction coefficient falls as the vertical load rises, so a tyre with twice the load makes less than twice the force. Pacejka's full model builds this in by making D and BCD functions of the load change dfz = (Fz − Fz0)/Fz0. MathWorks' load-dependent block uses the same terms.
Here is why it matters in a corner. Take two tyres on one axle at 4,000 N each, and assume the friction coefficient falls by 0.05 per 1,000 N from 1.0 at 4,000 N (an assumed, round number, not a measured tyre). Evenly loaded, the axle can make 8000 N. In a corner with 1,500 N of lateral load transfer the outer tyre carries 5,500 N at μ = 0.925 and makes 5088 N. The inner one carries 2,500 N at μ = 1.075 and makes 2688 N. Together that is 7775 N, 2.8% less than before.
The axle that takes more of the load transfer loses more grip. That is why a stiffer front anti-roll bar adds understeer and a stiffer rear one adds oversteer. See the ride frequency calculator for the spring and roll-stiffness side.
What this calculator does not model
- Camber, inflation pressure and speed. Full Magic Formula models such as MF 6.2 add camber thrust, pressure terms and speed effects with dozens of coefficients fitted to rig data. This calculator uses the four-coefficient formula with constant B, C, D and E.
- Transient behaviour. A real tyre needs to roll about one relaxation length before its force builds up after a slip change. All results here are steady-state.
- Combined slip in detail. The friction ellipse limits the total force. It does not predict how Fx and Fy share the grip at partial slip.
- Brush model simplifications. The model assumes a rigid carcass, one friction coefficient for sticking and sliding, and pure side slip. Real carcass flexibility lowers the cornering stiffness and moves the sliding point.
Skill-Lync's vehicle dynamics course has you code these tyre models yourself in MATLAB/Octave: Pacejka curves against slip angle and slip ratio, the combined brush model and a Magic Formula fit to test data. Then you use them in bicycle, ride and handling models.
Frequently asked questions
What is the Pacejka Magic Formula?
An empirical curve for tyre force against slip: y = D sin(C arctan(Bx − E(Bx − arctan Bx))). It is called "magic" because one formula with four coefficients fits measured longitudinal force, lateral force and aligning moment curves well. It is the tyre model in MATLAB/Simscape, Adams and most vehicle-dynamics codes.
What do B, C, D and E mean in the Magic Formula?
D is the peak force. C is the shape factor: it sets how far the force falls after the peak (the sliding value is D sin(πC/2) when E < 1). B is the stiffness factor: the slope at zero slip is B × C × D. E is the curvature factor: it moves the peak and sharpens or flattens it, and must stay at or below 1.
How do I get cornering stiffness from the Magic Formula?
It is the slope at zero slip, BCD. For the problem-set lateral coefficients (B = 0.27 per degree, C = 1.2, D = 2,900 N) that is 939.6 N/deg, or 53,840 N/rad. With MathWorks' dry tarmac coefficients at 4,000 N the longitudinal slip stiffness is 76,000 N per unit slip ratio.
At what slip ratio is braking force highest?
With MathWorks' typical Magic Formula coefficients the peak is at 18% slip on dry tarmac, 8.82% on wet, 31.1% on snow and 38.9% on ice. A locked wheel (100% slip) gives 91.5% of the peak on dry tarmac, and 78% on wet. ABS holds the wheel near the peak, which recovers that force and keeps the tyre able to steer.
What is the brush tyre model?
It pictures the tread as bristles. Each bristle sticks to the road at the leading edge and is dragged sideways as it moves back through the contact patch, until its force reaches the local friction limit μq(x) and it slides. The adhesion part gives the linear cornering stiffness; the sliding part bends the curve over. With parabolic pressure it gives the Fiala formula and full sliding at tan α = 3μFz/Cα.
What is pneumatic trail?
The distance the lateral force acts behind the centre of the contact patch. It produces the aligning moment Mz = Fy × trail. In the brush model it is a/3 (a sixth of the contact length) at small slip and shrinks as the rear of the patch starts sliding, so Mz peaks long before Fy does: drivers feel that as the steering going light near the limit.
What is tyre load sensitivity?
The friction coefficient of a tyre falls as its vertical load rises, so doubling the load less than doubles the grip. In a corner, load transfer moves weight onto the outer tyre; with friction falling 5% per 1,000 N (an assumed example), an axle with 1,500 N of load transfer has 2.8% less grip than the same axle evenly loaded. Roll stiffness distribution uses this to tune understeer and oversteer.
Related on Skill-Lync
- Vehicle dynamics project ideas (the tyre-model problem sets, with learners' reports and common mistakes)
- Learner report: Pacejka Magic Formula and the combined slip brush tyre model
- Learner report: longitudinal and lateral force with the Magic Formula in MATLAB
- Steering geometry calculator (Ackermann, turning radius, steering ratio)
- Ride frequency calculator (spring rates, motion ratio, roll stiffness)
- Gradeability calculator (traction limit on a slope)
- Regenerative braking energy calculator
- All free engineering tools
Sources
- MathWorks, Simscape Driveline documentation, "Tire-Road Interaction (Magic Formula)" (R2026b page, read 22 Nov 2026): Fx = Fz·D·sin(C·arctan{Bκ − E[Bκ − arctan(Bκ)]}), slope BCD·Fz at κ = 0, and the table "Magic Formula Coefficients for Typical Road Conditions" (dry tarmac 10 / 1.9 / 1 / 0.97, wet 12 / 2.3 / 0.82 / 1, snow 5 / 2 / 0.3 / 1, ice 4 / 2 / 0.1 / 1).
- E. Bakker, L. Nyborg, H. B. Pacejka, "Tyre modelling for use in vehicle dynamics studies", SAE 870421 (1987).
- H. B. Pacejka, Tire and Vehicle Dynamics, 3rd ed., Butterworth-Heinemann (2012): chapter 3 (brush model) and chapter 4 (Magic Formula).
- E. Fiala, "Seitenkräfte am rollenden Luftreifen", VDI-Zeitschrift 96 (1954): the brush lateral force with parabolic pressure.
- Skill-Lync vehicle dynamics problem sets, as quoted in two or more learners' public reports (see the project ideas page): Magic Formula coefficients, brush inputs for problem sets 2a and 2b, and the 598.70 N/deg answer.
- Skill-Lync, scripts/verify/tyre.py (NumPy): 36 checks. Magic Formula peaks by golden-section search, slopes by central differences, sliding values far out, the four-point fit round trip on 6 curves, and the brush model by numerical integration over the contact patch (14 cases, trail a/3, full-sliding angle). The page recomputes every one at build time.