CFD was being used in 1995 in F1 but inaccuracies meant that it was more research than application based. But it was a growth area then and continues to be a growth area now in F1.
Nowadays at Merc there are many people who work both on F1 and the Americas Cup Yacht project.
The big limitation when I was involved was wheel flow. Managing the wheel flow is a huge part of the aerodynamics of an open wheeled car and the CFD of my day could only deal with steady flow, and wheel wakes are anything but steady particularly higher up and predicting the separation point on top of the tyre was more lottery than science and getting that wrong also leads to B-I-G errors.
Have they sorted this or is it still a limitation?
Yes it is sorted.
Nowadays a small F1 team is likely to have at least 5 people working on correlation and another 5 people working on method development.
CFD is being used more and more to triage ideas before the best ones are put into the wind tunnel. So the balance in the CFD to wind tunnel balancing equation is shifting towards CFD usage now.
Do you think we’re in sight of a time where CFD is good enough that wind tunnels are redundant?
As the engineering adage goes, in theory theory and practice are the same but in practice they aren’t.
Is that because the code can now deal with unsteady flow or because there is a “better” fudge factor?
I have seen the separation point of the front tyre moved (by changing surface roughness) until the calculated forces matched real but the flow field was then nowhere near reality so it was a 
moment for me since it was obviously not good for concept work that way.
There are a series of steps to go through, but firstly I should clarify that the the wind tunnel and model are just another simulation of reality and as such have their own limitations. Whilst measurements of reality (the car running at the track) have another set of limitations.
For stable aero regulations:
With new aero regulations that contain significant changes, like for the 2022 season.
It is difficult to see wind tunnels *not (missed out in original post) being used at all (Virgin tried this for 2009 and made a big mess of it) or banned as they still provide a more accurate simulation of what is likely to happen at the track, but over time the effectiveness of CFD will continue to improve and there will be a push to move entirely to using CFD.
Yes it is because of large improvements in accuracy which has come from many cycles of research with universities and work in the teams.
There are still some tunable parameters that relate to surface roughness, but they have much less effect nowadays as the physics and the maths have improved so much.
Yes but does the CFD actually accurately deal with unsteady flow? There was some code purporting to do so which I believe BMW had and did a model for Williams but the forces were still way off and the software was mega expensive, 7-figures rental per month iirc, so it didn’t get used any more.
I am still intrigued to see if it is long-developed confident fudge or a genuinely more accurate flow calculation.
One theory I have is that the complex well developed bargeboards stabilised and controlled the front wheel wake well, so the fact that in isolation a wheel has a very messy wake wasn’t much problem with the old rules.
They don’t have any such devices available for the current rules so maybe back to wheel wake prediction being poor if the code doesn’t compute unsteady flow.
I am still unconvinced by CFD wheel wake accuracy but judging by the Ferrari pitot rake used in pre season testing, presumably generated from CFD flow lines, they believe it is steady which it definitely is not.
The fact that CFD is used to triage all ideas to select the best ideas for wind tunnel testing in at least one team and that there has been a move to using more CFD than wind tunnel testing in the balancing equation across the grid does suggest to me that accuracy has improved to a good enough point.
You need to talk to an expert about the detail of how it has been done. Whilst my field of expertise is as a simulation and analysis expert, I have never configured CFD runs or FEA runs (but have used the results from such sims) and always worked at the integration level of vehicle dynamics simulations.
I can put you in contact with an expert if you would like to discuss it directly?
Please do.
I worked extensively trying to get correlation and conceptual direction from CFD and it didn’t instil confidence - particularly the “triage of ideas” side. If we had relied on CFD what actually turned out to be the best solutions would not have been investigated.
I am concerned that after all these years confidence in CFD is higher than it deserves to be in its application to a completely new flow field - ie this year’s rules - and is maybe the reason some people have got it so very wrong.
Another reason for my scepticism is Adrian Newey building a new wind tunnel whilst advising others to use more CFD 
Expert contact sorted out.
The triage of ideas using CFD has grown over the past 15 years. With an enormous number of ideas (too many to test in the wind tunnel) there needs to be a triage process. That leaves 2 options for additional ideas; an aerodynamicist running a virtual wind tunnel in their head, or CFD.
I have come across many aerodynamicists who claim to be able to predict the outcome in their head. Generally if it is their idea then they predicted a positive result and for other ideas a far less positive view! With CFD being more successful in its predictions. Hence I did not support the idea of aeros using wind tunnels in their heads as the primary triage process.
You may be right that CFD is given a higher confidence level than it deserves (I would expect that cross the aero departments at the teams a lot of validation / correlation work has been conducted to increase the confidence, just like vehicle performance has done with vehicle simulation technology). We will find out more from the expert over the next week or so and hopefully this will give us a clearer understanding.
I am surprised that Adrian has taken so long to want a better tunnel as the Bedford tunnel had some significant limitations for F1 testing. Maybe these limitations have become more important with ground effect and ultimately if he does not have access to a high quality wind tunnel, then how does he get a lot of validation / correlation data ready for when wind tunnel use is restricted further?
I have observed wide variations in the quality of data (consistency with track data being my main measure) from 40%, 50%, 60% and 100% model results. Whilst the aeros focused on floor and diffuser pressure tappings for track data for improved validation, I have focused on forces and accelerations, with a vehicle model to help me understand the data at a more generic level.
This is indeed obvious but I am still concerned so looking forward to chatting with WT again.
It will still be the quality not quantity of ideas that leads to a competitive car.
Neither strictly F1-related, nor in continuation of the excellent ping pong with Ian and Frank, but…holy shit
Moron Piquet .
Had a great chat with Frank and WT regarding the state of F1 and the aero process.
I think that the simple conclusion is that teams need aerodynamicists with quality ideas and a quality selection process.
It should be noted that the further back the team is on the grid, the greater amount of wind tunnel time and CFD are available to them. So with the budget cap in place (my understanding is that only Haas do not spend the budget cap), this should be a massive opportunity for teams near or at the back of the grid, to move up the grid, but …
Would the McMurtry Spéirling give an F1 car a run for it’s money?
Bernie says, “Hold my beer”…
https://twitter.com/nazirafzal/status/1542406425200189440?s=20&t=68-Ei6qS2aqHBxZn9EnYWQ
What we’re they thinking / expecting having him on?
Christ alive. What a weapon
There were a LOT of racists working in F1, it was quite shocking.