CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   Drag coefficient is 1000 times smaller ! (http://www.cfd-online.com/Forums/fluent/40455-drag-coefficient-1000-times-smaller.html)

 Cyril April 5, 2006 10:45

Drag coefficient is 1000 times smaller !

Hi everybody !

I'm trying to calculate the Cd of a 30mm shell (2D axi).

A "hand" calculus gives me 0.44, and Fluent gives me 1000 times smaller : 0.0004!

I created my geometry with gambit, and changed the units with fluent ("...was created in mm"/"change unit"/"Scale"). Is there a special setting to obtain a good Cd ?

(I guess that if the Cd is 1000 times less than the reality, the other parameters sould be wrong too !)

Thanks !

 Jason April 5, 2006 12:19

Re: Drag coefficient is 1000 times smaller !

Check your reference values (Report->Reference Values).

Jason

 Anindya April 5, 2006 14:28

Re: Drag coefficient is 1000 times smaller !

Use the proper values of Reference Area and Reference Velocity to get the proper values for drag or lift coefficient. By Default these values are set to 1, so you should change them.

 Cyril April 6, 2006 04:12

Re: Drag coefficient is 1000 times smaller !

Should I compute from my "inlet" (pressure far field) or from my "shell" (wall) ?

Cause I don't know what these "Reference Values" means... I know it should be a basic knowledge, but I'm a quite new Fluent user :)

Thanks !

 Joe April 6, 2006 06:39

Re: Drag coefficient is 1000 times smaller !

These values come from a physical understanding of the problem. For Cd and Cl, the drag and lift coefficients, you need to know the density of the flowing fluid, the cross-sectional area of the bluff body (or whatever is the object), the depth (which should be set to unity for 2D computations) and finally a reference velocity scale (say the average u velocity or the u_max at the inlet). The rest I think can be safely set to zero, unless you're calculating moments etc.

Please search this forum before posting a new question. I see that this question has been answered quite a few number of times. Hope that helps.

Appendix: Cd = Fd / (0.5 * rho * A * u^2) Cl = Fl / (0.5 * rho * A * u^2)

Fd, Fl are the drag and lift forces Cd, Cl are the drag and lift coefficients rho is the fluid density A is the cross sectional area exposed to the flow (which can be set equal to the side length say for a 'square' obstacle in a channel) u is a characteristic velocity scale in the problem.

Personal rant: This is one of the things I totally hate about commercial codes like Fluent/CFX/COMSOL etc. They try to make life easier in their own way, and the person using it gets the impression that things are indeed as simple as they seem to be. Only after having gone through a rigorous CFD course where I had to literally code the entire problem (from dicretization to solvers to boundary conditions etc.), did I realize how 'locked' my knowledge was. Without Fluent, I was quite helpless. I still am, to a large extent.

I find myself faced with two choices. Either I choose to use Fluent for all my work and disregard any other compelling factor, thus seemingly saving on time, which the astute will quickly realize is the most irritating albeit important factor; (or) I wean myself away from this commercial convenience and take the hard route of coding for every problem I wish to solve.

 Joe April 6, 2006 06:40

Re: Drag coefficient is 1000 times smaller !

My apologies. I should have put in an extra carriage return for clarity. The relations for Cd and Cl are as follows:

Cd = Fd / (0.5 * rho * A * u^2)

Cl = Fl / (0.5 * rho * A * u^2)

 Joe April 6, 2006 06:42

Re: Drag coefficient is 1000 times smaller !

Darn, *Enter* the *Enter* key

Fd, Fl are the drag and lift forces

Cd, Cl are the drag and lift coefficients

rho is the fluid density

A is the cross sectional area exposed to the flow (which can be set equal to the side length say for a 'square' obstacle in a channel)

u is a characteristic velocity scale in the problem.

 Cyril April 6, 2006 07:30

Re: Drag coefficient is 1000 times smaller !

Thanks Joe.

I've already looked for this question in the forum, but I haven't found any thing...

Anyway... you said I should set, among the others, a bullet cross section area reference and an other about the velocity.

As I'm studying a static bullet in a moving athmosphere, sould I compute from the inlet of the shell ?

Thanks.

 Anindya April 6, 2006 09:54

Re: Drag coefficient is 1000 times smaller !

Use the velocity and area that you are using for your experimental calculations.

 Cyril April 6, 2006 10:02

Re: Drag coefficient is 1000 times smaller !

Hum... excuse me, but I do not understand...

I don't have any experimental calculations. I only have fluent, with an inlet, an outlet and a wall (my bullet).

When I set my reference values, I've got to compute from...? I would say from inlet, but I'm not shure...

 Cyril April 6, 2006 11:50

Re: Drag coefficient is 1000 times smaller !

In fact, I tryed to compute from every boundary I have (inlet, outlet, wall). The references values are always the same (or nearly), and each iteration gives the same Cd : 0.0004 (it is about 0.4 in reality !)

Could it be the boundary conditions ?? (pressure far field with gauge pressure=101325Pa, operating pressure=0Pa, inviscid or SA, ideal gaz, coupled explicit 2Daxi)

plzzz help me !

thanks !

 Anindya April 6, 2006 13:49

Re: Drag coefficient is 1000 times smaller !

How do you know that Cd is 0.4 ? See the book or paper from where you got that result and find out what values they have used to normalize the Drag or Lift to find the drag or lift cofficient. They must have stated what they have used for Velocity and Area. You should use those. Running Fluent blindy would not give give correct results. You have to understand the underlying physics to perform simulation correctly.

 Cyril April 7, 2006 03:04

Re: Drag coefficient is 1000 times smaller !

In fact, I calculate the Cd my self, using this theory : http://www.earmi.it/balistica/coefball.htm and http://www.earmi.it/balistica/formi.htm. (sorry for the italian, but that was the only one "well explained" I've found)

Anyway, I might be a little wrong about my Cd calculation, but Fluent IS wrong : 4e-4 is too low for a supersonic 30mm shell (airborne guns on Mirage2000 or Rafale).

 All times are GMT -4. The time now is 16:10.