Is this just good technology over old school or just being cheap? Comments? (from F-1 live) Windsor revealed that US F1, like Virgin Racing, will not be relying on a wind tunnel to design the car. "I noticed the other day that Nick Wirth [Virgin's technical director] issued a press release about the new Virgin F1 car being entirely designed on CFD - Computational Fluid Dynamics (rather than in the wind tunnel)," he added. "The same thing applies to our car, although we see this as a logical process for a new team rather than something about which to be particularly excited."
I have read that Adrian Newey sits down at a drafting table and does the firsy sketches by hand, then it is created as a 3D model and analyzed before being made as a model for wind tunnel testing. I'm not sure about only doing CFD, but USF1 has access to NASCAR facilities for their complete car and we will see how it matches the predictions. If CFD-only design proves acceptable, it will go a long way toward equalizing the teams. The million spent in tunnels could prove to be a marginal benefit and no advantage for the big teams.
generally, I think this is true. well down the scale of diminishing returns. however, to do nothing (at least do wind tunnel time to benchmark your CFD data versus a known methodology) seems a bit optimistic.
I heard this as well. From what I understand, USF1 won't build a scaled down model to be tested in the tunnel (no time/equipment) but will test the full size car. I guess they figured they either don't have the historical data to compare a new design or it's a financial/time critical situation that they simply cannot do and get on the grid with any meaningful changes to a design after it has been tested. Should be interesting.
That's not a bad point to make. If you're going to be a back runner anyway, a few tenths improvement for the millions spent is not worth it. But, you would think that something could be learned.
They have access to one of the largest wind tunnels in the world located close to their headquarters in Charlotte. It looks like they will be putting it to use once they have a full scale model of the car. Anyone remember when Ferrari's wind tunnel was out of commission for a few weeks 2 or 3 seasons ago and how much pace they lost. While computer may be the way of the future I still think the wind tunnel provides something that a computer monitor cannot.
I do some FEA (finite element analysis) which is somewhat similar to CFD in a way I guess... thing with FEA is that while it can predict how a mechanical component will react in use, it is only to get you in the ballpark, it can't ever replace actual empirical testing. It's like predicting the weather... there are so many variables... the more you add, the more complex you make the model, and the more powerful of a computer is required to process the data. Williams bought a giant supercomputer a year or two ago to do just this. I somehow doubt USF1 has a supercomputer for this More likely they are running it on some guys desktop machine Windsor saying this is a normal thing, if that's what he said, is more than a little disingenuous. It would be like claiming "well, we built a great electronic model of our car in the simulator so we're not going to do any on-track testing because we can do it all on the computer". It would be a lame excuse...
The problem with CFD is the fact that it requires lots of computer processing power and time to run 1 single simulation. This is why Sauber has there Albert supercomputer to run the CFD simulations. With CFD you don't have to make models to run in the wind tunnel. On the down side if your CFD model is inaccurate or does not have enough detail the results will be worthless AKA Garbage in = Garbage Out With a wind tunnel you can make changes to wing flap angles, wing cross section,....... quite quickly. You can also yaw the model to see the reaction in real time. The down side is if your tunnel is not properly calibrated and your operators are sloppy you will get poor results also. Both methods are usable in designing a F1 car. Most top teams run CFD simulation to get the basic package down and then run the wind tunnel models to fine tune and validate the package. The real final test is on the track during testing and racing.
Nick did the same thing for sports prototypes with the Acura ARX-2A LMP1 sports prototype using only CFD and no wind tunnel. This car is a race winner in ALMS - for Duncan Dayton (Highcroft) and Gil deFerran and also in LMP2 for Adrian Fernandez (cars now residing in Europe with Strakka Racing) - so it can be done successfully! Carol
The trend is away from wind tunnel testing and to only CFD. It can be done, but it requires the right model, computing power and most importantly the right people with experience of what works and what doesn't. Wind tunnel testing is btw also not a guarantee that the new parts work as Honda once learned the hard way: They miscalibrated the wind tunnel data and were stunned that their new parts were a dog on the real track. USF1 skipping the wind tunnel part however doesn't shock me much: Yes, they are skipping costs and time, but then again if you're not fighting for the top, it doesn't matter as much. Most of the overall design is dictated by the rules and can be done by copying other people's cars.
bear in mind that they are sposed to have two complete running cars on track beginning March '10. They will need all the time left to achieve building them, let alone have a full size wind tunnel model. Plus when you read Windshear press releases they seem to try to distance themselves from KA and his claim of having built it and been a partner in it so maybe the $$ needed aren't available either to rent time there. The new teams are the only ones in '10 who can use full size tunnels, others are restricted and must use 60% models. Interesting that they show Emco as the gearbox supplier, teams must use one of the two suppliers picked by FiA, Emco isn't one.
Another thing I wonder is that if they have access to a wind tunnel to test a full-size car perhaps it makes sense to do that rather than build models. I would think the cost of making a set of molds and scale models wouldn't be that much cheaper than building full-size components anyway. They seem to be trying to build a car as cheaply and quickly as possible. This is probably a better approach than blowing every penny on a new building with carbon fiber floor tiles and titanium bathroom sinks, only to never actually produce a car because they ran out of money somewhere between the model making and wind tunnel stage of development. As an aside, I read last week that Bernie is doubtful USF1 will make the grid. Wouldn't he be trying to help them find sponsorship since more cars on the grid improves his tactical position with the manufacturers? I have to wonder if USF1 really is an experiment sanctioned by Eccelstone to see just how little money is required to produce a viable F1 car, and he can then use that as leverage to lower the budget cap for all teams once it is proven out. If the teams spend less on car development, the more they have available he can extort from them.
True enough, but consider last years P1 competition. Except for Sebring & Petit against Audi and the Pugs they only ran against themselves and privateers. (The P2 car is modified Courage not a one-off Acura CFD designed chassis, is it not?) You can get away with just CFD when you're a full-on works team knowing that the only competition you'll be facing in 9 of 11 races is from gentleman drivers in 4 year old equipment. F1 is much to competitive and precise to cut corners...but I'm sure PW will prove us all wrong when his slippery toasters set the highest trap speeds in Bahrain!
To be fair to Honda, they were not the only team to find that what worked well in the wind tunnel didn't work on track. I seem to recall most the teams suffering this at sometime. With more and more restrictions being put on wind tunnel testing, CFD simulations are going to become the main source of aero testing and the more CFD is used, the more accurate it will become. The problem with copying other peoples cars is that to make it work, you need to fully understand the aerodynamic principles, balance and concept of the original car. Look at when Toyota copied the ferrari F2003-GA with their TF104 in 2004. Whilst the top half replicated the Ferrari, the underside didn't and the car was a bit of a dog because the balance was all wrong. It's ok to use the basic shape of other cars as your basis as the rules dictate that they are all similar but If you start adding in detailed design elements from several other cars, you can easily end up with a car that is undrivable. I would say though that if there was any budget at all available for wind tunnel testing, I would spend at least a day on it, if only to do a comparison with the CFD simulation. If they match up, great!, if they don't however, you could have a serious problem.
Therein lies the problem with wind tunnel modelling: You can't just "spend one day in there" or you'll get results without baseline. Wind tunnels don't tell you whether a car/part works IRL but only how it performs under the parameters you're using. Very similar to CFD. If you get these parameters wrong and/or misinterpret the results, you will only confuse yourself more. Only track time gives you absolute answers. And even there factors like grip on that day with that configuration and that driver skews the results as well.
I totally agree, you need a number of days in the wind tunnel ideally, to sort yourself out, but if your budget is that small, would one day be better than none?. It would be a complete gamble that could throw you completely, but I'm not sure I would want to turn up at a pre-season test session solely relying on CFD simulations, without any physical evidence of what the aerodynamics are or might be doing (assuming of course that USF1 are planning on pre-season testing and not just turning up at the first race of the season to run the car for the first time!). The on-track aero testing could then guide you (to a degree, what with all the variables available on track), as to whether the CFD or the wind tunnel gave the more realistic results.(Hopefully both CFD and the wind tunnel results would be similar but you never know). It makes you realise just how much the people involved in the aerodynamics departments earn their money when you start looking into it!.
This is true as the larger meshes require greater processing power. This is because even simple components effect the entire car and as such parts are not usually analyzed on an individual level. Super computers to run a sim are beneficial but for start ups I'd imagine they'd lease time/cycles from an existing computer. This is somewhat true but wind tunnels are only as accurate as the calibration as well. McLaren's early season problems were due to an improperly calibrated wind tunnel. This is why many of their mid season developments were used just after running CFD simulations. In terms of changing flap angles, these things are easy enough in FLUENT as well assuming your model was made intelligently. Aero development is different from any other development on the car in the sense that there is no closed form solutions. For example, if the desired goal is a wing with a l/d ratio of X and minimal disturbances downstream there's many ways to accomplish this and no simple formula. It's a very empirical field. AoA can be changed, profiles, gurneys etc. In terms of CFD models vs wind tunnel models it's of trivial difference. A CFD run requires a CAD model which is really only a step or two away from a physical model for a tunnel.
Maybe I'm naive, but to me it doesn't look as if there is much to decide on regarding the aero shapes if winning isn't the goal. About 80% of the cars look identical to me, partly defined by the rules, partly by aero. Given all the challenges USF1 is facing to make it to the grid a perfectly modeled front wing (as an example) is probably a minor issue. Besides: Just wait a couple of races and then copy what works best from the winners. Many teams have done that approach in the past.
