EC83 Wrote:
-------------------------------------------------------
> chet Wrote:
> --------------------------------------------------
> -----
> > Read above. More reliable to a point.
>
> So CFD is more reliable than a windtunnel? :P
An Alfa is reliable... to a certain degree or point in its life, but a BMW is far more reliable past the point the Alfa starts becoming a mess. Once CFD passes say the 50,000 mile mark a windtunnel becomes a much safer bet. Poor analogy I know
!!!
For simple things it takes less work with CFD to get reliable readings than with a windtunnel, however with complex things it takes less work for a windtunnel to be more reliable than CFD. CFD is more reliable if you look at a single aspect. Not the whole picutre
!
> By the way, at no point did I say windtunnels are
> unreliable. Don't put words in my mouth.
>
Appologies for that.
>
> Both techniques depend on human input, and since a
> session in a windtunnel involves the input of more
> people than the running of a computer programme
> does, the potential for human error is naturally
> greater, hence the greater practical potential for
> things to go wrong with the windtunnel process.
> Now who should've read above?
The human side of CFD requires more input than a windtunnel. GIGO
! Windtunnels require more maintence but on the most part much of that maintenence is ensuring the ideal flow conditions are achieved, in which the windtunnel has various devices to aid that.
With a windtunnel you naturally get as close to as real boundary conditions as you would in real-life and you can very easily (far more than CFD) adjust various aspects of the windtunnel to get the ideal conditions in a short time. Say such as at different attitudes (pitch, roll, yaw, rideheight). For CFD that kind of process takes a long time (you need to solve for one pitch angle, another, then another, and so on...). Windtunnel results are pretty much realtime. Much of the windtunnel work done by teams is mapping aero at different attitudes whereas the CFD is mostly testing various parts then possibly shoved into the tunnel. Best of both worlds i guess one for research one for validation. Using CFD to quickly model an update and see how it runs, then use the windtunnels realtime data gathering to map the aero and correlate. An hours run in the windtunnel testing different settings of say yaw could take days to solve in CFD. But essentially the windtunnel will give you the conditions, however with CFD you have to model the conditions. The way I see it theres alot more chance for human error to occur with CFD as it requires signifcantly more data input from the user. However we do not need to mention the idea of a incorrectly calibrated windtunnel.
As you'd expect the reliability of CFD depends on cost, and how much time you want to spend on things and at this stage I guess the limiting factor is time. Especially when despite a higher cost you could get results from a tunnel relativley quickly. Motorsport is analysis, results and application. All these needs to happen in a very short time. An example a guy I used to live with in the first year completed his placement at the then Honda last year. He said the one thing that amazed him was how he would be asked to draw something on computer, it would get a quick CFD run, a prototype would be made, then in the windtunnel to validate what they saw on computer, then to the track all within a week.
"Trulli was slowing down like he wanted to have a picnic" LOL