Dear All,

I have studied drag coefficient of a 2D ellipse form numerically. The Reynolds number applied is about 2.5 million and the coefficient is about 0.1 for L/D=6.0.

I do not have any comparisons with other data whether my result is correct. Through textbook, I can only find drag coefficient up to Rn 0.2 million which is about 0.3. If I extrapolated, the coefficient would be about 0.1 and 0.15 which is close to my result. However, I am not really sure if this is correct.

I wonder if anyone does have any data or information about this.

Thanks.

Ketut Utama

You may validate your calculation first with a circular form, there are plenty of experimental measurements. And I was told that the 2D model should overpredict the drag due to the three-dimensionality in the wake.

First, is the longer side of the ellipse along, normal or at an angle to the freestream? Flow along the long side will most probably not separate (it's streamlined) whereas the one normal to it will separate at the two ends. (In this regard checking the results against flow over a circle, while a good idea, is not going to help much).

Second, at such a high Reynolds number wouldn't you expect the drag coefficient to be (quite) independent of the Reynolds number? I'm curious what data coordinates you used to extrapolate? That is, are you taking a point at say Re=100 and another at Re=.2M and extrapolating, or are there sufficient data points (and most importantly) near the desired Re number to use for extrapolation with reasonable validity. In general, extrapolation is not a good idea unless you know something about the physics of the problem.

Third, I have to agree with Tony Chen that you are discounting 3-D effects, which will change the value of the drag considerably if taken into account.

Fourth, the convergence of the solution and the turbulence model you are using should obviously be verified.

Adrin Gharakhani

[MachZero]

I am trying to simulate a 3d ellipse at several angles of attack from 0 to 90 degrees. At low angles of attack I am getting reasonable drag values but at higher angles above about 30 degrees, my results are almost exactly double what we saw in the experiment. Is it possible fluent is incapable of predicting drag on such a highly separated flow? I'm using kw-sst for turb model.

[truffaldino]

What is your reynolds number? At low reynolds numbers (below about 10^6-10^5) rans turbulence modelling cannot do accurate predictions even for slightly separated and/or transitional flows . In such situations 100% error is not an exception but a rule.

[MachZero]

Thank you for your response. My Reynolds number based on x is right in that range. Is the inability to predict the drag coefficient due to the fact that the BL is transitioning along the chord and therefore it is inaccurately predicting separation location?

The weird part for me, and see if you can understand/explain this one, is that I am getting very reasonable values when I run it in steady state, values that agree with experiment. But when I switch it to transient, the values roughly double. I have timestep size such that it converged 2 order of magnitude on residuals in less than 20 iterations. Additionally y+ = 1.
Thanks in advance!

[90nash]

Hello All,

This is a relatively old thread. Sorry to dig it up!.. I am working on non spherical particles (ellipses) and was wondering if there are any correlations for drag on ellipses (2D). I was able to find some for 3D. Can anyone please direct me to an appropriate article.

Thanks.

[MachZero]

I have some NACA data for wind tears done in airfoils but I'm not sure that it's 2d? What do you mean you want 2d data

[90nash]

Hello MachZero,

Thanks for your reply. I am actually working on collision of ellipsoids and wanted to check my code for 2D first, restricting the rotation axis only perpendicular to plane since the equations become a lot more involved in 3D. To this effect i wanted to see if there are any equivalent drag models in literature for transport of ellipses 2D.