Air and water are both fluids and they behave in similar fashions as they flow around obstacles. The design of F1 cars, as we know, depends on efficient aerodynamics and we tend to think of designers in this field coming from the world of aerospace. But, those who study the flow of water can be just as useful. The current Technology Director of the all-powerful Mercedes AMG Petronas team, Geoff Willis, is a case in point. His F1 career began 26 years ago when he began to do some consulting work with Leyton House Racing, which was then under the technical leadership of a young Adrian Newey. The team had a consultant who also consulted with Britain’s America’s Cup consortium at the time, known as Blue Arrow. This had built a startlingly different kind of yacht to all the other challengers, known as “Radical”, for obvious reasons. The hydrodynamicist on Radical was Willis.
An engineering graduate from Cambridge, Willis had worked initially for the scientific research association known as British Maritime Technology, which had previously been a government institution. Based in Teddington, it studied all manner of different things, many of them secret, such as the design of submarines and torpedoes. While he was doing his research work, Willis also gained a PhD in Engineering Science from Exeter University, specializing in hydrodynamics. A while later he was approached by Blue Arrow and asked to join the design team. Unfortunately, Radical was rather too radical and in October 1988, while doing test runs off the coast of Falmouth, the vessel hit a large wave, while travelling at high speed. The bow dug into the water and the yacht pitchpoled, fracturing the hull. The boat was not rebuilt.
The design process had, however, introduced Willis to the new science of computational fluid dynamics, using computer models to simulate fluid flows and thus it was that he became a valuable person to Mr Newey. It was a short-lived relationship because after just a few months Newey was fired by Leyton House (just before the cars came good) and he landed at Williams. He called in Willis and so Geoff joined the team full-time and when Adrian left for McLaren in 1997, Willis became Chief Aerodynamicist of Williams and from there went on to become Technical Director of BAR and Red Bull Racing before landing at Mercedes in 2011.
joeblogsf1 
A good one again.
unreal!!!!
i would like to see or to know more of hia work…..
In the late 80’s, I worked for a computer company, one of the partners in which had created a piece of minicomputer software for marine architecture. He apparently used it in his consultations with the U.S Navy on the design of nuclear submarines. He also used it to help design America’s Cup boats. So, that connection is not unheard of. Laughably, I asked him to try designing a car with the software. Since the starting point was turning a sloop hull upside down, the result had a longitudinal roof ridge that looked less like that on a split-window Corvette and more like an abbreviated keel. It’s clear that Newey et al. have done a slightly better job.
Current WEC cars and several incarnations of F1 and Indy have pretty significant keels present on their bodywork.
I’m pretty sure Merc in the 90’s used hydrodynamics on it’s road cars, particularly the E class of the time.
Yes, I remember reading about that as well.
One of the ’90’s Japanese Le Mans makers (Nissan?) made a lot of use of a water tunnel.
Great read!
I love the FFF’s!
Seeing that you tell us about some of the personalities in F1, especially the engineers, do you have any interesting anecdotes about Mr Motormouth, Frank Dernie? I recall an Autosport interview, he was speaking about the influence of 3D-CAD modelling and mentioned (something like) computational tools make good engineers better as the real work, the thinking, methods and engineering science is still needed to extract the full benefit offered by computational tools (perhaps you even conducted the interview!?) Whatever happened to Mr Dernie & his 50-cuppa’s a day!? I think he last worked for Toyota in F1?
Thank you Joe.
I’m sure FD is out there and willing to comment.
His Motorsport Magazine Podcast is well worth listening too.
Geoff Willis is a bright guy. Also in my limited experience, he’s a very sensible and grounded one. I
t’s good that you’ve highlighted him in your latest fascinating fact, Joe.
Love the winter knowledge grabs, keep ’em coming.
It also goes the other way too. The British entry in the upcoming 2017 America’s Cup, Land Rover Ben Ainslie Racing, has a couple of well known F1 figures involved at the top, Martin Whitmarsh and Adrian Newey, along with several others from the Red Bull Racing design department.
I’m sure you already knew that.
Why on earth was Newey fired?!!
Because the people running the show were not awfully smart
As I recall it was either the floor of the wind tunnel model, or the belt in the tunnel what was moving under test, so had inconsistent results and so the car didn’t work as expected.
They fired Adrian Newey just as he had figured out what was wrong.
Early 90’s Leyton House GP had a pretty small aerodynamics department. Aside from draughtmen and model makers, I think it consisted of Adrian Newey and Chris Burkett (who now does some very nice aerobatics displays as well as co-running a CFD company)..
I was going to say, firing Adrian Newey is right up there with Decca not signing The Beatles…
The submarine analogy is not too good for vehicles, as they normally pass through the fluid, rather than on the surface.
The issue that fascinates race engineers is what happens at the boundary layer between the track and still air – ground effects.
That has developed steadily since the 60’s, but now, with huge amounts of computing capacity available to them, the minutia of what happens around mirrors, corner roll, steering angle etc. can be studied in great detail.
It’s a pity the the Strategy Group have again used their unfair advantage to block a proposal from Manor to have a bigger computer allowance to substitute for them using only 50% scale models in their wind tunnel.
Using CFD can be used to develop many products so it’s a pity that F1 want to stick with huge and expensive windtunnels.
Budget caps are long overdue.
One interesting innovation regarding ships, is the huge lump they’ve put on the front of them, just below the waterline. I’m waiting to see which F1 team copies that in 2017.
Sorry Rodger, which one travels on the surface, the submarine or the F1 car?
Teams can only use 60% models right now, so a reduction to 50% doesn’t seem to give much away. There is also the issue of verification of CFD results with reality. One of the teams that really lost out with the wind tunnel regulations is Sauber, with their (thanks BMW) full scale rolling road tunnel.
Is that (PR) “Full scale” or actual Full-scale (with acceptable blockage)?
Sauber has an amazing rolling road wind tunnel that can take full size F1 cars. It was operational in 2003 and cost $50 milllion. The test area is large enough to allow a full size care a yaw angle of 10° to airflow. It can also take two 60% models at the same time, allowing analysis of aero interactions between cars. I take this information from ‘The Science of Formula 1 Design’ by our own David Tremayne, a book worth having.
Yep, I’ve got that book around somewhere.
Saying it can fit in a full-size car isn’t the same as it being able to actually test a car repeatably and accurately at full size.
As a rough guide anything with more than about 7% blockage (the ration of car reference area to wind tunnel working section cross section area) and you start to have serious wall effects.
Many F1 tunnels use “passive” adaptive wall techniques to bump this figure up but not by a huge amount.
Think about it this way, it’s the difference between the aerodynamics of a car in fresh air, and one in a box with walls and a ceiling (and partial floor depending on the size of the belt).
So some F1 teams do limited testing on parts of the car with a full car on a stopped belt or fixed floor but it’s generally disruptive of a normal 60% scale test programme. Any shots you see of a 100% scale F1 in a tunnel are more likely to be for PR than for aerodynamic testing.
Going from 50% to 60% may not seem like much but it’s a big difference in reliability of results. If you get the stiffness of the wind tunnel model right (and models generally have to be much stiffer than a scaled down car) then going from 50% to 60% enables more accurate model shape, better representation of the aerolastic behaviour of the real car and better results (you’re not extrapolating the Reynolds number as much).
Ultimately, you’d have a 100% model (not a real car) in a full Reynolds number (full size, actual speed and temperature) wind tunnel but there are a number of problems with that:
(a) The cost and effort making the model, (which slows down how many design iterations you can test) sorting out the aerolastics of the model (does it behave like a real F1 car). Wen you’ve dome all that it’s almost as much effort as making the car.
(b) The F1 rules limit you to a 60% model in a atmospheric wind tunnel and something like 50m/s.
I’d explain why it’s a stoopid idea to try and use a pressurised wind tunnel for F1 (even if the FIA allowed it) but I fear Joe would ban me for being too esoteric and long-winded.
[I must take this anorak off now…]
🙂
Hi rmm,
without donning anorak, thanks! you indirectly answered a few Qs i had about the sensitivity of model feedback to CFD, at least conceptually.
But can you – or anyone – shed light on the resource limits?
The best I can get is vague limits e.g,. ca. 40 TFLOPS max theoretical, and frankly that’s under the GPGPU compute budget I’m evaluating for a entirely different task, but for a solo effort…
So, what is it with a arbitrary limitation saying F1 teams get less kit to play with than I do for a tinkering project..?
Something gives here, I know it does.. has to!
cheers/j
It’s amazing what you can learn from hydro/fluid dynamics. My PhD supervisor was an academic fluid dynamicist. My PhD was on the optimisation of calculations of the variations in the gravitational field of the earth … and the fundamental source of that calculation was one of Euler’s equations from fluid dynamics!
…and they still can’t predict the weather…
was watching a documentary, when a weather modeller related how they had it pointed out to them, that the cell sizes they were calculating for, conveniently “ignored” the Rockies…. up to which point, the narrator lamented his younger self had been quite chuffed with the developments… I’m fairly certain similar such revelations must occur frequently within F1. A optical engineer who worked for Detroit auto all his life, asserts more calculations for lens design are made in any recent 3 years, than in prior history, in a similar vein.
This is what worries me, that lack of talk in F1 on past form more likely means impossibility of policing limits keeps everyone quiet, but I’ve been trying to break up a very long comment draft all evening, how I think F1 is throwing away a really serious and readily realizable income source, due to silence about computing’s significance in the sport. I think enough to set the smaller teams budgets straight, no – more, as much as changes the money game all the length of the grid possibly, is being chucked straight out the window, and my mind is teeming with ways to stop this waste of opportunity, for a while now. Some of the data wrangling teams must do at pitstop speeds, intimidate companies into planning paralysis. The commissions paid to salespeople on bigger bleeding edge computing deals, would actually feed smaller teams. The knowledge i fret is kicking proverbial tires in F1 has immediately commutable value that would make vendor sales directors sit up.
Whereas a lot of theoretical fluid mechanics (as different from CFD) is borrowed from Electromagnetic theory.