CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLOW-3D (http://www.cfd-online.com/Forums/flow-3d/)
-   -   Aerodynamics Problem (http://www.cfd-online.com/Forums/flow-3d/62643-aerodynamics-problem.html)

ck5285 March 16, 2009 05:49

Aerodynamics Problem
 
Hi All
would like to know if Flow-3D can be use to understand drag & lift coefficient in aerodynamic problem? eg: F1 car wind tunnel testing.

H.C March 16, 2009 18:38

it sounds doable in FLOW-3D. the issue would be to find the right cell size to resolve the boundary layer as the reynolds number would be very high and the boundary layer thickness would be very small. you might need to utilize the nested mesh blocks in the setup. the other thing is to make sure that the domain is bigger enough so that you can apply symmetry boundary conditions on the sides and the top.

ck5285 March 17, 2009 02:05

Hi HC
Have you tired with this problem before? do you able to share some success case of Flow-3D in aerodynamic problem?

H.C March 17, 2009 10:55

you may find some information at http://www.xceng.com/esempi_e.html
They did a case to calculate coefficients on an airfoil in FLOW-3D.

ck5285 March 31, 2009 20:37

Hi HC
apart from this supplier, is there any customer really used flow-3d for aerodynamic problem? such as car design?

stefmascio April 7, 2009 11:10

yes, we did some aerodynamic simulations with FLOW-3D on a solar car.
It was designed by an italian university to race at the solar challenge next year. Results matches pretty well with wind tunnel data.

ck5285 April 10, 2009 02:00

Stefmascio
can you send me the link for further information? btw, Flow-3D don't have stead state solver so the wind tunnel testing must took very long time to reach steady state.

stefmascio April 10, 2009 03:12

dear ck,
the work was published on an italian magazine, and I think there's no the online pubblication. You can have an idea of the magazine watching the pdf at the link http://www.assomotoracing.it/CARTELL...R&T_2008_C.pdf .
About the steady state solver of course if you have it you can solve faster, but even if you don't have it times are not so long as you imagine: if you go out from the office in the evining, in the morning you can have your results.

ck5285 April 10, 2009 09:17

Stefano
the link only have 2 pages of PDF file but i don't see any F3D analysis been attached. so, can you send me details on work been done with F3D? btw, have you requested the steady state solver feature from F3D?

stefmascio April 10, 2009 09:31

yes, the link was just to have an idea of the magazine, but I think there is not any online article. This magazine is just a printed magazine, not an online one.
About the work of the car we have some informations but they are in italian language, so I think they will not be much usefull to you.
We also asked to FlowScience about a steady solver, but I think it is not so easy to add it. Anyway we often make steady state simulations with FLOW-3D, and it is not so bad.
What are the purposes of your steady state simulations? Which kind of simulations do you do?

ck5285 April 10, 2009 09:48

I believe some car picture with F3D analysis will be helpful, so can you help to attach the link? i'm interested to understand capabilities of FAVOR is resolve complex geometry such as Car model. i'm basically do both transient & steady state simulation, others generic CFD code that provided steady state solver is singnificant faster as compared to F3D that iterate through transient method.

stefmascio April 15, 2009 03:14

hello ck,
this is the link of the project http://www.ideasolare.eu and this is the page with the pictures with FLOW-3D simulations http://www.ideasolare.eu/progetto.html .
Tell me what you think about them.
Basically there are no problems in doing simulations like this, and the FAVOR doesn't imply nothing less (maybe "more" respect other structured grids software): you just have to take care to put one cell totally inside your object, even in small thickness part. This can be the limitation in your cells for cases like this.
So accuracy should not be compromised by the FAVOR, just the trade-off with the computational time, as you mentioned.

(but on the other hand, if you make transient simulation also, you can have a quite faster simulation with FLOW-3D than other "steady-state" softwares. So, depending if you make steady state or transient simulations, FLOW-3D can be faster or slower than the others. But for transient process: faster.)

ck5285 April 17, 2009 01:49

Stefano
thanks for the sharing :)

ck5285 April 27, 2009 03:51

Stefano
i know F3D generally have challenge in compute draft force for aerodynamic problem as it can't correctly capture the boundary layer effect. so, how to you performance lift & draft analysis for this car model?

H.C April 27, 2009 13:26

My guess it to apply a force windows to include the object, and the force window will output the force components in each direction. From there, you should be able to calculate the life coefficient based on the force component in Z direction and the drag coefficient based on the force component in X direction. You may need to switch to force components in other directions, depending on the setup.

Its just my rants. Stefano would be the best person to explain the process.

ck5285 April 28, 2009 20:46

This is just an approximation and may not an accurate data right?

stefmascio April 30, 2009 10:23

HC is right, even if I prefer to use the GMO feature to have in the output the forces acting on the body. I found them more accurate.
The drag force is also computed quite well, the important factor is the mesh size because we have experienced a bit of mesh sensitivity on the forces: but if you develop your know-how on a specific problem on the right cell size then you can have very good forces!
The boundary layer matter also is relevant mainly when you deal with smooth shapes in a separation/reattachment regime, that usually it is not the "design" regime. If your flow doesn't have separations along smooth shapes but it keeps attached (except for sharp corners) then the FLOW-3D turbulence model works well too.

For that car the drag coefficient matches pretty well with experimental data.

ck5285 May 2, 2009 08:32

thanks for sharing. :)

ck5285 May 7, 2009 22:37

Stefano
what turbulent model that you are using for this simulation? Btw, how do you define the TLEN (turbulent mixing length) for this wind tunnel like simulation?

stefmascio May 8, 2009 02:48

the turbulent model is the RNG.
About the TLEN we found that the values of the forces (expecially the drag) are more affected by the cell size than the TLEN value. TLEN value only affect the drag if it is particulary small (making the drag smaller), let's say almost smaller than the mean boundary layer (that is very small!), otherwise we experienced not much changes.


All times are GMT -4. The time now is 16:20.