Red Bull RB21

[venkyhere]

what's the layman summary of this new discovery ?

internal flow gets botched up under yaw conditions (corners) and affects the quality of air exiting the radiators and flowing towards the rear -> adversely affecting all the DF structures in the rear, ie, beam wing, rear wing, and thus in turn, indirectly affecting the air exiting the floor via the diffuser as well.

Did I understand it right ?
If yes, more questions :
- why hasn't it affected high speed corners much ? because of smaller yaw angles ?
- how does this explain 'rotation' issues in slow corners ?

[Silent Storm]

what's the layman summary of this new discovery ?

internal flow gets botched up under yaw conditions (corners) and affects the quality of air exiting the radiators and flowing towards the rear -> adversely affecting all the DF structures in the rear, ie, beam wing, rear wing, and thus in turn, indirectly affecting the air exiting the floor via the diffuser as well.

Did I understand it right ?
If yes, more questions :
- why hasn't it affected high speed corners much ? because of smaller yaw angles ?
- how does this explain 'rotation' issues in slow corners ?

You are correct...

At high speeds, yaw angles are smaller and the flow is more stable. Plus, there’s more mass flow overall, which can help overpower some of the low pressure recirculations or flow separation from internal ducts. Also, external flow structures (like floor edge vortices) dominate more in high-speed corners, potentially masking smaller rear aero disturbances caused by imperfect internal flow.

At low speed corners not only is the yaw angle higher, the car spends longer in a transient state. Internal flow may separate or stagnate unevenly, creating unpredictable outwash or asymmetric thermal plumes.

That turbulent, misaligned exit flow can cause rear downforce to fluctuate mid corner, reducing stability and making rotation either lazy or snappy depending on the phase. The car might feel fine on entry but vague or inconsistent at apex, especially if the beam wing and floor can’t stabilize the rear end due to dirty incoming flow. To stabilize the rear, low speed entry is usually sacrificed. If you remember there were times when both RB20 and RB21 performed well in low speed corners, could be that on some tracks the team was not forced to compromise their setup.

Basically, the rear structure of the car is exposed to hot turbulent (dirty) air from its own cooling exits, which can have an impact on rear deg too, not directly but through the domino effect it triggers. Teams usually compensate for this via setup (dialing in understeer to look after rears, etc) all this indirectly has an impact on low speed corners.

Also, one of the visible updates in Jeddah was the slightly wider opening of cooling exit.

[Paa]

You are correct...

At high speeds, yaw angles are smaller and the flow is more stable. Plus, there’s more mass flow overall, which can help overpower some of the low pressure recirculations or flow separation from internal ducts. Also, external flow structures (like floor edge vortices) dominate more in high-speed corners, potentially masking smaller rear aero disturbances caused by imperfect internal flow.

At low speed corners not only is the yaw angle higher, the car spends longer in a transient state. Internal flow may separate or stagnate unevenly, creating unpredictable outwash or asymmetric thermal plumes.

That turbulent, misaligned exit flow can cause rear downforce to fluctuate mid corner, reducing stability and making rotation either lazy or snappy depending on the phase. The car might feel fine on entry but vague or inconsistent at apex, especially if the beam wing and floor can’t stabilize the rear end due to dirty incoming flow. To stabilize the rear, low speed entry is usually sacrificed. If you remember there were times when both RB20 and RB21 performed well in low speed corners, could be that on some tracks the team was not forced to compromise their setup.

Basically, the rear structure of the car is exposed to hot turbulent (dirty) air from its own cooling exits, which can have an impact on rear deg too, not directly but through the domino effect it triggers. Teams usually compensate for this via setup (dialing in understeer to look after rears, etc) all this indirectly has an impact on low speed corners.

Also, one of the visible updates in Jeddah was the slightly wider opening of cooling exit.

This is all very interesting, thanks for sharing.
I'm wondering why RedBull would go this way of adding such complexity and risks.
Were they confident that they can confidently simulate all these effects? Or they just neglected the side-effects, trusting that the gains will be greater than the potential losses? Or they thought they don't need to consider these at all, trusting that they could perfectly control the internal flows?

[Silent Storm]

You are correct...

At high speeds, yaw angles are smaller and the flow is more stable. Plus, there’s more mass flow overall, which can help overpower some of the low pressure recirculations or flow separation from internal ducts. Also, external flow structures (like floor edge vortices) dominate more in high-speed corners, potentially masking smaller rear aero disturbances caused by imperfect internal flow.

At low speed corners not only is the yaw angle higher, the car spends longer in a transient state. Internal flow may separate or stagnate unevenly, creating unpredictable outwash or asymmetric thermal plumes.

That turbulent, misaligned exit flow can cause rear downforce to fluctuate mid corner, reducing stability and making rotation either lazy or snappy depending on the phase. The car might feel fine on entry but vague or inconsistent at apex, especially if the beam wing and floor can’t stabilize the rear end due to dirty incoming flow. To stabilize the rear, low speed entry is usually sacrificed. If you remember there were times when both RB20 and RB21 performed well in low speed corners, could be that on some tracks the team was not forced to compromise their setup.

Basically, the rear structure of the car is exposed to hot turbulent (dirty) air from its own cooling exits, which can have an impact on rear deg too, not directly but through the domino effect it triggers. Teams usually compensate for this via setup (dialing in understeer to look after rears, etc) all this indirectly has an impact on low speed corners.

Also, one of the visible updates in Jeddah was the slightly wider opening of cooling exit.

This is all very interesting, thanks for sharing.
I'm wondering why RedBull would go this way of adding such complexity and risks.
Were they confident that they can confidently simulate all these effects? Or they just neglected the side-effects, trusting that the gains will be greater than the potential losses? Or they thought they don't need to consider these at all, trusting that they could perfectly control the internal flows?

Every major package change has side effects : plumbing reroutes, mass distribution shifts, new vibration modes, even suspension cooling crosstalk. They gambled that the net would be positive and it was, in sim... On track the unmodelled side effects bit them in dynamic corners.

They had reasons to trust their tools, From extensive RANS to DES cfd runs, windtunnel validation of the same and simulator checks running the same aero map. All three showed the new packaging should have delivered the predicted downforce/drag balance. In a perfect world, the gains outweighed the theoretical risk.

Thermal stratification, hot air recirculation that alter pressure drop across radiators and yaw induced separation... All three would show up only if you do full conjugate heat DES with wall modeled LES. A high run time and resource cost even for Red Bull in this cost cap era.

[AR3-GP]

This is all very interesting, thanks for sharing.
I'm wondering why RedBull would go this way of adding such complexity and risks.

You have to take risk to stay ahead of the competition.

[AR3-GP]

You are correct...

At high speeds, yaw angles are smaller and the flow is more stable. Plus, there’s more mass flow overall, which can help overpower some of the low pressure recirculations or flow separation from internal ducts. Also, external flow structures (like floor edge vortices) dominate more in high-speed corners, potentially masking smaller rear aero disturbances caused by imperfect internal flow.

At low speed corners not only is the yaw angle higher, the car spends longer in a transient state. Internal flow may separate or stagnate unevenly, creating unpredictable outwash or asymmetric thermal plumes.

That turbulent, misaligned exit flow can cause rear downforce to fluctuate mid corner, reducing stability and making rotation either lazy or snappy depending on the phase. The car might feel fine on entry but vague or inconsistent at apex, especially if the beam wing and floor can’t stabilize the rear end due to dirty incoming flow. To stabilize the rear, low speed entry is usually sacrificed. If you remember there were times when both RB20 and RB21 performed well in low speed corners, could be that on some tracks the team was not forced to compromise their setup.

Basically, the rear structure of the car is exposed to hot turbulent (dirty) air from its own cooling exits, which can have an impact on rear deg too, not directly but through the domino effect it triggers. Teams usually compensate for this via setup (dialing in understeer to look after rears, etc) all this indirectly has an impact on low speed corners.

Also, one of the visible updates in Jeddah was the slightly wider opening of cooling exit.

This is all very interesting, thanks for sharing.
I'm wondering why RedBull would go this way of adding such complexity and risks.
Were they confident that they can confidently simulate all these effects? Or they just neglected the side-effects, trusting that the gains will be greater than the potential losses? Or they thought they don't need to consider these at all, trusting that they could perfectly control the internal flows?

Every major package change has side effects : plumbing reroutes, mass distribution shifts, new vibration modes, even suspension cooling crosstalk. They gambled that the net would be positive and it was, in sim... On track the unmodelled side effects bit them in dynamic corners.

They had reasons to trust their tools, From extensive RANS to DES cfd runs, windtunnel validation of the same and simulator checks running the same aero map. All three showed the new packaging should have delivered the predicted downforce/drag balance. In a perfect world, the gains outweighed the theoretical risk.

Thermal stratification, hot air recirculation that alter pressure drop across radiators and yaw induced separation... All three would show up only if you do full conjugate heat DES with wall modeled LES. A high run time and resource cost even for Red Bull in this cost cap era.

To add, Internal airflow inside the wind tunnel model is incredibly difficult to replicate because you can't just take a lifesize radiator and shrink it down, like the rest of the windtunnel model. The model radiators also don't produce any heat in the outlet, the way the real thing does. There's radiation of heat from the engine and exhaust which is not present in the model as well. So there are a lot of inaccuracies that can add. Not surprising if a team thought they "found something", and it turned out to be an illusion.

This might be why the aero testing of cooling systems is unlimited, because it's basically junk science compared to the rest of the car.

[ispano6]

The problem with the RB21 is the floor needs to produce more DF but they rely on the wings. It's particularly sensitive to the wind too.

RB18,19,20 poduced the 'highest downforce' of any team, from the floors. They could run with significantly less wings than other teams and not lose time in the corners (except slow speed, where the airspeed isn't meaningful enough to produce DF anyway) .

The 'philosophy' as per Pierre Wache, for RB21, was 'sacrifice peak DF, inorder to enjoy a wider working window across all kinds of tracks'. That has crashed like a house of cards - neither is enough DF present, nor is the working window any wider (in fact, narrower) . But someone like Rob Marshall (a disciple of Newey) has gone to McLaren and produced a floor that enjoys both peak DF as well as a wide working window.

Where are we drawing conclusions that this car doesn't produce enough floor downforce from?

The overall downforce is lower and that's all we know. The wings are still simple relative to others and other teams have flexi components that give free lap time, going beyond front wing.

I can understand the frustration but how do we know this?

The whole point of the ground effects era is for the floor/diffuser to produce the majority of the efficient DF. Relying on the front wing to generate vortices was frowned upon for this generation car. Additionally the rear wing regulations also were put in place to reduce the amount of dirty air shed from the car that would affect the car behind. Similar to the Lotus 79 that reigned supreme that had a manual skirt that acted as a curtain extending from the floor, the sealing of the floor low to the ground and keeping the platform at a consistent ideal height is what makes the floor work. Pitch moments and porpoising break that seal and cause the the platform to be nervous. The front suspension thus greatly mitigates the pitch moment with anti-dive geometry, as well as a pull-rod design to maximize the leading edge of the floor.

Vortices are still created to help improve downforce of the floor, but the main principle behind it is a pressure gradient leveraging the venturi vanes underneath the car and the low pressure below the diffuser creating a suction effect. Adding more rear wing to add downforce is thought to be a bandaid to the floor not producing enough steady downforce. The annular rear wing (closed loop) does aid in mitigating crosswind effects on the platform, but the more you stray from the ideal shape induces more drag and sensitivity for the amount of DF you gain. In fact, Red Bull have corrected their rear wing geometry in Jeddah to curve down the outer edge to return back to the annular loop. The more you rely on the rear wing to generate DF, the more likely the height changes from the ideal height as less downforce is applied through slower corners. As always, there is aeroelasticity at play and how the platform behaves but it isn't as critical as some make it to be. Especially when considering pressure gradients induced by temperature differences that can be influenced by the warmer air that comes through the body and over the beam wing, which is an extension of the diffuser, the suction effect is what is desired. In some ways it is reminiscent of the concept behind the blown diffuser. It appears Red Bull have realized this and returned the focus of downforce back to the floor and diffuser/beamwing.

Red Bull trimmed the tip of the rear wing to bring it closer to the ideal annular loop which also reduces DF and drag, and quite interestingly, simplified the beam wing with less chord but less camber as called out by planetf1.
https://www.planetf1.com/features/f1-un ... arabian-gp



It also introduced a new camera pod to observe the front tyres:

The relentless quest for performance also requires teams to think outside the box when gathering information and data on the car’s performance, with the team having developed a unique camera pod arrangement for use during Free Practice sessions in Saudi Arabia.

The shape of the pods, mounted on either side of the airbox, appear to position the cameras in such a way that the front tyres can be monitored more effectively, whether that be with a hi-speed camera to better understand the tyres deformation, or with a infra-red camera to gain further knowledge on how the tyre temperatures evolve during the course or a lap or stint.

[ispano6]

Correction : Okay, so the picture above of the central radiator is of racing bulls not the RB21. The person who tweeted it used racing bulls picture instead of Redbull.

Apologies for the error, I should have checked racing bulls gallery as well. Having said that, rest of the theory is still sound as they did move from cannons to normal, V radiator to slanted radiator and we will see in Imola if third change was made or not on RB21 that moved weight distribution of the car.

It started with having shortened the nose of the car back to the RB18/19, and they may start balancing the rest of the car to this change. Regarding the RB21 internal flow structures, they are doing something rather different from the RB19. They haven't reverted back to the baleen whale "throat" and have stuck to the stingray mouth and halo base openings. The RB19 louvres were quite a bit smaller and farther forward. The RB21's already sit well back and lower.





Shorter chord, less cambered beam wing, wider exit hole.


From Aramco's Raceteq website:
https://www.raceteq.com/articles/2025/0 ... -explained

[AR3-GP]

It also introduced a new camera pod to observe the front tyres:

The relentless quest for performance also requires teams to think outside the box when gathering information and data on the car’s performance, with the team having developed a unique camera pod arrangement for use during Free Practice sessions in Saudi Arabia.

The shape of the pods, mounted on either side of the airbox, appear to position the cameras in such a way that the front tyres can be monitored more effectively, whether that be with a hi-speed camera to better understand the tyres deformation, or with a infra-red camera to gain further knowledge on how the tyre temperatures evolve during the course or a lap or stint.

https://d3cm515ijfiu6w.cloudfront.net/w ... ameras.jpg

I saved a video over the weekend because I noticed the tire sidewall flexing. Look at the left front tire behind the brake duct outlet. I don't think it is specific to Red Bull (in some ways, yes, but all cars have some amount of deflection depending on the amount of lateral acceleration that the car is able to generate from the downforce and steering angle)

[Silent Storm]

This is all very interesting, thanks for sharing.
I'm wondering why RedBull would go this way of adding such complexity and risks.
Were they confident that they can confidently simulate all these effects? Or they just neglected the side-effects, trusting that the gains will be greater than the potential losses? Or they thought they don't need to consider these at all, trusting that they could perfectly control the internal flows?

Every major package change has side effects : plumbing reroutes, mass distribution shifts, new vibration modes, even suspension cooling crosstalk. They gambled that the net would be positive and it was, in sim... On track the unmodelled side effects bit them in dynamic corners.

They had reasons to trust their tools, From extensive RANS to DES cfd runs, windtunnel validation of the same and simulator checks running the same aero map. All three showed the new packaging should have delivered the predicted downforce/drag balance. In a perfect world, the gains outweighed the theoretical risk.

Thermal stratification, hot air recirculation that alter pressure drop across radiators and yaw induced separation... All three would show up only if you do full conjugate heat DES with wall modeled LES. A high run time and resource cost even for Red Bull in this cost cap era.

To add, Internal airflow inside the wind tunnel model is incredibly difficult to replicate because you can't just take a lifesize radiator and shrink it down, like the rest of the windtunnel model. The model radiators also don't produce any heat in the outlet, the way the real thing does. There's radiation of heat from the engine and exhaust which is not present in the model as well. So there are a lot of inaccuracies that can add. Not surprising if a team thought they "found something", and it turned out to be an illusion.

This might be why the aero testing of cooling systems is unlimited, because it's basically junk science compared to the rest of the car.

In my view, It's not junk science but hard science. It means you have to treat windtunnel cooling results with care, fuse them with on track temperature probes and use correction factors for heat transfer. In practice you end up with a mixed strategy:

1) Cold model tunnel + RANS/LES CFD to nail bulk flow paths
2) Heated radiator rigs for pressure drop calibration
3) On track instrumentation (thermocouples, IR cameras) to close the loop

So yes, early tunnel success can be illusory if you don’t model the heat, and that’s likely how a “breakthrough” internal duct concept can look great in sim/tunnel but misbehave on track. Teams are very aware of it and build entire sub programs (and the “unlimited” cooling tests) just to catch these errors before deployment.

[AR3-GP]

That could be the reason for the abandonment of the cannons and the octo-inlets, as you said earlier. Was anyone actually buying that one set of cooling covers was for low downforce and the other set was for high downforce?

Its more likely that one set didn't produce any downforce (unintentionally... ).

[Silent Storm]

Correction : Okay, so the picture above of the central radiator is of racing bulls not the RB21. The person who tweeted it used racing bulls picture instead of Redbull.

Apologies for the error, I should have checked racing bulls gallery as well. Having said that, rest of the theory is still sound as they did move from cannons to normal, V radiator to slanted radiator and we will see in Imola if third change was made or not on RB21 that moved weight distribution of the car.

It started with having shortened the nose of the car back to the RB18/19, and they may start balancing the rest of the car to this change. Regarding the RB21 internal flow structures, they are doing something rather different from the RB19. They haven't reverted back to the baleen whale "throat" and have stuck to the stingray mouth and halo base openings. The RB19 louvres were quite a bit smaller and farther forward. The RB21's already sit well back and lower.

https://www.raceteq.com/-/jssmedia/race ... ab0352.jpg

https://www.raceteq.com/-/jssmedia/race ... ab0274.jpg

Shorter chord, less cambered beam wing, wider exit hole.
https://www.raceteq.com/-/jssmedia/race ... ab9356.jpg

From Aramco's Raceteq website:
https://www.raceteq.com/articles/2025/0 ... -explained

External inlet geometry and internal packaging are related, but not one and the same. Redbull can keep the stingray style inlet while quietly reverting internal ducts back to RB19 philosophy.

Why they don't need baleen whale throat?

The sidepod inlet controls where air enters, but the internal plenum (volume, splitter vanes, collector shape) and radiator position dictate how that air is routed. They can keep the same mouth while reshaping the inside. RB21 did move to slanted radiator of RB19 without touching the inlet.

External inlets set the boundary, but internal packaging writes the rules for radiator mass flow, heat rejection, and downstream plume behavior.

There is a reason why RB19 is brought out of retirement to run around Silverstone...

What can Redbull do in this test?

Use pressure tap arrays, thermocouple grids and mass flow sensors, and run these tests:

1) Heater mat radiator tests - Before track running, team will mount heater mats on the RB19’s radiators to bring them to realistic operating temperatures in the pit lane. Record how the pressure drop curves change when the cores are hot vs. “cold model” conditions.
2) Coolant loop profiling - Team will run a full engine cooling circuit on a dyno bench under race representative thermal loads, capturing detailed coolant side flow, pressure, and temperature maps. These are then cross referenced with data from on car validation runs to ensure correlation in real world operating conditions.

After all this it's back to factory where they will start simulator and cfd re-alignment. Basically update internal boundary conditions in their RANS/DES runs with the real pressure loss and mass flow curves. Enhance the simulator’s thermal models so that driver in loop setups now reflect the actual heat rejection and plume behavior observed on track.

By running these comprehensive cooling loop and on car validation tests, Red Bull can systematically close the gap between their digital models and the real RB21. By nailing down every nuance of internal cooling and its interaction with the external aero, Red Bull will regain the predictability they enjoyed with the RB19.

This test will also help Redbull in brake duct development. Something they said they are working on.

[Silent Storm]

That could be the reason for the abandonment of the cannons and the octo-inlets, as you said earlier. Was anyone actually buying that one set of cooling covers was for low downforce and the other set was for high downforce?

Its more likely that one set didn't produce any downforce (unintentionally... ).

Haha exactly, That one set lowered our hope more than it lowered drag.

[f1isgood]

RB18,19,20 poduced the 'highest downforce' of any team, from the floors. They could run with significantly less wings than other teams and not lose time in the corners (except slow speed, where the airspeed isn't meaningful enough to produce DF anyway) .

The 'philosophy' as per Pierre Wache, for RB21, was 'sacrifice peak DF, inorder to enjoy a wider working window across all kinds of tracks'. That has crashed like a house of cards - neither is enough DF present, nor is the working window any wider (in fact, narrower) . But someone like Rob Marshall (a disciple of Newey) has gone to McLaren and produced a floor that enjoys both peak DF as well as a wide working window.

Where are we drawing conclusions that this car doesn't produce enough floor downforce from?

The overall downforce is lower and that's all we know. The wings are still simple relative to others and other teams have flexi components that give free lap time, going beyond front wing.

I can understand the frustration but how do we know this?

The whole point of the ground effects era is for the floor/diffuser to produce the majority of the efficient DF. Relying on the front wing to generate vortices was frowned upon for this generation car. Additionally the rear wing regulations also were put in place to reduce the amount of dirty air shed from the car that would affect the car behind. Similar to the Lotus 79 that reigned supreme that had a manual skirt that acted as a curtain extending from the floor, the sealing of the floor low to the ground and keeping the platform at a consistent ideal height is what makes the floor work. Pitch moments and porpoising break that seal and cause the the platform to be nervous. The front suspension thus greatly mitigates the pitch moment with anti-dive geometry, as well as a pull-rod design to maximize the leading edge of the floor.

Vortices are still created to help improve downforce of the floor, but the main principle behind it is a pressure gradient leveraging the venturi vanes underneath the car and the low pressure below the diffuser creating a suction effect. Adding more rear wing to add downforce is thought to be a bandaid to the floor not producing enough steady downforce. The annular rear wing (closed loop) does aid in mitigating crosswind effects on the platform, but the more you stray from the ideal shape induces more drag and sensitivity for the amount of DF you gain. In fact, Red Bull have corrected their rear wing geometry in Jeddah to curve down the outer edge to return back to the annular loop. The more you rely on the rear wing to generate DF, the more likely the height changes from the ideal height as less downforce is applied through slower corners. As always, there is aeroelasticity at play and how the platform behaves but it isn't as critical as some make it to be. Especially when considering pressure gradients induced by temperature differences that can be influenced by the warmer air that comes through the body and over the beam wing, which is an extension of the diffuser, the suction effect is what is desired. In some ways it is reminiscent of the concept behind the blown diffuser. It appears Red Bull have realized this and returned the focus of downforce back to the floor and diffuser/beamwing.

Red Bull trimmed the tip of the rear wing to bring it closer to the ideal annular loop which also reduces DF and drag, and quite interestingly, simplified the beam wing with less chord but less camber as called out by planetf1.
https://www.planetf1.com/features/f1-un ... arabian-gp

https://d3cm515ijfiu6w.cloudfront.net/w ... -wing2.jpg

It also introduced a new camera pod to observe the front tyres:

The relentless quest for performance also requires teams to think outside the box when gathering information and data on the car’s performance, with the team having developed a unique camera pod arrangement for use during Free Practice sessions in Saudi Arabia.

The shape of the pods, mounted on either side of the airbox, appear to position the cameras in such a way that the front tyres can be monitored more effectively, whether that be with a hi-speed camera to better understand the tyres deformation, or with a infra-red camera to gain further knowledge on how the tyre temperatures evolve during the course or a lap or stint.

https://d3cm515ijfiu6w.cloudfront.net/w ... ameras.jpg

Most of the downforce in every generation of F1 cars comes from the floor — it's not specific to the ground effect era. As you rightly said, that's where most of the efficient downforce is, but it’s still unclear to me why there's this belief that Red Bull's floor isn't doing its job. Ground effect becomes far more pronounced at high speeds, and from what we’ve seen, Red Bull is still right there with the best teams in that area. They’re also more efficient than McLaren, which makes me think McLaren are gaining more through their wings, and that would naturally make them more draggy — which isn’t insignificant. So I’m not convinced McLaren are producing more floor downforce at all. If anything, corner performance seems to show this: in floor-dominated, high-speed sections, Red Bull are still right there, but in other corners, not so much.