CFD - Computational Fluid Dynamics, Motorsport, Formula 1

[TDH]

Yes I did everything like the tutorial said but I use ansys 14 instead of 12.

[Blackout]

I have a question. How do you call that 'technique' when you smooth the angles/add a radius to an angle to improve the airflow
(red in this rough illustration)


Has someone a better illustration ?

[stez90]

Fillet?

[Blackout]

exactly. thanks

[ajdracing]

Wow Russell, absolutely fantastic work. Its great to see what is really going on. I have been looking at one of the other forum topics - William rear wing - that is discussing rear wings. There are two items there that have been discussed.

Serated gurneys (supposed to give similar downforce with less drag over a regular gurney) and end plates with cut outs that follow the upper surface of the wing profile (suppossed to be more stable when the wing is in yaw but with a slight reduction in straight ahead performance, and apparently use in IRL and CART on ovals. scarbs thinks the cutouts are more to do with vortex management than yaw performance).

I don't suppose you could model the effect of these so we could see whats happening? I think the rear wings are two element affairs and not the three element one you have used here, making the endplate cutouts easier to form.

Cheers

Nick

I know this is old (on page 1) but I came on here looking for the answers to the exact same questions and i am unable to pull up the picture of the results. If someone could explain it to me or republish the results that would be greatly appreciated.

Thanks,

AJ

[Blanchimont]

Right now, i'm getting in touch with OpenFoam CFD ( http://www.openfoam.com/ http://en.wikipedia.org/wiki/OpenFOAM ).

It has the capability to generate a mesh around a specified geometry (.STL file) with the tool snappyhexmesh (http://www.openfoam.org/docs/user/snappyHexMesh.php).

So i need to create the geometry i want to analyse and therefore need a CAD software, that is able to export the geometry as .STL file.What opensource/freeware CAD software can you recommend for this task?

[Nando]

is CFD so advanced that it calculates individual molecules, or groups of molecules or does it work by other principles?

Where do they draw the limit?

[Just_a_fan]

CFD doesn't look at individual molecules, no. It looks at points in space or at volumes created by a mesh of points.

In general, the more accurate the model required, the smaller the volumes have to be. The smaller the volumes, the greater number of them for a given model and the greater amount of computational power that is required.

It is unlikely that CFD will ever look at individual molecules except in very restricted research e.g. combustion where understanding how individual molecules interact is fundamental.

[Nando]

Thanks

[Just_a_fan]

Additionally, it appears that the largest DNS (Direct Numberical Simulation) run ever was carried out on 4096^3 mesh points. That's 6.87x10^10 points. That required the Japanese Earth Simulator supercomputer to run it (the most powerful supercomputer in the world at the time).
http://www.cfd-online.com/Wiki/Direct_n ... tion_(DNS)
http://en.wikipedia.org/wiki/Earth_Simulator

22.4l of air at STP contains 6x10^23 molecules (Avogadro's number). So the DNS run looked at a number of mesh points equal to 2.5x10^-12 of a litre of gas.
(Note: DNS doesn't model molecules rather it solves the Navier-Stokes equations directly without using a simplified turbulence model so this comparison is purely for fun.)

We're going to need to wait a long time before we can do DNS on big things like cars...these are big numbers

We'll probably never need to carry out molecular-CFD on things like cars because the RANS/LES/DNS models will be "close enough". This is just as well as we'll probably never be able to do molecular-CFD on cars...the numbers are just stupidly huge!

[enginerd364]

One additional explanation, most of the methods we use to solve fluid dynamics rely on the assumption of continuum, i.e., the flow properties across an infinitesimally small volume are continuous. This assumption is valid for flow fields of non-reacting (low temperature) fluids, where the chemical composition does not change over time. To increase the "accuracy" or reveal additional details within the flow field, we simply reduce the size of the volume, which results in a larger grid. The only examples where tracing molecules would be necessary would be in high temperature or mixing flows, such as combustion or hypersonic flows. Even then, we still make assumptions with regards to molar/partial pressure/etc fractions of the amount of each molecule in the volume and calculate the properties accordingly. It is unlikely that we will see the application of molecular based analysis in anything other than specific fields of science. You will see a lot more work being applied to improving the current meshing/modeling techniques currently used before ever seeing molecular based physics applied aerodynamics.

[Just_a_fan]

Good points.

[Nando]

Great info, thanks!

[kosioBG]

All this CFD is so interesting, but so scary... i don't know where to start(i have no experience whatsoever) And i can only make 3d models with google sketchup, would that be ok or would i have to learn cad?

[KeithYoung]

You most likely need to learn CAD. I haven't used mainstream CFD in a while but typically they required the ability to import STEP or Parasolid files etc. I don't know that Google can make those files.