CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   FLUENT (https://www.cfd-online.com/Forums/fluent/)
-   -   Meshing a Sphere (https://www.cfd-online.com/Forums/fluent/115316-meshing-sphere.html)

 elina March 28, 2013 04:40

Meshing a Sphere

1 Attachment(s)
Hi Friends

I have to simulate a heat transfer problem in Fluent for that I have to use a Sphere where I have to do a structured meshing...so that number of elements are very less..Can anyone tell me how to do that...I guess since its having a contact point I have to edit the geometry...But I dont know how to do that..Can anyone tell me how to do this....
I have the picture of that....

 oj.bulmer March 28, 2013 08:54

Have you tried meshing, do you get any errors?

Here is a thought, try running a small cylinder from one sphere to another, such that the axis of cylinder passes through the point of contact. Make sure the diameter of cylinder is small enough for you to be able to mesh. This way, you have avoided bad mesh at the contact point, yet creating a contact.

OJ

 elina March 28, 2013 09:41

Ah I dint try meshing using ICEM CFD but I am using ansys workbench , so I hope I can use FLUENT and in that icem Interactive will be helpful...but yes I will try that also some how..Ah then I have to give that in all the other faces also right..because here you are seeing just a one quarter of the sphere since i considered symmetry..so in the other side ther is supposed to be another sphere but though I dont have to draw that still that small cylinder have to constructed right..? ..and again can you give me an idea to mesh it like small concentric spheres one inside the other..becoz there will be conduction in spheres and hoping that we have to consider just the first small layer rest inside the core we can make it a say 3 to 4 element meshing...thereby reducing the number of elements in the solid region and have to see the impact of that number of elements in the outlet temperature....

 oj.bulmer March 28, 2013 10:22

I think you have created two threads on same subject. Can we keep the discussion only in this thread? The two spheres are not concentric, so why do you mention about meshing them concentric?

Since only two spheres are touching, there can be only one point contact. Hence we require only one cylinder.

As I asked earlier, did you try meshing as it is, using ANSYS meshing package in workbench? If it fails there you may explore the cylinder approach, just to mitigate meshing problems.

OJ

 elina March 28, 2013 11:19

Ah yes I hope so...
Yes I tried meshing but that is not what we require..My professor asked me to mesh it in such a way like very less number of elements and also structured meshing ( Hexahdral mesh probably) and the concentric meshing what i talked about is for the spheres..Like the each spheres has to be meshed like say 3 or 4 layer where conduction on the spheres talks place and also thereby we can achieve less number of elements in the sphere..What he suggested me is to have say 4 or 5 elements (ie) split the sphere into say 5 or 6 elements hope you got it now.....
The ting is I am doing for 6 layers of spheres and the thing is there will be 1000s of such balls so in simulation the number of elements to be reduced and computation time also..And hence they were telling to have very less number of elements for both spheres and fluid region and also have to check the result(Outlet Temperature) based on the number of elements...where according to my knowledge after some number of elements there wornt be any change in the temperature of outlet....If I am correct.

 oj.bulmer March 28, 2013 12:48

I am not sure if you can have structured meshing in the region where spheres reside. It would be tricky. But it is possible to do structured meshing easily in rest of your pipe. You can cut the pipe before the two spheres such that you have pipe without spheres - which can be hex, and pipe with spheres with tet. Have a look at meshing tutorials to know how to achieve this

If you want to divide/cut the sphere to have more elements, it should be easily possible with Design modeller in Workbench.

OJ

 elina March 28, 2013 20:39

Hi Sir
I hope you saw my Geometry in that the contact point between the 2 spheres has to be edited to get a good structured meshing i hope so....and the thing is I want the symmetry case so I dont want the entire sphere instead just a 1/4th of the complete sphere ,which you can see from my geometry picture..And then I want the sphere portion to be meshed in such a way that it has only very less number of elements like ( the meshing to be done like one layer after the other ,so say consider the entire sphere has 3 layer of different thickness and each layer is considered to be one or 2 elements or cells...) hope you got it...and then fluid will be flowing from the top and have to check the outlet temperature ,and based on changing the number of elements in both I have to check the outlet temperature which tells about the grid interdependent study I hope so....So this is the basic aim..And also I want to change the diameter of the sphere so based on that I have to see the results ..but the elements and all should be the same ( so it may be called as giving the scaling factor ,so that I can get the results for different dimensions..) can you tell me that also how to do..kindly help me ..its urgent..:(

 oj.bulmer March 29, 2013 11:25

Can you attach the Workbench project file (along with the folder) zipped ?

OJ

 elina March 29, 2013 11:44

1 Attachment(s)
Hello Sir

You just give me your Email Id i will send you indetail...but anyways here is my files...jsut have a look and email me @ elinamathew@ymail.com

 oj.bulmer March 29, 2013 13:01

You'll need to attach the wbpj file as well, otherwise the project can't be opened.

OJ

 elina March 29, 2013 13:40

1 Attachment(s)
sorry here it is...

 elina March 30, 2013 08:02

What happened did you check the files..? any idea..??

 oj.bulmer March 30, 2013 08:50

Quote:
 What happened did you check the files..? any idea..??
Patience is a virtue!

I have tried a bit. There are two cases, one with a small patch of cuboid between two spheres to create contact, one without a patch. I have decomposed the spheres with two layers, and putting few elements for each layer. You can use the one that suits your criteria. You may want to refine the mesh sizes to your liking and put inflation etc as needed.

I don't clearly understand what you'd achieve with coarse meshing on spheres. Also, the assumption of symmetry doesn't quite seem correct since the geometries aren't exactly symmetric about those symmetry planes.

The workbench project: http://www.gigasize.com/get/xzmod5tzk5b

OJ

 elina March 30, 2013 16:33

Hi Sir

I am not able to download it....And yes so you are telling that I wornt get good heat transfer results if i give less elements in the spheres..?? And yes even the geometry symmetry is bit confused..:) And can you send the file to my Email Id..?? Its not able to download..and another thing is can you give me a small explanation how you did..becoz I have to do like the same for 6 balls kept one above the other....And kindly tell me if there is any chance to change the diameter of the sphere so that I can see the different results without doing anything with the meshing,,....Like some scaling factor or something,,,,

 elina March 30, 2013 17:12

1 Attachment(s)
Yes I donwloaded it...the thing is I hope you created a new sphere inside the bottom sphere...and revolved it..I dont understand why you did that for...?? And our aim is to reduce the number of elements and get the good result..and hence we needed structured meshing..as hexahedra...and the spheres are to be meshed like layers of hexa nd not that it has to be created as layers...just the meshing has to be done..and geometry editing is not a problem..you can give a contact surface as you gave in it...then only we will get good mesh I heard that...and all the faces outside is a symmetry..and not just the 2 sphere sides...becoz we consider another sphere left and right of the sphres...and we have an inlet on the top and outlet on the bottom....here is the image you can see there...

 oj.bulmer March 30, 2013 20:29

I see, the snap you attached makes situation much clearer. That's what I suggested earlier, a picture is worth thousand words and helps avoid misunderstandings.

I understand that your problem is about meshing while this forum is about FLUENT. I suggest you post this thread onto Meshing forum:

http://www.cfd-online.com/Forums/ansys-meshing/

Try to be brief, to-the-point in your post at meshing forum and post the last image. There are lot of meshing experts in that forum and you will surely get a lot of ideas over your problem. But I doubt if someone will do your assignment for you. These forums are there for guidance but finally it is YOUR assignment!

Judging by the new snap, the assumption of symmetry you made initially seems unreasonable. Assuming your flow is in Z direction, You can at most divide your geometry in 4 parts, two parts each in X and Y direction. You will need to keep all spheres in Z direction. You say 6 spheres, but you show layers of 5 and 4 spheres after another so there is a discrepancy in your description. Also, if you want to use a fully hex mesh, you will need a multiblock approach in ICEM CFD, and from the looks of your geometry, it is not going to be piece of cake. Before you attempt that, you should have done basic ICEM tutorials. If you have no idea about multiblock mesh, am not sure how you are going to create mesh independent CFD solution.

It is possible to scale the mesh in ICEM , have a look at documentation.

OJ

 elina March 30, 2013 20:42

Hi
This image was just a sample of the arrangement ..out of these I have to take just one sphere in each y axis so that its like one sphere kept above the other...only thing is slight variation in the x coordinates...hope you got it..and then I have to take the symmetry of that complete sphere arrangement in y direction in such a way that I have only one quarter of the geometry so that it makes the meshing and simulation time more easy and less time....And regarding the mesh interdependent study is like I need to change the size of elements and see different outlet temperatures and have to plot a graph for outlet temperature vs the number of elements in the geometry then we can see after some number of elements the temperature remains the same...I hope that I can do just by giving the parameter for number of elements...and element size if i am correct..And now the problem is just that I need to reduce the elements in meshing and in such a way that I can mesh the sphere like small concentric circles...or can be said as Hexahedra mesh.....This is what I have to know to do....

 oj.bulmer March 31, 2013 10:35

Good to see you have posted the thread on Meshing forum. Am sure you will have some suggestions there.

OJ

 elina March 31, 2013 10:44

Ah thanks sir
But still dint get any advice or helps.....:(

 elina March 31, 2013 15:44

Hallo Oj Blumer..

Your Idea seems to be interesting...see this entire spheres are not necessary ot mesh.. i will take only one sphere in each layer and will take the symmetry of that...so the spheres will be arranged vertically...initially what the geometry you did was for 2 spheres..like that I have to keep 4 more spheres on the top or bottom..its ok wherever it is... and the flow is from the top that is positive y axis and not z axis..nothing to do with z axis....and I have to calculate the outlet temperate based on the temperature I give at inlet as 363 K ,..so based on the mesh element sizes I can see different temperature at the outlet...so I can come to a conclusion that from 2 layer of sphere the mesh sizes and these mesh sizes can be used for my 6 layer of spheres...hope you got it..and since the number of spheres is more there will less temperature at outlet which is the required aim....and will give the inlet flow at different velocities..hope I made you clear...and is your mesh which you send me is a good quality one...??

 All times are GMT -4. The time now is 15:10.