Low Drag coefficient! :(
 

Hi all!

I am trying to do find the Drag on a small underwater marine vehicle. Its .8 m in Length and comprises of two bodies: One ellipsoid on top, and a cylinder below it.

The total area is .8 sq-m and the speed at which this vehicle is supposed to travel is 0.4 m/s. Now when I solve for this vehicle using the K-epsilon realizable model for the X component of force on the vehicle, I get really low values for the force and the drag coefficient( 4.4 N and Total coefficient(pressure + viscous) as given by FLUENT is .0054).Could you please help me out with what am I doing wrong?

Also, when I had a look at the center of pressure of the object, the solver showed me a value which was like REALLY far away from the vehicle. Would you please please please help me out with this?

Thanks a lot.

DT

Anyone? Please :|

Never mind the center of pressure problem. I got what was wrong there. But can anyone please help me out with the drag coefficient? What could I be doing wrong? Should I increase the roughness height on the hull? (Its already quite high.. 0.8 mm) Please! Any advice would be more than welcome.

Thank you.

I tried increasing the roughness height, decreasing the roughness height, but the forces dont change by much. Please please help me out. What am I doing wrong? The solution converges everytime, but the forces are not different at all. With a roughness height .0005 mm, the Cd was .067 and with roughness height .005, it remained almost the same!! Please please help me out. :( What am I doing wrong over here?

Hello,

just to avoid unnecessary discussion, you know that force and moment coefficients use the reference values in the report/reference panel when evaluated? They should be consistent with your actual physical problem.

cheers

Hey Max,

Thanks for the reply. Yes, the reference values are in accordance with the problem.

Is there anything else that I can look into? The forces the solver reports after convergence, the viscous part in them is VERY low. Only about .5-.7 Newtons. This can not be right. Is there anything else that I should check?

Thanks a lot in advance.

On second thoughts, as the vehicle is totally gonna be submerged totally in water, and its gonna travel at a low velocity, around 0.5 m/s, so that is like a Re of 3*10^5. So shouldnt the drag coefficient be low? As it will only comprise of the viscous part? And by the ITTC line(can I use it?), the drag coefficient comes out to be of this order approximately. Please please correct me if I'm wrong.. Is a drag coefficient of .02 for a body (an ellipsoid and a cylinder) travelling at 0.5 m/s totally submerged in water, OK?

Please please help me out. Someone..also, this is the link to the velocity profile which I get for the vehicle after the solution has converged.
http://picasaweb.google.com/lh/photo...eat=directlink
As you can see, the wake region is very long..is this normal? And also, do you think I should use a longer flow domain? Could that be causing the problem?

some thoughts to ponder...

drag on an axial cylinder is Cd ~1 for wide range of Re (check Hoerner's book on drag, or White, Introduction to Fluid Dynamics). Elipsoid can be low ~ 0.2, but you are getting unrealistic values. I would not expect your contraption to have a very low drag value.

As mentioned by coglioni, are you sure you understand how to get from the force to the coefficient and vice versa. perhaps show your equations.

It looks as if you are running a 2D case, this would (in my opinion) be a VERY BAD idea for the geometry shown in the Picassa picture. You need a 3D simulation. What is your cell count?

Start with a cylinder (2D) or sphere (3D) where there are numerous benchmarks and see how your answers compare. This might help you to gain confidence with your case.

Your simulation domain is too small. I would suggest at least 3-5 times your characteristic length L (e.g. height of vehicle) upstream and 10-15 diameters downstream. Upper and lower boundaries should also be more than one L removed from object (unless this geometry represents a physical situation). Again, running with a known configuration (e.g.sphere or cylinder) will help to test sensitivity of answers to the extents of the domain.

I lack the time and motivation to expound on the near wall meshing and turbulence modeling. Bluff drag is one of the places where CFD tend to fall over very easily because it appears as if it should be a simple problem...

May convergence be with you.

Hi Mystic,

Thanks for taking out time to reply to my post. :)

I dont think I can increase the domain size, because as it is, I have 2.1 Million cells and when I go up from this number, FLUENT starts giving me "malloc" faults. And btw, I am running a 3D case on the vehicle. And I think that I'll do this drag simulation for a sphere, as you suggested, because that might help me find out whats wrong.

But I think you should have a look at this thing.
Autosub: An autonomous unmannedsubmersible for ocean data collection
Search for this paper on IEEE. OR I can send it to you if you dont have access. I was going over this paper yesterday, and its given in the paper that AutoSub is a 7 m long AUV, 0.9 m in Diameter, a velocity of 2m/s and its mentioned in the paper that the thruster, 5 bladed, is "able to provide the required 130 Newtons thrust." Dont you think that the Cd for this is pretty low too? Please help me out.. I'm so confused. I know this is an unrealistic value, but what about the AutoSub paper?

Thanks in advance. Will wait for your replies..

Hey Mystic,

I tried modeling the Sphere like you suggested me to. I looked up on the forum and came across this thread.
http://www.cfd-online.com/Forums/cfx...g-too-low.html
So I thought why not model the same problem: For an inch diameter sphere,in airflow at 60m/s, I got a drag force of 0.1 N when it should have been .46 N. :(( what did I do wrong? Why the low drag coefficient?

BTW, did you have a look at that AutoSub paper? I did, just to be sure. And tit says that the thrust required by that 7 m long AUV,0.9 m in diameter, while traveling at 2 m/s is 130 N. That means the Cd is pretty low for this vehicle too, right?

Thanks in advance.

You might have to trust your answer as being correct. I had a look at this page:
http://en.wikipedia.org/wiki/Drag_coefficient
(don't have textbook at hand for the moment).

they list many drag coefficients, and all aeroplane bodies are in the range that you first reported, i.e. Cd = 0.02 - 0.1, so a minisub should be there ot thereabouts.

The 2million cell limit means you are running 32 bit, i would really urge you to try and get onto a 64 bit platform which would allow for a larger mesh to test sensitivity.

hope my previous post did not lead you too far down the wrong avenue.

Are you freaking kidding me? Thank you for your replies. I love it when people actually try and help others over here :) And yes, I think you are right. The answer, it maybe right. And in the lab I'm in, they dont have 64 bit platforms so theres no way I can have more than 2 million cells right now. I have to run the whole thing on my computer and it almost kills it. :|

Any suggestions for the flow around a sphere? I am getting 1/5 th of the answer I should get.

Thanks again. :)

did you include roughness on your sphere model?

according to my "very reliable" wiki page, for a smooth sphere we have:
Cd = 0.1

this equates to F = 0.11N if I use V=60m/s; d = 25.4mm; rho = 1.225 kg/m3; mu = 1.8e-5 Pa.s.

values of Cd = 0.4(5) for rough spheres... would give you higher force in line with what you seem to be expecting.

you might be right again :cool: - i found that SST-kw and reliazible ke gave quite accurate drag results for bluff body work that i did in the past.

Aah..yes. About that, what range do these values(roughness height, roughness constant) lie in? I mean how do enter the value of the roughness height? Is there any page on the internet which gives the values of these for different materials?

Thanks. :)