[mecarlg]

Hi Everyone,

I've been using Fluent for years now and I have seen a bug in the software - it cannot find a sensible center of pressure (COP) for bluff body aerodynamics. I have seen this multiple times and wondered if any body else has (a) seen this as well and (b) found a workaround?

To give an example, I've got a case of flow past a bluff body in ground effect. The lift and drag values are sensible and it stands to reason that the COP is somewhere within the volume of the object. However, the COP coordinates, projected onto the symmetry plane (it is a symmetric vehicle) is outside of the object volume which is physically incorrect. In other cases, I have found the location to be on the order of kilometers behind the object or even underground!

I have looked at the COP, per component on a bluff body. Most of the COP coordinates give a sensible value. However, for surfaces which are exactly parallel to the x-axis (direction of free-stream) the vertical component of the COP gives impossible values of 10e+11 meters! Likewise, surfaces which are exactly perpendicular to the x-axis have a horizontal COP coordinate which is totally wrong.

Attached is an image of the COP coordinate per component and the non-physical values are highlighted in red. With this issue being for surfaces which are at 0 or 90 degrees to the flow direction, it seems like there is a bug in Fluent when calculating COP coordinates, something wrong with matrix calculations perhaps?

I'm trying (and failing) to use Ensight to post-process for me but I'm new to this software and I just get error warnings about COP calculations only being possible on 2D objects.

Has anybody else seen this problem? I found an old post on here from 2012 with a similar complaint to mine but nobody ever replied.

Suggestions very welcome!

[LuckyTran]

If the COP lies on the line (for 2D) or plane (for 3D) about which the moment is being calculated then that's what happens. And this happens when you have a perfectly flat surface that is perfectly parallel to the moment generating line/surface.


What it means is you have a degenerate case where the COP is not a point but an entire line or an entire surface. The bug is your problem setup. The rest is algebra.

[mecarlg]

Hi thanks for the reply. If the bug is my problem set up, how do I rectify it? I thought about rotating the domain by a few degrees so that there are no surfaces parallel or perpendicular to the x, y and z unit vectors. That does seem a bit extreme. Any other suggestions welcome...

In other cases I have looked at, I get perfectly reasonable force measurements but when the COP is a few km underground, that's not helpful!

[LuckyTran]

You are asking me to come up with a solution as to why you cannot divide by 0! The problem is you are trying to divide by 0 and wondering why division is bugged

Division by zero is senseless no matter the numerator. 0/0, 1/0, and a few km underground divided by 0 are all equally senseless. None are better than others.

[mecarlg]

Thanks for the reply LuckyTran.

To be fair, it's not me dividing something by zero, I'm relying on a tool in the software to give me an answer! However the number comes about, the unrealistically large (or small) numbers for certain components massively skew the COP for the vehicle as a whole.

Unfortunately, I can't remove these surfaces from the overall COP calculation. For example, the whole rear face of the bluff body. Taking that out means my answer is junk. It is also junk with it included.

To try to find the longitudinal COP coordinate (x), one method is to guestimate the vertical coordinate (z) and systematically calculate the pitching moment from nose to tail. If you plot the moment vs x, where the moment changes sign coincides with the COP. This has worked for me in the past but not always. I'm putting this here in case it helps anybody.

If anybody has any tips for how to find COP in Ensight, any help would be welcome. If nothing else, I would like to sanity check Fluent.