Note: This discussion is about an older version of the COMSOL Multiphysics® software. The information provided may be out of date.

Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.

Urgent : Boundary layer meshing problem

Please login with a confirmed email address before reporting spam

Hi,

I need a help to understand the boundary layer meshing in COMSOL.

I have an axisymmetric model of a simple cylinder of diameter radius 6 mm. For the penetration depth of 0.14 mm at the outside surface forces me to use a meshing of about 0.014 mm. Next mesh “circle” should be at about 1/5 under the surface, then 1/3, next ½, then at 1, 1.5, 2 and 3 times Pen_Depth.

Please let me know how this can be achieved in Boundary layer meshing.

--
Ishant Jain


9 Replies Last Post 26 ott 2012, 17:27 GMT-4
Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 23 ott 2012, 01:24 GMT-4
Hi

check the boundary mesh info in the doc. Normally it's a 2 step operation: 1 select a domain where the boundary will appear (inside) then select ALL required boundaries where to apply the mesh, these coul be a connected polyline

then use rather a low layer number with a stretching factor >1. you should not restrict your adjustment factor to the depth, but to 3-5 times your skin depth, as with a stretching factor > 1 your mesh steps will increase gently , hence produce a smooth density transition

--
Good luck
Ivar
Hi check the boundary mesh info in the doc. Normally it's a 2 step operation: 1 select a domain where the boundary will appear (inside) then select ALL required boundaries where to apply the mesh, these coul be a connected polyline then use rather a low layer number with a stretching factor >1. you should not restrict your adjustment factor to the depth, but to 3-5 times your skin depth, as with a stretching factor > 1 your mesh steps will increase gently , hence produce a smooth density transition -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 23 ott 2012, 02:40 GMT-4
Dear Ivar,

I presume that you must have checked the model I have shared.

Which doc you have referred ? Where is the 2 step in the 2 step approach ?

I will request you to please reply considering a new user as I am unable to get what you have written in your reply.

I sincerely request you to please reply as earliest.




Hi

check the boundary mesh info in the doc. Normally it's a 2 step operation: 1 select a domain where the boundary will appear (inside) then select ALL required boundaries where to apply the mesh, these coul be a connected polyline

then use rather a low layer number with a stretching factor >1. you should not restrict your adjustment factor to the depth, but to 3-5 times your skin depth, as with a stretching factor > 1 your mesh steps will increase gently , hence produce a smooth density transition

--
Good luck
Ivar





--
ishant.jain@tatasteel.com
Dear Ivar, I presume that you must have checked the model I have shared. Which doc you have referred ? Where is the 2 step in the 2 step approach ? I will request you to please reply considering a new user as I am unable to get what you have written in your reply. I sincerely request you to please reply as earliest. [QUOTE] Hi check the boundary mesh info in the doc. Normally it's a 2 step operation: 1 select a domain where the boundary will appear (inside) then select ALL required boundaries where to apply the mesh, these coul be a connected polyline then use rather a low layer number with a stretching factor >1. you should not restrict your adjustment factor to the depth, but to 3-5 times your skin depth, as with a stretching factor > 1 your mesh steps will increase gently , hence produce a smooth density transition -- Good luck Ivar [/QUOTE] -- ishant.jain@tatasteel.com

Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 23 ott 2012, 03:23 GMT-4
Hi

start to open your model, select the "Mesh, Boundary Layer - Boundary Layer Properties" and select the related
boundaries (probably 4, 6 ,8). then set 20 boundary layers (not 400) a stretch value of 1.5, and a use Automatic and thickness of 0.001, then change gently the variables util you get as desired.

You could also start from a defaut boundary mesh and select the domaind and the respective bundaries (but NOT the on axis in 2D-axis)

Another way is to use a mapped mesh and use a distribution on the edge selections seed points

--
Good luck
Ivar
Hi start to open your model, select the "Mesh, Boundary Layer - Boundary Layer Properties" and select the related boundaries (probably 4, 6 ,8). then set 20 boundary layers (not 400) a stretch value of 1.5, and a use Automatic and thickness of 0.001, then change gently the variables util you get as desired. You could also start from a defaut boundary mesh and select the domaind and the respective bundaries (but NOT the on axis in 2D-axis) Another way is to use a mapped mesh and use a distribution on the edge selections seed points -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 23 ott 2012, 07:46 GMT-4
Dear Ivar,

I have encountered with an error "Inverted element occurs" with every change I am making.

Please let me know the meaning and significance of
1. Boundary layer stretching factor
2. Thickness Adjustment factor


Also please let me know, if it is possible to do the successive boundary layer meshing with different thickness between layer(s) group i.e. I have mentioned earlier :

For the penetration depth of 0.14 mm at the outside surface forces me to use a meshing of 10 layers of about 0.014 mm. Next mesh “circle” should be at about 1/5 under the surface, then 1/3, next ½, then at 1, 1.5, 2 and 3 times Pen_Depth.


--
Ishant Jain
Dear Ivar, I have encountered with an error "Inverted element occurs" with every change I am making. Please let me know the meaning and significance of 1. Boundary layer stretching factor 2. Thickness Adjustment factor Also please let me know, if it is possible to do the successive boundary layer meshing with different thickness between layer(s) group i.e. I have mentioned earlier : For the penetration depth of 0.14 mm at the outside surface forces me to use a meshing of 10 layers of about 0.014 mm. Next mesh “circle” should be at about 1/5 under the surface, then 1/3, next ½, then at 1, 1.5, 2 and 3 times Pen_Depth. -- Ishant Jain

Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 23 ott 2012, 11:00 GMT-4
Hi
you can only make 1 layer per boundary side, if you want several to cover a domain, you need to make several domains one inside the next (like a Russian doll ;)

Normally, if you select the 3 sides of a rectangle with a default mesh of at lest 2 elements across the domain then the default boundary mesh domain should work, then you can start to change and adapt the different parameters

--
Good luck
Ivar
Hi you can only make 1 layer per boundary side, if you want several to cover a domain, you need to make several domains one inside the next (like a Russian doll ;) Normally, if you select the 3 sides of a rectangle with a default mesh of at lest 2 elements across the domain then the default boundary mesh domain should work, then you can start to change and adapt the different parameters -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 26 ott 2012, 08:14 GMT-4

Hi Ivar,

Could you please see and let me know how this mapped meshing would be corrected ?

Boundary layer meshing is working fine, as we had discussed but I have to run the model to check the same.
--
ishant.jain@tatasteel.com
Hi Ivar, Could you please see and let me know how this mapped meshing would be corrected ? Boundary layer meshing is working fine, as we had discussed but I have to run the model to check the same. -- ishant.jain@tatasteel.com


Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 26 ott 2012, 14:36 GMT-4
Hi

but why such a special mesh ? Do you have different materials per layer later on ? with very different alpha (heat diffusivity ?)

else why not a mapped mesh with a distribution of the "predefined type, a rather high number of elements and a larger number for the element ratio and why not a geometric distribution (you might nee to revers the direction, or have 2 Distribution nodes, one for the top edge, one for the bottom edge, if their direction is opposite

Before you should not use mapped mesh along a 2D axi axis, I believe ths is not a limitation today, if so, try to add a thin layer and mesh it with tri, as last item in the meshing node list.

Boundary leayers are rather optimised for flow, or one skin depth resolution. I use it also for high diffusivity gradients boundaries

--
Good luck
Ivar
Hi but why such a special mesh ? Do you have different materials per layer later on ? with very different alpha (heat diffusivity ?) else why not a mapped mesh with a distribution of the "predefined type, a rather high number of elements and a larger number for the element ratio and why not a geometric distribution (you might nee to revers the direction, or have 2 Distribution nodes, one for the top edge, one for the bottom edge, if their direction is opposite Before you should not use mapped mesh along a 2D axi axis, I believe ths is not a limitation today, if so, try to add a thin layer and mesh it with tri, as last item in the meshing node list. Boundary leayers are rather optimised for flow, or one skin depth resolution. I use it also for high diffusivity gradients boundaries -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 26 ott 2012, 14:53 GMT-4
Hi Ivar,

Indeed I am using this meshing for boundary layer diffusivity i.e. why such a special mesh. I have the same material but I have followed what you have mentioned as Russian Doll so this type of meshing. Please note this does to compute the Induced current developed for the skin depth and consecutive Heat developed. I will request you to please comment.


I also tried to to Mapped meshing but didn't find it suitable to do it as I am encountering with "Inverted Mesh" Error. Please help me to understand the concept for element ration and GP and how it can be best used.

Thanks
Ishant




Hi

but why such a special mesh ? Do you have different materials per layer later on ? with very different alpha (heat diffusivity ?)

else why not a mapped mesh with a distribution of the "predefined type, a rather high number of elements and a larger number for the element ratio and why not a geometric distribution (you might nee to revers the direction, or have 2 Distribution nodes, one for the top edge, one for the bottom edge, if their direction is opposite

Before you should not use mapped mesh along a 2D axi axis, I believe ths is not a limitation today, if so, try to add a thin layer and mesh it with tri, as last item in the meshing node list.

Boundary leayers are rather optimised for flow, or one skin depth resolution. I use it also for high diffusivity gradients boundaries

--
Good luck
Ivar





--
ishant.jain@tatasteel.com
Hi Ivar, Indeed I am using this meshing for boundary layer diffusivity i.e. why such a special mesh. I have the same material but I have followed what you have mentioned as Russian Doll so this type of meshing. Please note this does to compute the Induced current developed for the skin depth and consecutive Heat developed. I will request you to please comment. I also tried to to Mapped meshing but didn't find it suitable to do it as I am encountering with "Inverted Mesh" Error. Please help me to understand the concept for element ration and GP and how it can be best used. Thanks Ishant [QUOTE] Hi but why such a special mesh ? Do you have different materials per layer later on ? with very different alpha (heat diffusivity ?) else why not a mapped mesh with a distribution of the "predefined type, a rather high number of elements and a larger number for the element ratio and why not a geometric distribution (you might nee to revers the direction, or have 2 Distribution nodes, one for the top edge, one for the bottom edge, if their direction is opposite Before you should not use mapped mesh along a 2D axi axis, I believe ths is not a limitation today, if so, try to add a thin layer and mesh it with tri, as last item in the meshing node list. Boundary leayers are rather optimised for flow, or one skin depth resolution. I use it also for high diffusivity gradients boundaries -- Good luck Ivar [/QUOTE] -- ishant.jain@tatasteel.com

Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 26 ott 2012, 17:27 GMT-4
Hi

sorry for having induced you in error then, but there you are overdoing it.
Fine mesh OK, but you can also obtain interesting progressing mesh density with the "dimensions" tab and using non linear functions on the two short radial edges

even you have the adaptive mesh that improves the mesh per iteration too. Meshing is complex one need to make many to learn, and it's worth to take one of the comsol courses, if you have the opportunity

For youre case, if you start by adding a 2nd "size" node and select the internal rectangle, then set it to extra fine, and leave the rest as default normal mesh you can also act locally and it should give you a first result. Then solve it, and observe the result, then decide to perhaps increse still the density

I often start with a default mesh, then make it more dense, and finally only if really I see problems, I go to very extreme cases, such as succesive layers.

Another important point is to have a rather regular mesh shpae, but there are also many exceptions where highly elongated mesh gives quicker and as precise results

--
Good luck
Ivar
Hi sorry for having induced you in error then, but there you are overdoing it. Fine mesh OK, but you can also obtain interesting progressing mesh density with the "dimensions" tab and using non linear functions on the two short radial edges even you have the adaptive mesh that improves the mesh per iteration too. Meshing is complex one need to make many to learn, and it's worth to take one of the comsol courses, if you have the opportunity For youre case, if you start by adding a 2nd "size" node and select the internal rectangle, then set it to extra fine, and leave the rest as default normal mesh you can also act locally and it should give you a first result. Then solve it, and observe the result, then decide to perhaps increse still the density I often start with a default mesh, then make it more dense, and finally only if really I see problems, I go to very extreme cases, such as succesive layers. Another important point is to have a rather regular mesh shpae, but there are also many exceptions where highly elongated mesh gives quicker and as precise results -- Good luck Ivar

Note that while COMSOL employees may participate in the discussion forum, COMSOL® software users who are on-subscription should submit their questions via the Support Center for a more comprehensive response from the Technical Support team.