CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Phoenics (http://www.cfd-online.com/Forums/phoenics/)
-   -   drag calibration for cube (http://www.cfd-online.com/Forums/phoenics/78042-drag-calibration-cube.html)

ga5aq July 11, 2010 09:48

drag calibration for cube
 
Hi all,
in order to simulate drag over my experimental vehicle, I wish to first simulate the domain and the turbulence model on a cube for which there is plenty of info on the web.
Cd for a cube at Reynolds10000 to 1000000 is 1.05

I am using the kemodl and an air stream velocity of about 11m/s
but I am far off at about cdx=2.3 in the result

I admit I haven't reached convergence and I am flying on 262500 cells and 4000 iterations, took me 12 hours:p.

Has anyone got any suggestions on what to change in order to correlate the simulation with the empirical cube results?
before I run it on the vehicle cross section.

Cheers and have a great week you all.

g.;)

gallant77_cfd July 11, 2010 22:37

Hi Gai

Do you use "drag and lift object" to get the drag force ?

Richard

ga5aq July 12, 2010 03:27

well I am now simulating it with almost 500000 cells and 4500 iterations, and it takes 23 hours

to answer your point
On earth I go through the domain settings and click on output, output of forces and blockage (ON)
and, in settings "calculate drag coeff. from total forces (ON)

then in the results file towards the bottom , under DRAG COEFF. BASED ON TOTAL FORCES" I have a couple of lines for cdx and cdy or cdz
eg.
Cdx = 2.389233E+00

Cdz = 6.366532E-01

I think that there is quite some turbulence behind the cube, and by inserting the probe within that turbulence, the solver finds it hard to settle down on some value, causing the graphs to go bonkers.

CFD is becoming an art!

G.

ga5aq July 13, 2010 04:25

I have done a further simulation
with a 1m cube within a wind tunnel, at almost 500000 cells and 4500 iterations and the values for drag are ridiculous 0.179 while it should be 1.05

I wonder if someone in this forum can suggest a possible solution for calibrating a KEmodel wind tunnel where the cube occupies less than 3% total cross section to have a value of 1.05 Drag coefficient!

Real values for a cube in Re=10000-> 1000000 is 1.05

PLEASE HELP ME!

Linh.Hoang July 14, 2010 05:37

Hello
Im not sure I can help you because I have an older Version of Phoenics and when I wanted to calculate the lift and drag coefficient I calculated it myself with the forces on the blockage (they were listet in the result file).
But first of all is the flow and pressure field right? What about convergence?

Linh

ga5aq July 14, 2010 06:22

not really, it hasn't converged, it took 24 hours to go through 4500 iterations and it had about 450000 cells

but I think that the major issue is because I have run the model as a 2d model, I will have a go with a 3d model with a bit less cells tonight or towards the end of the week and a KEmodel and let you know, or otherwise try with an ahmed model.

Thanks a lot for your kind help and interest.

if you wish I will e-mail you my q1 file for the previous case. let me have you e-mail address please.

Best wishes
Gai

ga5aq July 14, 2010 06:23

my e-mail address is ga5aq@yahoo.com

ga5aq July 15, 2010 08:18

succeeded!!! thank G-d!!!!
 
I 've finally managed to calibrate the model, now I can use it to simulate my vehicle aerodynamics.

it was just a small mistake, in the model setting

THE START:
I have set up the simulation as a 1 m cube within a domain of 15m x15m x 25m

the free stream velocity is 10m/s - in order to keep reynolds below 1E6 (1000000)

At first I have set the simulation with an inlet and oUtlet and the cube in the middle

it didn't work out well

then I have enclosed the whole cube within 4 plates resembling the walls of the wind tunnel and in/outlet

and it still didn't work out as Cdx was below 1 (Cdx = 1.116530E-01), while it had to be 1.05 for that reynolds number

then I have looked and found out that the reference velocity was mistakenly set to 30m/s, I have changed that to the free stream velocity (10ms) and after only 800 iteration (for 230000) , the values converged and the drag coefficient was :


Cdx = 1.004902E+00
Cdy = 1.708112E-04
Cdz = 2.131931E-05

I believe that further mesh refinement and appropriate wall functions can make up for the discrepancy from the original value of CDX 1.05

IF ANYONE WISHES TO HAVE THE Q1 FILES OR ANYTHING THEN PLEASE DO NOT HESITATE TO LET ME KNOW, I'LL BE PRETTY HAPPY TO HEAR VIEWS AND SHARE MY FINDINGS

by the way to set simulation in order to obtain the drag , lift coefficients
you have to click in "Domain settings" "OUTPUT" and then put "output of forces and moments on blockage objects" ON the reference velocity was in the settings tab AND IT WAS THERE MY MISTAKE!

thanks to all of you that helped in some sort or another.

Please do let me know if this has been useful to anyone.

G.


All times are GMT -4. The time now is 18:43.