Ground Effect, Floor Physics, and the Undercut

The floor is the car

From 2022 through 2025, a Formula 1 car made most of its downforce underneath itself. The front wing and rear wing are what the broadcast shows. The lap time is mostly two tunnels you cannot see, a throat where the air speeds up, and a diffuser that lets that air rejoin the freestream without the whole floor stalling.

Wings are adjustable in the garage and, within limits, from the cockpit. The floor is the part you commit to when the car leaves the garage. Change the wing and you move the balance a notch. Miss the floor's ride-height window and the car stops producing the load the tyres were brought into the weekend to use.

The 2026 cars cut overall downforce and add active aero on the straights. The pitwall problem does not retire with that ruleset. A stop still only works if the lap after the stop puts the tyre and the floor in the same window. If it does not, the undercut was a model.

A venturi, not a skirt

A venturi is a constriction. Air that is forced through a smaller section has to speed up. Faster air, lower static pressure. That is Bernoulli, and it is the whole trick: pressure under the car drops, pressure over the car does not, and the difference is a force pushing the car into the track. The tyres are what that force pushes against.

The practical shape on these cars has three jobs:

• Inlet. Feed the tunnel clean air from under the nose and ahead of the sidepod.
• Throat. The narrow, low section where the air is fastest and the pressure is lowest. This is where the downforce is made.
• Diffuser. A careful expansion. It raises the pressure back toward ambient so the low-pressure region in the throat can exist without the flow separating.

None of that survives if ambient air bleeds in from the side and spoils the throat. The floor edges and the vortices they shed are there to seal the tunnel. Lose the seal and the floor is just a flat panel with extra drag. Teams spent the ground-effect era chasing millimetres of edge geometry because the seal is the downforce.

Downforce scales with dynamic pressure, which scales with the square of speed. A corner taken at 260 km/h does not ask the tyre for a bit more load than a 130 km/h hairpin. It asks for about four times the aero load, before you even count the mechanical load from the corner itself. That is why a floor that looks genius in a fast esses can cook a rear tyre, and why a floor that looks asleep in a hairpin can still be the reason the car traps fast at the end of the straight.

Ride height is a strategy variable

The tunnel only works inside a short band of ride height. A few millimetres too high and the throat stops choking: the seal weakens, downforce falls away, and the driver reports a car that will not turn in. A few millimetres too low and the floor either kisses the track or the flow separates. Both feel, from the cockpit, like "no grip". They are opposite problems.

Fuel makes this a moving target. A heavy car at the start of a stint sits lower. As the tank empties the platform rises, unless the team has ride-height and spring choices that keep the throat in its band across that weight loss. A driver who says the car "goes away" after ten laps is often describing the floor walking out of the map, not a tyre that has suddenly lost its tread. The trace to ask for is ride height and tyre surface temperature together, not a single degradation number.

Porpoising and the plank

Porpoising is what happens when that window is unstable. Downforce pulls the platform down into a ride height where the floor flow changes. Load collapses. The springs push the car back up. The flow reattaches, the load returns, and the car does it again. The driver feels a vertical oscillation down the straight. The tyre feels a load that is switching on and off, which is a miserable way to keep a contact patch in a temperature band.

The 2022 cars did this in public. Later floor-edge and flexibility constraints were written so the oscillation would stop being a safety problem. The underlying coupling did not vanish. Any car that makes its downforce with a sealed tunnel can be pulled into the height where the tunnel misbehaves.

There is also a legal floor under the aerodynamic one. The plank is a post-race measurement. Wear it below the minimum and the car is disqualified after the flag, which is a slower kind of strategy error than pitting one lap early, and a permanent one. A setup that is fastest over a single lap because it kisses the ground is not automatically the right race setup. Race ride height is the lowest height you can hold while the plank still passes and the driver can still finish the stint without the floor stalling as the fuel comes off.

Tyres close the aero map

Pirelli's usual range runs from C1, the hardest compound, to C5, the softest. The red, yellow, and white you see on a Sunday are not fixed compounds. They are the soft, medium, and hard of that weekend's allocation. A 🔴 Soft in Monaco can be a C5. A 🔴 Soft at a high-energy circuit can be a C3. If a note does not name the C-number, it is talking about a color, not a tyre.

This worked example uses one allocation, not a universal map: 🔴 Soft C5 for a short spike, 🟡 Medium C3 as the opening race tyre, ⚪ Hard C2 as the undercut target. 🟢 Inter and 🔵 Wet are a different car. Standing water breaks the seal the tunnels need. You are suddenly on a mechanical-grip car with a ride height the dry floor was never mapped at, which is why a wet stop is a compound call and an aero call at the same time.

Two failure modes get blurred in commentary:

• Graining. The surface is too cold or the tyre is sliding. Rubber shears and rolls into grains. Grip falls even though plenty of tread is left. More sliding makes it worse.
• Blistering. The carcass is too hot. The surface bubbles and chunks. This is an energy problem. A floor that keeps adding load with the square of speed can push a rear tyre from "fine" to blistered inside a handful of high-speed laps.

A ground-effect car can show both in one lap. The slow corner slides a tyre that is not yet in its window (graining). The fast corner, a few seconds later, dumps v squared of aero load into the same tyre (heat). The compound that looked durable in the low-speed sector is the one that blisters at the end of the straight. That is the floor and the tyre being one system.

How to read the lap

Sector times are a better instrument than the lap. Sector 1 and Sector 3 on a lot of circuits are where the straight and the traction live. Sector 2 is often where the floor is supposed to pay you back. A speed trap is the honesty check: more floor usually means more drag, and drag shows up as a slower trap even while the corner sector improves.

The numbers below are a worked push lap, not a claimed result. Negative delta is faster than the reference lap. Positive is slower.

Read it as a trade, not as a verdict. Sector 2 took two tenths. The trap gave some of that back as drag, and Sector 3, the traction and the last straight, is slightly worse. The lap is still faster. The question for the strategist is whether those two tenths in the floor sector are being bought by cooking the rear tyre. If the following laps give Sector 2 back and the trap does not recover, the floor is fine and the tyre is leaving the window. If Sector 2 itself fades, the platform has risen, the seal is leaking, or the tyre is sliding so much that the floor never gets a clean entry. Those are different calls. One of them is "box". The other is "he is faster if he stops sliding".

stint 1   compound  MEDIUM C3
lap 12    S2 -0.20   trap 313   rears in window
lap 15    S2 -0.11   trap 314   surface temp climbing
lap 17    S2 +0.04   trap 316   sliding, floor not the limit
gap ahead 1.4s       undercut overlap available: 2 laps

The undercut lives in the out-lap

You stop before the car ahead. Both of you will pay a pit-lane loss. If the stops are the same length, that loss cancels. What does not cancel is the overlap: the laps you spend on a fresh tyre while they are still on the used one, minus whatever your first lap out of the box gives away.

Call the overlap k laps, the pace gap d seconds per lap (how much slower their old tyre is than your true race pace), and the out-lap penalty p (how much slower your out-lap is than that same true pace). With equal, clean stops and no safety car:

undercut gain ≈ k × d − p

You jump them only if that gain is bigger than the gap you have to close, and only if you come out ahead on the road. A positive gain that is smaller than the gap just means you will see them sooner. It is not a pass.

The second row is the ground-effect failure. The compound is fresh. The model says you should be fast. The tyre is cold, the contact patch is not the one the aero map was built on, and the car is not in the ride height where the tunnel seals. The out-lap is slow because the floor is not working, and a slow out-lap is exactly the term that kills k × d − p. Boxing into traffic, or boxing onto a compound that grains when it is cold, makes p worse. Staying out is then the faster call even though the tyre looks "done" on the television graphic.

The in-lap mostly cancels too, until it doesn't. Both cars lose time entering the pit lane if both pit in free air. If you pit from a queue of backmarkers and they pit in clean air, your in-lap is part of p and their in-lap is not. The inequality does not know that you "deserved" the undercut.

Team radio

Gap is 1.4 and stable. His rears are starting to slide. Out-lap on this compound has been about a second off. The undercut is this lap and the next. Box this lap.

That message is the inequality said out loud. Gap 1.4 seconds. Overlap of two laps at 0.75 seconds is 1.5 seconds. Out-lap penalty about 1.0 second. Gain about half a second, which does not clear 1.4. Read it again: the engineer said the undercut is live, and the arithmetic says it closes the gap and does not complete the pass unless the car ahead's next lap is worse than d, or your out-lap is cleaner than p. A good pitwall states which of those it is betting on. "Box" by itself is not a reason.

A card the pitwall can run

Same stint, three plans. Assumptions, said once: equal pit-lane loss, a clean stop, no safety car, and the Sector 2 fade in the trace above is tyre slide rather than the floor leaving the map. If a safety car appears, every one of these cards is wrong and the cheap stop is the one you take.

Plan A is the television call. Plan B is the call the arithmetic prefers when the out-lap on a ⚪ Hard C2 has been poor all weekend. Plan C throws both away, on purpose. The mistake is blending them: boxing for an undercut because rain might come, onto a dry Hard, with an out-lap you have not measured.

What to watch on Sunday

You do not need a team sheet to see this. Five checks, in order:

1. Name the C-number. The color is the weekend label. The C-number is the tyre.
2. Watch Sector 2 and the speed trap together. A faster corner sector and a slower trap means the floor is working and the drag went up.
3. If Sector 2 then fades while the trap rises, ask whether the tyre is sliding or the platform has risen. The first is a stop. The second might be fuel and ride height.
4. Before you believe an undercut, invent p. If you cannot say how slow the out-lap is, you do not have a call. You have a habit.
5. If the track is wet, stop using the dry inequality. The seal the tunnels rely on is already gone.

That is the whole note. The floor is a venturi that only works in a thin band of ride height. The tyre is the other half of that band. The undercut is just whether the lap you gain while they stay out is larger than the lap you lose while your own floor and tyre wake up.

More notes live on the F1 hub. This one is a mechanism, not a prediction, and the lap in it is a worked example rather than a result.