# Banking, grade and vertical curvature effects

> Banking, uphill/downhill grade and crests/compressions change both the normal load on the tyres and what the accelerometers read. Correct for them before comparing grip, balance or load-transfer data between corners or tracks.

Area: Weight Transfer & Balance · Starting level: L2 Practitioner · Updated 2026-11-16
Source page: https://race-engineer-academy.vercel.app/lesson/wt-banking-grade-effects

## L1 · Foundation

On a flat corner the tyres must supply all of the centripetal force, and gravity is perpendicular to the road. On a **banked** corner (angle $\alpha$) gravity has a component that helps push the car towards the centre, and the centripetal acceleration has a component *into* the road surface. Both help:

$$V^2 = gR\,\frac{\mu + \tan\alpha}{1 - \mu\tan\alpha}$$

**Example:** $R = 200$ m, $\mu = 1.6$ (no aero). Flat: $V = \sqrt{1.6 \times 9.81 \times 200} = 56$ m/s (202 km/h). Banked 11°: $\tan\alpha = 0.194$, so $V^2 = 9.81 \times 200 \times 1.794/0.689 = 5110$ and $V = 71.5$ m/s (**257 km/h**). Banking is worth a lot.

**Grade** (uphill/downhill slope $\theta$) does two things:
- reduces the normal load slightly ($mg\cos\theta$; 5° costs only 0.4%);
- adds a gravity component along the road ($mg\sin\theta$; 5° ≈ 0.087 g), which slows the car uphill and pushes it downhill, and **shifts load** between axles like a longitudinal acceleration: uphill loads the rear, downhill loads the front.

**Vertical curvature** (crests and compressions) changes normal load by $m V^2/R_v$:
- A **compression** (valley) adds load: more grip, lower ride heights, risk of bottoming.
- A **crest** removes load: less grip and the car goes light. At 200 km/h over a crest with a 600 m vertical radius, $V^2/R_v = 0.52$ g, so the tyres lose half the static load (before aero).

Corners on crests, braking zones downhill, and compressions at fast corners are where tracks become difficult. Know them from the track survey or GPS elevation.

## L2 · Practitioner

**What the accelerometer sees.** A chassis-mounted accelerometer measures acceleration *plus* the gravity component along each axis (specific force). On a banked road with the car roughly parallel to the surface:

$$a_{y,meas} = \frac{V^2}{R}\cos\alpha - g\sin\alpha, \qquad a_{z,meas} = g\cos\alpha + \frac{V^2}{R}\sin\alpha$$

The measured lateral channel is what the **tyres** must actually produce, which is less than $V^2/R$. The vertical channel reads above 1 g in banking.

**Segers §10.3** works through an IndyCar at Phoenix (11° banking) with a measured lateral of 3.13 G. Segers divides by $\cos\alpha$ to get 3.19 G. The full specific-force relation above also adds back $g\sin\alpha$: $a_c = (3.13 + 0.19)/0.982 \approx 3.38$ g of path acceleration. Know which convention your channel uses: "path" lateral ($V^2/R$, from speed × yaw rate or GPS) vs "tyre" lateral (chassis accelerometer). Mixing them when comparing a banked corner with a flat one gives misleading grip numbers.

**Practical checks:**
- Compute **$V \cdot r$** (speed × yaw rate) and compare with the lateral accelerometer. The difference in steady corners reveals banking (and accelerometer offsets or roll misalignment).
- Longitudinally: $a_{x,meas} - \dot V \approx g\sin\theta$, so the difference between the accelerometer and the derivative of wheel speed estimates **grade** (Segers' slope equation). Integrated over distance it gives an elevation profile you can sanity-check against the track map.
- **Ride height and load-cell traces** jump in compressions without any change in speed or G. Without grade or vertical-curvature correction you may blame aero or springs.

**Setup implications:**
- Compressions (Eau Rouge-type) need heave stiffness or bump-rubber margin. Bottoming there is often the limiting ride-height constraint for the whole lap.
- Crests at turn-in (e.g. blind braking zones) demand stability on a light rear axle: brake bias, coast diff and rear damping.
- Downhill braking zones increase front load share and lock-up risk on the inside front.

## L3 · Advanced

**Full force balance on a banked turn** (tyre forces in the road plane, $a_c = V^2/R$ horizontal):
- Normal force: $N = m(g\cos\alpha + a_c\sin\alpha)$.
- Required lateral (road-plane) force: $F_y = m(a_c\cos\alpha - g\sin\alpha)$.
- Lateral load transfer is driven by $F_y$ (road-plane), so in banking the load transfer per unit path acceleration is **reduced** while total normal load is **increased**. Both help grip.

Example, Segers' Phoenix point (measured lateral 3.13 g at $\alpha = 11°$): the accelerometer reads the road-plane demand directly, $F_y/m = 3.13$ g. The path acceleration is $a_c = (3.13 + 0.191)/0.982 \approx 3.38$ g, and $N/m = 0.982 + 3.38 \times 0.191 \approx 1.63$ g (before aero). The required friction coefficient is $3.13/1.63 \approx 1.9$ before any aero load, which is why oval cars depend on downforce to raise the denominator. Note that Segers' simpler correction divides the measured lateral by $\cos\alpha$ only; the full specific-force relation includes the $g\sin\alpha$ term. Check which your logger's channel implements.

**Grade:** axle loads with slope $\theta$ and longitudinal acceleration $a_x$:

$$W_f = mg\cos\theta\,\frac{b}{L} - m\,(a_x + g\sin\theta)\,\frac{h}{L}$$

(uphill $\theta > 0$). The accelerometer reads $a_x + g\sin\theta$ directly, so a load-transfer math channel built on the **accelerometer** already includes grade transfer, while one built on $dV/dt$ doesn't.

**Vertical curvature:** $\Delta N = m V^2/R_v$ (compression positive). The accelerometer's vertical channel captures it. Aero load also changes through the ride-height change it causes: a compression lowers the car and may stall a diffuser, while a crest raises it and reduces downforce exactly where the mechanical load is lowest.

## L4 · Expert

**Edge cases engineers get wrong.**
- **Oval setups** exploit banking with asymmetric static loads (stagger, cross-weight, left-side weight bias) because the load-transfer and normal-load structure differs from a flat road course. Translating road-course LLTD intuition directly to a banked oval misleads.
- **Camber relative to the road** is unchanged by banking, but the car's roll attitude relative to gravity changes. Inclinometer-based roll channels must be referenced to the road, not to gravity.
- **Sensor alignment.** A 0.5° accelerometer misalignment in roll injects ~0.009 g per g of vertical acceleration into lateral, and the vertical acceleration is up to 1.6 g on a banked oval. Calibrate offsets on a level surface and check symmetry on the straight.
- **Simulation tracks** must include 3-D geometry (banking, grade, camber of the road, vertical curvature). A flat-track QSS lap sim will be wrong by seconds at Spa, Laguna Seca, COTA T1 or any oval. Correlation discrepancies concentrated at such places are track-model errors, not car-model ones.
- **Load transfer math channels** based on accelerometers inherently include gravity components; channels based on $V \cdot r$ don't. When building Segers-style total wheel-load channels, decide deliberately which reference you use and add the missing terms explicitly.

## Key formulas

**Max speed on a banked corner (no aero)**

$$V^2 = gR\,\dfrac{\mu + \tan\alpha}{1 - \mu\tan\alpha}$$

- \alpha: banking angle
- R: radius (m)
- \mu: friction coefficient

**Accelerometer readings on banking**

$$a_{y,meas} = a_c\cos\alpha - g\sin\alpha, \quad a_{z,meas} = g\cos\alpha + a_c\sin\alpha$$

- a_c: path centripetal acceleration V²/R (m/s²)

**Grade estimate from data**

$$\sin\theta \approx \dfrac{a_{x,meas} - \dot V}{g}$$

- \theta: road grade (uphill positive)
- \dot V: derivative of speed (m/s²)

**Vertical curvature load change**

$$\Delta N = \dfrac{m V^2}{R_v}$$

- R_v: vertical radius (m), compression positive

## Trackside takeaways

- Banking adds normal load and gravity help: 11° on a 200 m corner raises the no-aero limit from ~202 to ~257 km/h at μ = 1.6.
- The lateral accelerometer measures tyre demand (a_c·cosα − g·sinα), not V²/R. Know which you are plotting.
- Accelerometer minus dV/dt gives the grade; integrate it for an elevation sanity check.
- Crests unload the car (m·V²/R_v); compressions load it and set ride-height/bottoming limits.
- Correct for banking and grade before comparing grip or load-transfer stats across corners or tracks.

## By car category

### IndyCar

Ovals are the extreme case: banking up to ~20°+ on some tracks, and sustained vertical loads well above 1 g. Asymmetric setups (stagger, cross-weight, weight jacker) and accelerometer corrections are fundamental. Segers’ Phoenix example uses an IndyCar.

### GT3

Road courses with big elevation (Spa, Bathurst, Laguna Seca, Nordschleife) have compressions that limit ride height and crests that unload the rear in braking zones. Grade-aware channels prevent misdiagnosis there.

### LMP2

Compressions at high speed are critical for plank wear and aero stall, and fast crests reduce downforce exactly when mechanical load is lowest. Heave stiffness is tuned around these points.

### FIA F3

Light, stiff, ride-height-sensitive cars bottom in compressions. Plank wear limits and the risk of aero stall make vertical curvature a setup driver, not just a driving issue.

### Super Formula

Suzuka and Fuji have significant elevation changes. Corners on crests (e.g. the Suzuka esses sections) show reduced grip factors that are track geometry, not car performance.

## References

- Segers, Analysis Techniques for Racecar Data Acquisition, 2nd ed. (SAE 2014): §10.3 Banking and Grade Effects (Phoenix IndyCar example; slope equations)
- Milliken & Milliken, Race Car Vehicle Dynamics (SAE 1995): Ch. 18 Wheel Loads
- Milliken & Milliken, Race Car Vehicle Dynamics (SAE 1995): Ch. 9 "g-g" Diagram (Envelope and track geometry)
