# Load sensitivity: why load transfer costs grip

> Tyre friction coefficient falls as vertical load rises, so an axle with unequal wheel loads makes less total force than the same axle with equal loads. This is the root of LLTD balance tuning and of why downforce gives less than proportional grip.

Area: Tyres · Starting level: L1 Foundation · Updated 2026-11-16
Source page: https://race-engineer-academy.vercel.app/lesson/tyres-load-sensitivity

## L1 · Foundation

Double the load on a tyre and its maximum lateral force **less than doubles**. The friction coefficient $\mu = F_{y,max}/F_z$ falls with load; this is **tyre load sensitivity**. A simple linear model is

$$\mu(F_z) = \mu_0 - k\,F_z$$

**Worked example (axle pair).** Take $\mu_0 = 1.8$, $k = 5\times10^{-5}\,\text{N}^{-1}$, so $\mu = 1.6$ at 4 kN, 1.5 at 6 kN, 1.7 at 2 kN.

| Case | Outside $F_z$ | Inside $F_z$ | Outside $F_y$ | Inside $F_y$ | Axle $F_y$ |
|---|---|---|---|---|---|
| No transfer | 4000 N | 4000 N | 6400 N | 6400 N | **12 800 N** |
| 2000 N transfer | 6000 N | 2000 N | 9000 N | 3400 N | **12 400 N** (−3.1%) |
| 3000 N transfer | 7000 N | 1000 N | 10 150 N | 1750 N | **11 900 N** (−7.0%) |

The outside tyre gains less than the inside tyre loses. The loss grows with the **square** of the transfer: 1.5× the transfer gave about 2.25× the loss.

Consequences you will use all the time:
- **Balance:** the axle that takes the larger share of lateral load transfer loses more grip relative to the other. More front roll stiffness → more front load transfer → more understeer. This is the mechanism behind every anti-roll-bar change.
- **Low CG and wide track** reduce total load transfer, so they raise total grip.
- **Downforce:** adding load raises grip, but μ falls as load rises, so the cornering g does not rise in proportion to downforce. **Mass** works the same way: a heavier car corners at lower g on the same tyres.
- If μ were independent of load, lateral load transfer would not cost grip at all (apart from camber and geometry effects). That is worth remembering when someone claims otherwise.

## L2 · Practitioner

**Trackside implications.**

1. **Anti-roll bars do not reduce total weight transfer**; they move it between axles. The total is fixed by $m\,a_y\,h/t$. Stiffening the front bar moves load transfer to the front, the front pair loses grip through load sensitivity, and the car understeers more.
2. **Size of the effect.** With the numbers above, moving about 5% of LLTD from rear to front on a GT3 at 1.6 g changes front axle capacity by about −1.7% and rear by about +1.3%. That is a balance swing equivalent to a significant aero balance change, which is why a couple of ARB clicks are felt.
3. **Track width and CG height** are "free grip" for the designer, not usually for you. But **ride height** changes CG height a little (a few mm out of 450 mm is under 1%), so lowering for mechanical grip mostly acts through aero and roll centre, not CG.
4. **Ballast placement** (BoP ballast in GT3, success ballast) raises mass and, if mounted high, CG. Teams fight to place it low and central.
5. **Tyre pressure** changes load sensitivity: higher pressure generally increases load sensitivity (μ falls faster with load), so an axle that already sees large load transfer suffers more from over-pressure.

**What load sensitivity is *not*:** it is not the inside wheel "lifting". A car can lose grip through load sensitivity well before any wheel unloads. It is also not only about lateral transfer: longitudinal transfer under braking unloads the rear pair together. Both rear tyres lose capacity, and the front pair, now unequal in braking-plus-cornering, also suffers.

**Measuring it:** you rarely get the coefficient from data directly. Use the supplier tyre model or flat-track data. In car data, you can compare max lateral G in low- vs high-speed corners against estimated vertical load; $a_y$ vs total load flattening out is load sensitivity plus aero balance effects mixed together.

## L3 · Advanced

**Quadratic loss derivation.** With $F_y = \mu_0 F_z - k F_z^2$ per tyre and wheel loads $F_{z0} \pm \Delta F$:

$$F_{y,pair} = 2\mu_0 F_{z0} - k\big[(F_{z0}+\Delta F)^2 + (F_{z0}-\Delta F)^2\big] = F_{y,pair}(0) - 2k\,\Delta F^2$$

The loss is independent of static load and quadratic in transfer. Check with the table: $2 \times 5\times10^{-5} \times 2000^2 = 400$ N. ✓

**Balance sensitivity.** Let total lateral load transfer be $\Delta W$ and front share $\chi$ (LLTD), with $\Delta F_f = \chi \Delta W$, $\Delta F_r = (1-\chi)\Delta W$ (per side). Then

$$\frac{\partial F_{y,f}}{\partial \chi} = -4k\,\chi\,\Delta W^2, \qquad \frac{\partial F_{y,r}}{\partial \chi} = +4k\,(1-\chi)\,\Delta W^2$$

For a GT3 at 1.6 g with $\Delta W \approx 5.6$ kN and $\chi \approx 0.5$, each 1% of LLTD costs the front pair ~30 N and gives the rear ~30 N (about 0.3% of each axle's capacity, so a 5% LLTD change is a ~3% relative balance swing). The effect is strongly $a_y$ dependent ($\Delta W^2 \propto a_y^2$). A bar change felt strongly in a 1.8 g corner is far weaker in a 1.0 g corner.

**Real tyres are not linear in $\mu(F_z)$.** Magic Formula uses $D = (p_{Dy1} + p_{Dy2}\,df_z)\,F_z$ where $df_z = (F_z - F_{z0})/F_{z0}$, which is the same linear-μ structure about a nominal load. Cornering stiffness has its own load dependence: it rises to a maximum at high load and then falls ($C_\alpha \propto \sin(2\arctan(F_z/p_{Ky2}F_{z0}))$ in MF). That matters for **linear-range** balance (understeer gradient) and can make the sign of an ARB effect at low $a_y$ different from at the limit.

**Aero interaction.** With downforce, static $F_{z0}$ grows with $V^2$, and $\Delta W$ grows with the total lateral force, which grows with it. The pair loss depends only on $\Delta F$ and $k$, so in newtons it scales with $a_y^2$; relative to axle capacity (proportional to $m\,a_y$ at the limit) it scales with $a_y$. An ARB change therefore moves *more* balance in the highest-g corners, which on a downforce car are the fast ones. Mechanical tools are nevertheless the main balance lever in slow corners because there the aero-balance term, which dominates at speed, is small. At speed, roll-stiffness changes also move the aero platform, which mixes the two effects.

## L4 · Expert

**Where the simple model breaks.**
- **Camber and inclination under load.** The outside tyre's effective camber changes with load and roll, while the inside tyre's camber goes further negative (roll minus the droop camber recovery), so it runs on its inner shoulder. Part of what engineers call "load-transfer loss" is really camber loss on the inside tyre. Separating them requires a tyre model with $\gamma$ and $F_z$ coupling.
- **Thermal load sensitivity.** Heavily loaded outside tyres run hotter. If the outside front is above window (Segers §8.6 example: the loaded front tyre spikes out of range while the inside stays cold), the effective $k$ is much larger than the isothermal value. A softer front ARB then helps twice: less transfer, and a cooler outside tyre.
- **Very light inside loads.** Near zero load, μ can be high but force is tiny, and relaxation length and camber dominate. Inside-wheel lift (TCR rear, GT4 inside front on kerbs) makes the pair term saturate; further roll stiffness then changes nothing on that axle and only acts on the other.
- **Transient.** Load sensitivity applies to instantaneous loads. Damper-driven transient LLTD (higher front bump/rebound → more front transfer during roll-in) creates a transient balance different from steady state.
- **Model identification.** In sim correlation, an error in $k$ shows up as balance error that scales with $a_y^2$. If your sim's balance error is speed independent but $a_y$ dependent, suspect load sensitivity before aero maps.

## Key formulas

**Linear load-sensitive friction**

$$\mu(F_z) = \mu_0 - k\,F_z$$

- \mu_0: friction coefficient extrapolated to zero load
- k: load sensitivity (1/N)
- F_z: vertical load (N)

**Axle force loss from load transfer**

$$\Delta F_{y,pair} = -2k\,\Delta F_z^2$$

- \Delta F_z: load transferred from inside to outside wheel (N)

**Balance sensitivity to LLTD**

$$\dfrac{\partial F_{y,f}}{\partial \chi} = -4k\,\chi\,\Delta W^2$$

- \chi: front lateral load transfer distribution
- \Delta W: total lateral load transfer (N)

## Trackside takeaways

- μ falls with load, so the outside tyre gains less than the inside loses.
- Axle grip loss is quadratic in load transfer: 1.5× the transfer gives ~2.25× the loss.
- ARBs and springs redistribute total load transfer between axles; they do not reduce it.
- The axle carrying more load transfer loses relative grip. That is why front roll stiffness gives understeer.
- The LLTD effect on balance grows with a_y (∝ a_y² in newtons, ∝ a_y relative to axle grip): judge ARB changes in the highest-g corners. Slow corners are "mechanical" because the aero-balance term is small there.

## By car category

### GT3

At ~1300 kg with h around 0.45 m, total lateral transfer at 1.6 g is ~5.5 kN. A 5% LLTD swing takes ~150 N of capacity off one axle and gives ~150 N to the other, enough to change entry/mid balance clearly. BoP ballast adds mass, so it also costs cornering g through load sensitivity.

### TCR

TCR runs very high rear roll stiffness so the rear pair absorbs most of the load transfer (inside rear often lifts). That keeps the front pair more evenly loaded and protects front grip and traction.

### FIA F3

Light car, low CG, wide track: load transfer as a fraction of weight is moderate. The tyres are run near peak load sensitivity at high downforce, so aero load beyond a point gives diminishing lateral G.

### LMP2

Heavy downforce dominates wheel loads at speed. In slow corners the mechanical LLTD is the main balance tool, and in fast corners it is aero balance. Engineers split balance work by corner speed for this reason.

### IndyCar

On ovals, asymmetric setups (stagger, cross-weight, weight jacker) deliberately set unequal static loads so the dynamic loads come out more equal in the turn. That uses load sensitivity in your favour.

## References

- Milliken & Milliken, Race Car Vehicle Dynamics (SAE 1995): Ch. 2 Tire Behavior (Load sensitivity)
- Milliken & Milliken, Race Car Vehicle Dynamics (SAE 1995): Ch. 7 Steady-State Pair Analysis (Pair force vs load transfer)
- Milliken & Milliken, Race Car Vehicle Dynamics (SAE 1995): Ch. 18 Wheel Loads
- Segers, Analysis Techniques for Racecar Data Acquisition, 2nd ed. (SAE 2014): §8.6 Lateral Load Transfer and Tire Temperature
