07/02/2011
NEWS STORY
 |
Mat Coch writes:
It's fair to say Monaco doesn't hold too many fond memories for Nick Wirth. It was on the Principality's streets that his Simtek team took part in its final race before its untimely demise a few short weeks later.
That weekend in May 1995 is one Wirth will have tried hard to erase from his memory. Both drivers, Jos Verstappen and Mimmo Schiattarella, retired from the race. The handful of laps the pair completed were to be the last Simtek would ever turn, the team going into voluntary liquidation shortly after the Canadian Grand Prix, an event the team didn't even travel to. It was the end of Nick Wirth's tenure as a Formula One team boss.
Wirth started his Formula One career at March, working alongside Adrian Newey until he and Max Mosley founded Simtek Research in 1989. After the venture collapsed on the back of the racing team's failure he joined Benetton as chief designer, a position he held until 1999.
His path then took a turn away from motor sport as he founded RoboScience, a robotics company which developed, of all things, a robotic dog. Then, in 2003, he set up Wirth Research, a design consultancy company with a focus on motor sport.
The company has been a bit of a rebirth for Wirth's motor sport career, one that ultimately returned him to the streets of Monaco with his own team some fifteen years after the demise of Simtek. A director and share holder in the Virgin Racing team, Wirth has returned to the sport's biggest stage, with Wirth Research finally getting the chance to sink its teeth into a Formula One contender having established a reputation for itself in the US.
Virgin's philosophy is a little different to most other teams, while being entirely consistent with the minimalist approach Wirth seems to favour in his business ventures. While some teams, not to mention the Formula One Teams Association, carry on about cost reduction and controlling expenditure, Virgin Racing has been built from the ground up with those very principles in place. A lean operation, the development budget Wirth enjoys is a fraction of that spent even by their nearest rivals.
Much of that is a result of the approach that Wirth Research has adopted in the design process, heavily investing in computational fluid dynamics (CFD). "I like CFD because it has allowed my company to design championship winning sports cars," Wirth explains. "We have won pretty much everything there is to win in sports cars with one notable exception, and which we're working on at the moment, which is LMP1, the 24 hours of Le Mans."
It was while working on a sports car project for Acura in North America that Wirth Research made a discovery that set the wheels in motion for the company's move towards Formula One, though they didn't know it at the time.
"On the sports car front, the Acura sports car programme starting in 2006, we used for the first time huge scale CFD," Wirth recalls. "Because we were resource limited we were spending so much money wind tunnel testing. We were spending, in parts of the programme, probably $60,000 a day on wind tunnel model parts, and of those probably $54,000 of those bits ended up in the skip because they wouldn't be any good.
"For every ten iterations you do nine of them wouldn't work, it doesn't matter if it's bodywork or suspension or a wing. So we were haemorrhaging money, like everyone who uses a wind tunnel does, on parts, and so we set ourselves out a target to try and improve the resolution and fidelity of CFD to at least match a wind tunnel, but actually we exceeded it. We ended up learning things which enabled us to turn what was a reasonable sports car, the Acura sports car at the beginning of the programme, into a championship winning and record setting series of sports car which continues to this day."
A win at Le Mans in the LMP2 class by six laps for Acura in 2010 is strong evidence that the CFD approach has potential, and while its application is common among other teams already, Virgin Racing are the first team to have taken the completely virtual approach.
"It's basically the science of modelling airflow, thermal heat transfer and those type of things mathematically," Wirth patiently explains when asked for a simple explanation of CFD. "So basically it's a technology that enables us to simulate digitally what happens in reality.
"A traditional approach would be designing something from a drawing board, obviously everyone's moved on now and everyone models things in CAD, or computer aided design," he continues. "Everything that's on the car is modelled in three dimensions down to the nuts and bolts, washers that hold everything together; pipes, electrical cabling, etc. Literally what you then do is take that model of the car, or its components, and then scale it down to say fifty or sixty percent of its full size.
"With these wind tunnel models that are used to get aerodynamic data traditionally you would manufacture components of the car, depending on what they are, in various technologies. Quite a lot of the scale models, fifty or sixty percent, are made in composites just like the real car, because they run at quite a high speed in the wind tunnel and they suffer from stiffness issues and bits and pieces.
"There's actually quite a lot of design to actually manufacture, and manufacturing to manufacture, a component for a model. Sometimes wing sections that might be carbon fibre on the real car will be manufactured in aluminium, or steel, machined from solid - so you take a lump of metal and you machine it out and you end up with a small profile or suspension component or whatever. All these parts are built for a model, so you have a model design department which is like, essentially, a scaled version if you like of the real design office that manufactures real parts, and then these parts are assembled and tested in the wind tunnel as a wind tunnel model."
According to Wirth that's where a CFD approach has its benefits. Once the car is modelled in CAD it is transferred into a super computer which performs calculations simulating what would happen in a real world environment. By using a virtual car in a virtual world the team is able to use full scale components, whereas the FIA has restricted wind tunnel testing to sixty percent.
"When it (wind tunnel testing) first started everyone was using quarter scale models, and then they went on to third scale models, and there was a huge hurrah when everyone went to half scale and some teams could afford to go to sixty percent. Now the rules are you aren't allowed to go bigger than sixty percent; sixty percent is the regulation limit. But we actually test full scale which us why CFD is such an advantage.
"We don't use wind tunnels at all," he continues. "Your development is more efficient, more financially efficient and more performance efficient to do your, what we call heavy lifting development in CFD."
There's a catch however. It's not simply a case of loading the CAD model into a computer and going to make a coffee while the process runs, there's still a lot of work that needs to be done to get the model ready.
"The process is essentially the surfaces or shapes of the car, instead of having to manufacture patterns, moulds or full model components sixty percent scale we essentially create a mesh which is like, it's almost like coating - if you can imagine a car, coating it in chicken wire, but very fine.
"We're talking wheel spokes, brake callipers, all of these items exist in mesh and then basically that three dimensional surface mesh, out of that we grow, in all the areas we want to model airflow, we grow a whole new mesh which contains all the internal components of the car, the space between the car and the ground, and then away from the car into what we call the 'far field'. So you end up with a very, very, very big number of three dimensional points which represent lattice, if you like, which contain the surface of the car, where the air's going to flow. That mesh is basically loaded into a super computer and we solve equations of motion of the fluid, quite complicated ones, which describe how the airflow would go over millions and millions of three dimensional points. And that's basically how CFD works. It's a very, very mathematical process."
That reliance on computing power means the hardware Wirth Research uses scarcely resembles anything you'll pick up down at the local computer store. "It's in another universe," replies Wirth when asked how similar a home computer would be to the machines he uses. "People at home, who've had home computers, have followed the revolution of computers and we've all heard of multicore processors these days. At a certain point they got to a certain performance with a processor and then ran out of clock speed, so one of the ways of speeding up calculations in a computer is essentially to run the processor faster and faster and faster at a higher and higher frequency so it does multiple calculations.
"Essentially when they hit that limit the best way for the computer industry to increase power was to, instead of having one processor, have two. Now that's quite tricky because getting two processors to work together and splitting the tasks up, splitting the job up if you like, between two processors is quite tricky. And so processors ended up being what they call dual-core, and now quad-core and then you can get six-core and eight-cores. The number of individual processors on a chip is increasing. But if you expand that to a huge amount that's what cluster super computing is. We have a very, very, very large number of processors, each one of which is as good as, if not better, than the fastest processor you can buy today.
"To give you an idea of how much data we generate, people understand iPods, or people understand laptops, and a good iPod these days can hold sixty four gigabytes of data, which is a substantial amount; 10,000 songs or something like that. If you can imagine, and people know a good laptop is maybe two hundred and fifty gigabytes, every twenty-four hours we generate over one thousand gigabytes of data. That's the equivalent of filling thirty or forty iPods every twenty four hours, non-stop.
"We have a hard disk storage system, I'm not going to go into the size of it, but since the beginning of the Virgin Racing project we've generated five-hundred terabytes of data, which is half a petabyte. It's basically 500,000 gigabytes, or 5,000,000 megabytes of data. It is a monumental amount of data, and a lot of the technology at Wirth Research is around data processing and analysing the results, it's staggering."
Staggering it may be, but it still works out cheaper than using a wind tunnel, even with the world's third largest CFD installation. That in itself is quite an achievement, given the aerospace industry is perhaps the biggest user of CFD. What it means is that the boffins at Wirth Research have very nearly the same computing grunt as those that design space shuttles and satellites.
"If you look at an industry, our CFD installation, which is being upgraded again for 2011, according to our software supplier who are a world player who works on the CFD code with us, will be the third biggest CFD installation on the planet running their code. And if you realise they supply all the big players in the aerospace industry, space agencies, whatever, that's a pretty amazing statistic for a small supplier in Formula One. We're not a team, we're a supplier, so that tells you the level that F1 is at."
That's all well and good but the fact of the matter is that Virgin Racing were, according to the history books, the worst team in Formula One last season. They finished at the bottom of the Constructors' Championship with only a couple of fourteenth place finishes to show for a year of toil. The record books don't show the whole picture though, because the car was competitive and results at the bottom of the timing screens were as much down to luck as outright performance on a Sunday afternoon. If the Virgin car lasted the distance, - and it became apparent early on that it didn't have a big enough fuel tank to do - the best they could hope for was to be next best behind the established teams. If some of those dropped out Virgin Racing could pick up an extra spot here and there. In reality their fate was more in the hands of others than their own.
Still, to have a car designed purely on a computer, a car that had never seen the inside of a wind tunnel, competitive against their immediate rivals was an encouraging sight. "2010 was very much a learning year; understanding how we take and expand the level of CFD and industrialise it if you like," Wirth believes. "It's a commodity now and we have to compete with the fruit that wind tunnel testing is bearing. So we can't rely on the fact its just better than wind tunnel testing, we have to be able to do as much testing and try as many variations, and that's the challenge this year, but we're definitely getting there.
"The challenge for us was to come into Formula One with this technology with the greater difficulties you have with open wheel cars, but also what excited us about Formula One more was not trying to compete with the top teams on budget terms, but trying to do that with a limited budget. To give you an idea of the limitations of our budget our entire research and development budget, which includes essentially everybody's salaries, all of the R&D done, the aerodynamic development, all of that is... There are individual engineers in Formula One who are paid more than the entire budget of our R&D. It is tiny in comparison to other people but we believe it is appropriate to Formula One."
"Any technology born in the white heat of battle is far quicker than normal industry. That's true in war time, you think of what came out of the world wars in terms of engineering. There are technologies that came out of the space race in terms of the Apollo technologies."
The wind tunnel isn't dead however, even Wirth believes there's a need for the technology if only to validate data derived from CFD. In terms of Formula One, especially under the current regulations which limit teams to scale models, Wirth firmly believes his virtual approach is superior. "I think that the full scale wind tunnel definitely has a future, but in terms of scale model testing I realised some years ago that I didn't think it was an efficient way of working, and I continue to believe that.
"There will always be a need to have wind tunnels. If there wasn't a restriction in full scale wind tunnels where you can go in, validate things, you can build a real vehicle, see the real affects or the field, you can see the quality of your final surfaces compared to computer theory."
For the moment computer theory is all Virgin Racing has to go on. If 2010 was a learning year 2011 must be the year in which they act on those lessons and begin making strides up the grid. The team's first target will be knocking Team Lotus off their perch as 'best of the 2010 new comers' and securing at least tenth in the constructors title, while escaping the first phase of qualifying must also be high on the agenda.
Bottom of the heap last year wasn't such a disaster, though it will be if it's repeated in 2011.