2022 cars 'porpoising' at high speed

[Greg Locock]

I'd just add the inerter is trying to control the unsprung mass, of the order of 20 kg (WAG) rather than half the car.

[Sulman]

Wrong again. See my other prev posts. The agency of the pilot/driver is irrelevant. The physical phenomenon is governed by the same forces. What's known as a type of bounced landing. PIO is a general term and could happen at various altitude, be applied to numerous types of faults.

https://en.wikipedia.org/wiki/Bounced_landing

The increasing and decreasing angle of attack of the aircraft on approach and as they approach touchdown is classic POI. The issue started well before the wheels first touched the ground. Then the pilot tried to correct the situation with control inputs.

Even the Shuttle pilots had an issue with PIO during the early test flights. The control logic was altered to add some damping of the pilots' control inputs to help prevent PIO in future by slowing the response of the controls to inputs.



You can see the PIO starting before the wheels touch the runway. Once they do, the pilot then over controls and the vehicle under goes further oscillations in the air.

Anyway, aircraft PIOs or undercarriage bounces are off topic for the aero-created porpoising of the current F1 cars.

I don't know if it's related but motorcycle wobble can be driver induced but is determined by the motorbike's characteristics AND aero. Specially if the wavelength surpasses normal human reaction times.

BTW, I experienced it in a very drastic way at "not so high" speeds, and suspected the unusual positioning of the bmwF800R fuel tank (rear). It stopped when I changed my backpack: I think it acted as a moving shark fin.

I used to get it on my Kawi around 48mph, but only if I didn't damp it with a very light touch on the bars. It would stop above 50. Generally things on the rear of the bike, or anything that reduces weight on the front exacerbates it.

[Michael]

(New poster, hi all).

I’m an aerodynamic non expert, but I was wondering if one approach towards constructing a floor that doesn’t generate porpoising is to engineer the floor such that it starts to ‘spill’ more flow from the sides as the ride height decreases, decreasing the flow through the throat section of the diffuser.

The teams have a range of bits and bobs along the sides of the floor. Is there a way they could introduce elements that as the ride height decreases effectively ‘switch on’ (once they get into a strong/stronger ground effect regime) that then start to ‘extract’ air from teh sides of the floor?

This would basically be a sort of ‘aerospring’ that decreases the rate of increase in downforce with decreasing ride height, thus making the porpoising.

Is this a concept that would work, or even one that’s being used?

[joseff]

I'm afraid that sounds like introducing 2nd order vibrations. I.e. another oscillation on top of the porpoising.

[cooken]

To our CFD subject matter experts: do the models these days include any kinematics and elasticity, or is everything still limited to rigid bodies and fixed ride height? In other words, is there any accounting for real time suspension movement within the sim. If so is that statics only or are viscous and inertial effects included?

I am assuming they set a fixed rigid suspension position based on estimated or iterated lift and centre of pressure. In this limited framework predicting/simulating porpoising is clearly very difficult, its a violation of some fundamental assumptions.

[Big Tea]

While paging experts ( )

Are fluid switches totally banned? Or could a pressure activated vent be used to stop the proposing by regulating the pressure and release based on the pressure difference?

[Hoffman900]

To our CFD subject matter experts: do the models these days include any kinematics and elasticity, or is everything still limited to rigid bodies and fixed ride height? In other words, is there any accounting for real time suspension movement within the sim. If so is that statics only or are viscous and inertial effects included?

I am assuming they set a fixed rigid suspension position based on estimated or iterated lift and centre of pressure. In this limited framework predicting/simulating porpoising is clearly very difficult, its a violation of some fundamental assumptions.

I brought this up during the first test, but extremely unlikely for a few reasons

1) CFD really struggles with turbulent air / vortex shedding. It’s a very very hard thing to model accurately. Even NASA, Boeing, Airbus, Lockheed Martin, etc all struggle with this.
2) You would need a suspension / chassis model.This would be extremely complex and being a model, it would have a lot of errors as well. Basically everything is a spring and they are hell to model. To quote a PhD friend “infinite stiffness only exists in undergraduate text books”.
3) no F1 team has the computing power to do both. The simulators probably reduce the resolution of all of it a lot. This is clear because Merc obviously can’t get a hold of it despite their resources and I’m sure tireless efforts at the factory.


Wind tunnel models are scaled and again no suspension. Even if they did, a scale chassis won’t go in resonance like a full scale.

[cooken]

To our CFD subject matter experts: do the models these days include any kinematics and elasticity, or is everything still limited to rigid bodies and fixed ride height? In other words, is there any accounting for real time suspension movement within the sim. If so is that statics only or are viscous and inertial effects included?

I am assuming they set a fixed rigid suspension position based on estimated or iterated lift and centre of pressure. In this limited framework predicting/simulating porpoising is clearly very difficult, its a violation of some fundamental assumptions.

I brought this up during the first test, but extremely unlikely for a few reasons

1) CFD really struggles with turbulent air / vortex shedding. It’s a very very hard thing to model accurately. Even NASA, Boeing, Airbus, Lockheed Martin, etc all struggle with this.
2) You would need a suspension / chassis model.This would be extremely complex and being a model, it would have a lot of errors as well. Basically everything is a spring and they are hell to model. To quote a PhD friend “infinite stiffness only exists in undergraduate text books”.
3) no F1 team has the computing power to do both. The simulators probably reduce the resolution of all of it a lot. This is clear because Merc obviously can’t get a hold of it despite their resources and I’m sure tireless efforts at the factory.


Wind tunnel models are scaled and again no suspension. Even if they did, a scale chassis won’t go in resonance like a full scale.

Yea pretty much as I thought. So testing and track measurements are really crucial for solving this.

By the way it doesn't take a PhD to realize that many of the assumptions we lean on in undergraduate education go out the window very quickly in real world application. The most egregious one I come across all the time is assuming small displacement and rotation / linear behaviour. Simplifying assumptions are great and often necessary to suit scope and budget etc but they are very easily abused.

[Hoffman900]

To our CFD subject matter experts: do the models these days include any kinematics and elasticity, or is everything still limited to rigid bodies and fixed ride height? In other words, is there any accounting for real time suspension movement within the sim. If so is that statics only or are viscous and inertial effects included?

I am assuming they set a fixed rigid suspension position based on estimated or iterated lift and centre of pressure. In this limited framework predicting/simulating porpoising is clearly very difficult, its a violation of some fundamental assumptions.

I brought this up during the first test, but extremely unlikely for a few reasons

1) CFD really struggles with turbulent air / vortex shedding. It’s a very very hard thing to model accurately. Even NASA, Boeing, Airbus, Lockheed Martin, etc all struggle with this.
2) You would need a suspension / chassis model.This would be extremely complex and being a model, it would have a lot of errors as well. Basically everything is a spring and they are hell to model. To quote a PhD friend “infinite stiffness only exists in undergraduate text books”.
3) no F1 team has the computing power to do both. The simulators probably reduce the resolution of all of it a lot. This is clear because Merc obviously can’t get a hold of it despite their resources and I’m sure tireless efforts at the factory.


Wind tunnel models are scaled and again no suspension. Even if they did, a scale chassis won’t go in resonance like a full scale.

Yea pretty much as I thought. So testing and track measurements are really crucial for solving this.

By the way it doesn't take a PhD to realize that many of the assumptions we lean on in undergraduate education go out the window very quickly in real world application. The most egregious one I come across all the time is assuming small displacement and rotation / linear behaviour. Simplifying assumptions are great and often necessary to suit scope and budget etc but they are very easily abused.

They are and I think F1’s over reliance on simulations makes for a worst product.

With these big rule changes they should have allowed more cap space for on track testing. Teams having to develop their cars heavily over the first 1/4 - 1/3 of the season isn’t conducive to a good show and nullifies any work to make the cars follow closer.

[cooken]

I brought this up during the first test, but extremely unlikely for a few reasons

1) CFD really struggles with turbulent air / vortex shedding. It’s a very very hard thing to model accurately. Even NASA, Boeing, Airbus, Lockheed Martin, etc all struggle with this.
2) You would need a suspension / chassis model.This would be extremely complex and being a model, it would have a lot of errors as well. Basically everything is a spring and they are hell to model. To quote a PhD friend “infinite stiffness only exists in undergraduate text books”.
3) no F1 team has the computing power to do both. The simulators probably reduce the resolution of all of it a lot. This is clear because Merc obviously can’t get a hold of it despite their resources and I’m sure tireless efforts at the factory.


Wind tunnel models are scaled and again no suspension. Even if they did, a scale chassis won’t go in resonance like a full scale.

Yea pretty much as I thought. So testing and track measurements are really crucial for solving this.

By the way it doesn't take a PhD to realize that many of the assumptions we lean on in undergraduate education go out the window very quickly in real world application. The most egregious one I come across all the time is assuming small displacement and rotation / linear behaviour. Simplifying assumptions are great and often necessary to suit scope and budget etc but they are very easily abused.

They are and I think F1’s over reliance on simulations makes for a worst product.

With these big rule changes they should have allowed more cap space for on track testing. Teams having to develop their cars heavily over the first 1/4 - 1/3 of the season isn’t conducive to a good show and nullifies any work to make the cars follow closer.

Eh, as an FE analyst I kinda disagree. FSI and multiphysics are continuously gaining traction, and I would rather see F1 innovating in this area too rather than being artificially handicapped.

I do also think the testing restrictions are way overkill. End of the day it's still a race car and needs track time.

[Hoffman900]

Yea pretty much as I thought. So testing and track measurements are really crucial for solving this.

By the way it doesn't take a PhD to realize that many of the assumptions we lean on in undergraduate education go out the window very quickly in real world application. The most egregious one I come across all the time is assuming small displacement and rotation / linear behaviour. Simplifying assumptions are great and often necessary to suit scope and budget etc but they are very easily abused.

They are and I think F1’s over reliance on simulations makes for a worst product.

With these big rule changes they should have allowed more cap space for on track testing. Teams having to develop their cars heavily over the first 1/4 - 1/3 of the season isn’t conducive to a good show and nullifies any work to make the cars follow closer.

Eh, as an FE analyst I kinda disagree. FSI and multiphysics are continuously gaining traction, and I would rather see F1 innovating in this area too rather than being artificially handicapped.

I do also think the testing restrictions are way overkill. End of the day it's still a race car and needs track time.

They are, but you would need to;

1) model the tub accurately and its stiffness / resonance, which will change when things are bolted onto it
2) model all the joints of all suspension accurately in terms of flex,
3) model body work resonance (think flutter) accuruatetly
4) model the shocks
5) model the springs , flex , resonance in the rest of the suspension
6) model the tire

The run this model in concurrence with CFD. You would need to rent time at Los Alamos to even have a sniff at this. And with CFD missing or incorrectly getting turbulence wrong and how hard vortex shedding induced resonance is to model (think skyscrapers and turbulence induced acoustics) and the model would be wrong anyway.

[cooken]

They are and I think F1’s over reliance on simulations makes for a worst product.

With these big rule changes they should have allowed more cap space for on track testing. Teams having to develop their cars heavily over the first 1/4 - 1/3 of the season isn’t conducive to a good show and nullifies any work to make the cars follow closer.

Eh, as an FE analyst I kinda disagree. FSI and multiphysics are continuously gaining traction, and I would rather see F1 innovating in this area too rather than being artificially handicapped.

I do also think the testing restrictions are way overkill. End of the day it's still a race car and needs track time.

They are, but you would need to;

1) model the tub accurately and its stiffness / resonance, which will change when things are bolted onto it
2) model all the joints of all suspension accurately in terms of flex,
3) model body work resonance (think flutter) accuruatetly
4) model the shocks
5) model the springs , flex , resonance in the rest of the suspension
6) model the tire

The run this model in concurrence with CFD. You would need to rent time at Los Alamos to even have a sniff at this. And with CFD missing or incorrectly getting turbulence wrong and how hard vortex shedding induced resonance is to model (think skyscrapers and turbulence induced acoustics) and the model would be wrong anyway.

I reckon the tires are the biggest challenge there. I don't think deformable bodywork would be all that necessary, maybe the floor ahead of the rear tyres for these cars but some are adding stays to keep that part rigid so maybe not. Perhaps the FW elements...for dynamics substructuring could be put to good use here.
The suspension and joints could be effectively modeled with very few additional DoFs.

[Vanja #66]

Had a discussion on why and how the teams missed ground-effect bouncing during car development with jjn and Vyssion this morning. Reading some comments from teams during pre-season, to me it seemed like they expected to catch any kind of dynamic effect in wind tunnel, but this is limited by "low" speeds of 180km/h and wt models are only 60% so aerodynamic forces are relatively low.

Vyssion shared some insights on what kind of simulations teams mostly do with CFD, the most important part here is that teams don't really run straight-line simulations at all any more, as those are practically useless for cornering phenomena and results. Also, teams might be running simulations mostly at 180km/h as well, so lack of straight-line and "high" speed simulations could have easily led to missing the extent of porpoising.

Another thing jjn added is the outboard floor flexing, creating strong floor sealing which also adds up to overall porpoising phenomena. This would have also been hard to show in wind tunnel, as models are mostly made of metal to preserve the thin edges and thickness, which is usually less pliant than actual composite floors. And, as mentioned, forces in wind tunnel are really low compared to actual car going 300+ km/h... Teams managed to agree to allow floor stays, this obviously helped a bit with those problems, so floor flexing was a big part of it as well.

[Hoffman900]

The 180km/h limit blows my mind still. I was shocked to be reminded of that (must have forgotten).

I really do think with the wholesale changes, they should have allowed more on track testing this year.

[maxxer]

The 180km/h limit blows my mind still. I was shocked to be reminded of that (must have forgotten).

I really do think with the wholesale changes, they should have allowed more on track testing this year.

first time i knew about this , wow so limited time and limited speed , why invest in a tunnel even