CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   ANSYS Meshing & Geometry (http://www.cfd-online.com/Forums/ansys-meshing/)
-   -   [ICEM] Bad cells from O-grid in cylinder (http://www.cfd-online.com/Forums/ansys-meshing/99833-bad-cells-o-grid-cylinder.html)

Kristin_N April 12, 2012 14:53

Bad cells from O-grid in cylinder
 
1 Attachment(s)
Hello,

I have a problem with bad quality cells resulting from an O-grid in a cylinder and I wonder if there is a way to improve them?

As shown in the attached picture, I have associated the edges of the block to the circular curve in order to ensure that the centre remains "round" (I've been asked to do this for ease of post-processing). This seems to result in the poor quality cells. Is there a way to improve the cells whilst retaining the round O-grid in the centre?

Furthermore, is there a reason why the geometry retains the correct dimensions in length, but goes from a diameter of 0.01537m to one equal to 0.0153623? I would really like to fix/avoid this as the latter (wrong) dimension is being read by the CFD code.

Thank you in advance for your help,

Kristin

diamondx April 12, 2012 15:25

From what i see, you need another o-grid inside . can you attach tin and blk ?

Far April 12, 2012 22:40

diamondx has suggested right approach.You need another O-grid inside

Kristin_N April 13, 2012 07:00

2 Attachment(s)
Thank you for your replies Ali and Far.

I have attached the tin and blk files for the original model with the bad cells. I also tried adding a further O-grid without linking it to a circular curve (I assumed that's what you meant, but I could very well be wrong) and it does improve the cell quality, however I lose the circular shape I was asked to retain.

Could you please let me know if I have misunderstood and if there is another way to maintain a high cell quality with a "circular o-grid"?

Also, would you be able to let me know what bounding dimensions you obtain from my files? I just don't understand why the diameter goes from 0.01537m to 0.0153592m, but I really must avoid this.

Thank you,

Kristin

diamondx April 13, 2012 10:21

2 Attachment(s)
kristin, the o-grid you added is perfect. see my attempt in the pictures below. I don't know why you wanted to keep the circular shape at first place ? i didn't investigate the diameter problem, but hey if you have 0.01537 m, it equals 15.3 mm, the .3 is already very very precise. May be i'm wrong, my mistake if you are modeling nano-scale project but 0.37 mm or 0.359 mm will not change much in your solution in a regular large geometry. i could not get the ruler to get me distance between one point in the curve and one point in the center my icem is bugging somewhere? i hope i could hope...

Far April 13, 2012 12:47

1 Attachment(s)
Radius is coming out to be 0.01537m,no problem at all.

Files are attached.

http://img594.imageshack.us/img594/888/53232376.png

Kristin_N April 13, 2012 13:19

1 Attachment(s)
Thank you both Ali and Far.

When I mesh the geometry of the files I attached to the previous post, the min and max dimensions are only correct for the length. I know some rounding is to be expected, but as shown in the image attached, the limit values in the x and y direction are 0.0153558, as opposed to 0.01537 (and this value seems to change depending on what size of mesh cell I specify).

This wrong value is what's being read by the cfd code, which leaves me uncertain of where I am locating monitoring point probes. This is just a basic geometry, the one I need to use is a bit more complex, but with it changing these "bounding box" values I have no control over the interior ones either.

Please let me know if you can think of a way to obtain higher control of the dimensions used, I do need them to be precise.

Thank you,

Kristin

PSYMN April 17, 2012 15:19

Far asked me to take a look at this thread.

First, if you force it to map the center of the Ogrid to a circle, you are demanding that it have poor quality at the corners. Your solver doesn't need this, so unless you are actually trying to capture circular geometry at that location, don't worry about it. The Ogrid is trying to do you a favor, let it ;) If you are trying to capture some sort of circular geometry at that location, then Far is right about just putting another Ogrid inside to improve the quality at the "corners" of the circle.

As for the tolerance issue, yes, there are several things you can do to tighten the tolerance...

1) For the geometry tolerance... settings => Model => Triangulation Tolerance... Set this to something smaller and see the difference... the catch is that display and rotation times are also affected.

2) For the hexa meshing tolerance, you can have it project to the underlying bspline geometry instead of the "efficient" faceted geometry level as set by the triangulation tolerance... Go to "Settings => Meshing Options => Hexa Mixed => and turn on the option to "project to Bsplines".

3) You could also adjust "Settings => Display => Float Display Precision". Actually, we found a defect in this recently and improved it for 14.5 (it was being overruled by internal settings in some cases).


But seriously, as FAR and diamondx said, it might not matter. Are you expecting solver accuracy down to that many decimal places?

Kristin_N April 18, 2012 10:11

Thank you for your input PSYMN, my understanding of the O-grid quality improvement methods has now been confirmed.

Regarding the accuracy I require, I understand that it may not affect my solution excessively, but if possible I would prefer to import a mesh that has the min and max coordinates it should as opposed to slightly below the actual value and with increased decimal places.

I tried making the changes you suggested and reducing the triangulation tolerance has improved things slightly. I suspect that adjusting the Float Display Precision setting would solve my issue, but regardless of what number I change it to, it has no effect. Perhaps my icem v14 has the defect you mention? If so, is there any way I could prevent it from being overruled by the internal settings?

If there is really nothing I can do, I'll just have to accept the compromise. Since it's for my PhD project, I believe it's worth trying to achieve accuracy though.

Thank you,

Kristin

Shomaz ul Haq February 27, 2016 03:21

Quote:

Originally Posted by Kristin_N (Post 354554)
Thank you for your replies Ali and Far.

I have attached the tin and blk files for the original model with the bad cells. I also tried adding a further O-grid without linking it to a circular curve (I assumed that's what you meant, but I could very well be wrong) and it does improve the cell quality, however I lose the circular shape I was asked to retain.

Could you please let me know if I have misunderstood and if there is another way to maintain a high cell quality with a "circular o-grid"?

Also, would you be able to let me know what bounding dimensions you obtain from my files? I just don't understand why the diameter goes from 0.01537m to 0.0153592m, but I really must avoid this.

Thank you,

Kristin

Hi Kristin_N. Hope you are well. I wanted to ask how can I create an O grid inside the inner cylinder as you did. Would be grateful.


All times are GMT -4. The time now is 15:43.