[sujay]

Dear All,

My geometry is similar to water tank with inlet at top and outlet at bottom (both at off center) along the central axis.

For Meshing i am using smaller size at inlet and outlet (Max size 4 mm) through part mesh setting and for rest region through Global mesh size (Max Size).

With fixed max mesh size for inlet and outlet (Max 4 mm) , I am setting global mesh size. The change in total elements and nodes are as below,

Global Mesh Size (Max) Total Element Nodes
64 mm 414383 72421
32 mm 461344 80527
16 mm 841709 145118

How total element/nodes depend on Global mesh size? Please guide to select mesh size for mesh in-dependency.

Thanks,

Sujay Patil

[energy382]

Originally Posted by sujay

Dear All,

My geometry is similar to water tank with inlet at top and outlet at bottom (both at off center) along the central axis.

For Meshing i am using smaller size at inlet and outlet (Max size 4 mm) through part mesh setting and for rest region through Global mesh size (Max Size).

With fixed max mesh size for inlet and outlet (Max 4 mm) , I am setting global mesh size. The change in total elements and nodes are as below,

Global Mesh Size (Max) Total Element Nodes
64 mm 414383 72421
32 mm 461344 80527
16 mm 841709 145118

How total element/nodes depend on Global mesh size? Please guide to select mesh size for mesh in-dependency.

Thanks,

Sujay Patil

global scale factor means grow ratio in x,y,z direction.

e.g, you have a mesh with 100.000 elements and want to achieve a 300.000 elements mesh. You need an total grow ratio of three, which simply means cube root of three and you have approx. your global scale factor (not absolutely exact, but very close).
you can also scale spacing, but be aware: it's only enlarged up to 33%

speaking for turbomachinery, I normally goes in my example for 100k, 300k and 900k

[energy382]

Originally Posted by energy382

global scale factor means grow ratio in x,y,z direction.

e.g, you have a mesh with 100.000 elements and want to achieve a 300.000 elements mesh. You need an total grow ratio of three, which simply means cube root of three and you have approx. your global scale factor (not absolutely exact, but very close).
you can also scale spacing, but be aware: it's only enlarged up to 33%

speaking for turbomachinery, I normally goes in my example for 100k, 300k and 900k

btw: i was speaking about hexa

[sujay]

Originally Posted by energy382

btw: i was speaking about hexa

Dear Chirstoph,

Thanks.

Meshing Details :

Shell Meshing: Patch dependent with Quad dominant Mesh
Volume Meshing : Tetra/mixed with Robust(Octree)

Can you please guide for tetra mesh.

Sujay Patil

[energy382]

Hey,

it's better to ask someone more experienced (with tetra), as I usually use hex mesh for my simulation, sorry.

[PSYMN]

I would guess that this has to do with "powers of 2" and the octree process.

Do a search and you will find lots of other posts with details about how this works.

It appears to be seeing your 64 and 32 mm as the same basic size. Then when you go down to 16, it subdivides again. The Octree algorithms uses logic to decide if it should subdivide down to the next level or not. The solution is in your setup.

If you can post your tetin file with the 32 bit setting, I could take a quick look.

[sujay]

[sujay]

Dear Sir,

Please find the tin file.

Inlet, outlet, pipe submerged and pipe top assigned with smaller size while rest with 32 mm.

During the meshing i am not at all used the shell meshing, directly volume meshed.

Please clarify on the mesh quality also. Regarding the single edges, multiple edges and delauncy violation for the geometry.

Thanks,

Sujay Patil

[sujay]

Originally Posted by PSYMN

I would guess that this has to do with "powers of 2" and the octree process.

Do a search and you will find lots of other posts with details about how this works.

It appears to be seeing your 64 and 32 mm as the same basic size. Then when you go down to 16, it subdivides again. The Octree algorithms uses logic to decide if it should subdivide down to the next level or not. The solution is in your setup.

If you can post your tetin file with the 32 bit setting, I could take a quick look.

tin file added

[ashish agrahari]

hello frnds
I am working on cfd simulation of in-cylinder flow in SI engine, for which i am using dynamic mesh(layering and smoothing method) for movement of piston and valves. I have done full tetra meshing which is giving negative volume during compression. therefore it is required to have hybrid meshing i.e. tetra along with hex. Hex is required mainly at regions which is deforming eg engine cylinder. But, i am not able to mesh both hex and tetra on same geometry for some region. So can anyone help me in this case? how to mesh tetra and hex both in one geometry?

[Far]

Played with global scale factor (kept every thing else same, surface mesh from octree and volume mesh from denauly)

a) 1 >>>> fine mesh (500,000 nodes)
b) 2 >>>> Coarse mesh (250,000 nodes)


Coarse mesh:





Fine mesh:

[PSYMN]

Hey Far, Did you figure out what he did wrong, or did you just leave the size alone and play with the scale factor?

[Far]

I just played with scale factor.

I did steps in following manner:

1. Surface sizing through part mesh setup.

2. Density box at few locations

3. Gave tetra width and ratio

Once above parameters were set (though were random, may be he requires some different settings), then I played with scale factor : 1 and 2.

For scale factor 1, I had the fine mesh (double size) ~500,000
and for scale factor 2, mesh was coarse ~ 250,000

[sujay]

Originally Posted by Far

I just played with scale factor.

I did steps in following manner:

1. Surface sizing through part mesh setup.

2. Density box at few locations

3. Gave tetra width and ratio

Once above parameters were set (though were random, may be he requires some different settings), then I played with scale factor : 1 and 2.

For scale factor 1, I had the fine mesh (double size) ~500,000
and for scale factor 2, mesh was coarse ~ 250,000

Dear Sir,

Thanks.

can you show these steps with snaps in ICEM

1. Surface sizing through part mesh setup.

2. Density box at few locations

3. Gave tetra width and ratio

Sujay Patil

[Alia Amer]

Hi Everyone,
i am using ansys flotran for studying airflow inside buildings... i want to ask a question about generating the mesh ...after meshing the whole model....( i dont mesh the entire volume at once, i mesh it volume by volume ) .... how can i know the number of elements used in the mesh ????? i want to display it in the same way that you did:
Global Mesh Size (Max) Total Element Nodes
64 mm 414383 72421
32 mm 461344 80527
16 mm 841709 145118

is the element size measured in milimeters???

and i want to ask another question... is the scaling factor in ansys the same like grid refinement ratio???? and how should i use it????

[Far]

Originally Posted by Alia Amer

Hi Everyone,
i am using ansys flotran for studying airflow inside buildings... i want to ask a question about generating the mesh ...after meshing the whole model....( i dont mesh the entire volume at once, i mesh it volume by volume ) .... how can i know the number of elements used in the mesh ????? i want to display it in the same way that you did:
Global Mesh Size (Max) Total Element Nodes
64 mm 414383 72421
32 mm 461344 80527
16 mm 841709 145118

Element Size
is the element size measured in milimeters???

and i want to ask another question... is the scaling factor in ansys the same like grid refinement ratio???? and how should i use it????

Scaling Factor vs Grid Refinement:

Roughly speaking yes. but technically by reducing the scale factor, mesh sizes on the surfaces become half. But this may or may not be translated into double the mesh from previous one. As it largely depends on volume mesh and it is difficult to predict it.

Units of Element size
Element size in measured in the units you are using for your model.

No nodes vs element size
64 mm 414383 72421
32 mm 461344 80527
16 mm 841709 145118


As you can see , decreasing the element size is increasing the no of nodes, but it is not linear. And you can also see that same size (64 * 0.5 of scale factor = 32 and 64*0.25 of scale factor = 16) can also be obtained with scale factor. So roughly speaking chaning element size through scale factor or manually does change the mesh density, but you cannot be sure what is gonna happen precisely. Hit and trail is the way to go

[Alia Amer]

thank you so much Far for your informative reply
but what i want to know about the element number is how can i know it (i mean the way by which i can display it or the command that shows me the ni of elements used ) i use ansys flotran..can you help me to know from which command or main menu can i know the no of elements in my mesh?
thank you

[Far]

Originally Posted by Alia Amer

thank you so much Far for your informative reply
but what i want to know about the element number is how can i know it (i mean the way by which i can display it or the command that shows me the ni of elements used ) i use ansys flotran..can you help me to know from which command or main menu can i know the no of elements in my mesh?
thank you

If you are talking about flotran, I do not know it.

[Ugee]

In fluid flow problem does the mesh size of fluid domain affect the Y + value.IF it does then how to select appropriate size as in large geometry changing fluid domain size by 1 mm added million of elements for tetra meshing.

[AtoHM]

The mesh spacing close to the wall does affect your y+ value, yes. Search the forum for threads about yplus values, there is a ton.
I dont understand your second sentence, proper punctuation might help.