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.

Basic Model Feedback

Please login with a confirmed email address before reporting spam

Hello,

I am trying to learn COMSOL to model a frisbee. My previous post may have been to broad to receive feedback so I am listing some specific details and uploading the model here.

Details:
Laminar Flow - initial value of 8.9 m/s y direction
Material - air
Wall- No Slip
Mesh - Fine

Should I be including Volume Force?

My intent is that this be a basic model of a frisbee flying through the air, but there are a lot of details to check.

2d plot:
-spatial frame
-arrow surface
- Time .5 (out of one second run)

I have tried to make a 2d plot of a YZ cut plane to show the Y velocity of air over the disc, but strange arrows are appearing coming out of the disc going vertically.

I may not have created a disc moving through air, but a disc made of air I am not sure how to interpret the model.

Any feedback will be really REALLY appreciated\. If you live near Buffalo I will come shovel your driveway.

Thanks,

Alex


4 Replies Last Post 11 gen 2011, 16:54 GMT-5
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 8 gen 2011, 17:21 GMT-5
Hi

indeed it looks like a fresbee of "air" floating in vacuum, I believe you should "invert" your volum to start with. Box it in a cube 3-5 times greater than the fresbee, and then subtract the fresbee volume from the box to only have the air outside the fresbee volume (and not inside) in 1 doman.

Then I suppose your fresbee is rotating with a fixed axis along Z passing by its CoG that I see is around (0,0,soemthing_in_z), and the fresbee is assumed rigid w.r.t the air, OK ?

So you must simulate the rotation by rotating velocity of the no-slip wall. Unfortunately I do not see any easy way to use a cylindrical coordiante in NSF, wounder why ? but you can define some variables such as rxy = sqrt(x^2+y^2) and write out the transformation vx=-y/rxy, vy=x/rxy (pls check I might have out it up wrong, or missed a sign).

Note do not forget to set some initial conditions compliant with the rest (including some lateral air flow, start with laminar)

I believe there are some airfin example somwhere, perhaps it was on the forum or the model exchange, that can give you also some clues

Also working directly in 3D wil ldemand for quite som RAM, you might be better off in 2D, but I'm not sure how symmetric one can define the rest of the BC's, not sure what to propose for that

Hope this helps on the way, have fun Comsoling

--
Good luck
Ivar
Hi indeed it looks like a fresbee of "air" floating in vacuum, I believe you should "invert" your volum to start with. Box it in a cube 3-5 times greater than the fresbee, and then subtract the fresbee volume from the box to only have the air outside the fresbee volume (and not inside) in 1 doman. Then I suppose your fresbee is rotating with a fixed axis along Z passing by its CoG that I see is around (0,0,soemthing_in_z), and the fresbee is assumed rigid w.r.t the air, OK ? So you must simulate the rotation by rotating velocity of the no-slip wall. Unfortunately I do not see any easy way to use a cylindrical coordiante in NSF, wounder why ? but you can define some variables such as rxy = sqrt(x^2+y^2) and write out the transformation vx=-y/rxy, vy=x/rxy (pls check I might have out it up wrong, or missed a sign). Note do not forget to set some initial conditions compliant with the rest (including some lateral air flow, start with laminar) I believe there are some airfin example somwhere, perhaps it was on the forum or the model exchange, that can give you also some clues Also working directly in 3D wil ldemand for quite som RAM, you might be better off in 2D, but I'm not sure how symmetric one can define the rest of the BC's, not sure what to propose for that Hope this helps on the way, have fun Comsoling -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 11 gen 2011, 12:29 GMT-5
Ivar thank you so much! Also thanks in advance to anyone else who has the time to advise.

I now have a disc made of silicon and a box made of air.

I am assuming away rotation for the time being, but thanks for the information I will bring it in later.

I have the volumes of the two, but cannot find how to subtract the volume of the disc from the box.
do you mean the volume listed in the measurements window for the box should actually be [Box volume-disc volume]?

Within Laminar flow I select only the Box domain correct?
Under laminar flow > Wall 1 I cannot choose which walls are no-slip and they are all selected.
Initial velocity set, should I add inlet and outlet velocities equal in magnitude and direction to the initial?

Finally in regards to working in 2d - do i just need to make a new model and select 2d at the beginning or can i switch somehow? You are right the processing takes forever as it is.

If I work in 2d will it only examine the flow over the center profile of the disc ( the outline that I rotated into 3d)? Now the model wont even solve due to lack or memory so I will need to switch or figure out the error.

Is there any reason for me to use time dependent solving? It seems everything should remain constant, and it might be easier to compute without it, however it cannot find a solution in this way either.

I have uploaded the updated model. Thanks again for your help i apologize for how rudimentary my questions must seem.
Ivar thank you so much! Also thanks in advance to anyone else who has the time to advise. I now have a disc made of silicon and a box made of air. I am assuming away rotation for the time being, but thanks for the information I will bring it in later. I have the volumes of the two, but cannot find how to subtract the volume of the disc from the box. do you mean the volume listed in the measurements window for the box should actually be [Box volume-disc volume]? Within Laminar flow I select only the Box domain correct? Under laminar flow > Wall 1 I cannot choose which walls are no-slip and they are all selected. Initial velocity set, should I add inlet and outlet velocities equal in magnitude and direction to the initial? Finally in regards to working in 2d - do i just need to make a new model and select 2d at the beginning or can i switch somehow? You are right the processing takes forever as it is. If I work in 2d will it only examine the flow over the center profile of the disc ( the outline that I rotated into 3d)? Now the model wont even solve due to lack or memory so I will need to switch or figure out the error. Is there any reason for me to use time dependent solving? It seems everything should remain constant, and it might be easier to compute without it, however it cannot find a solution in this way either. I have uploaded the updated model. Thanks again for your help i apologize for how rudimentary my questions must seem.


Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 11 gen 2011, 13:21 GMT-5
I have created a 2d model, which is working better, although it still wont solve due to not enough netwon iterations.

Ivan i found one of your old posts on how to increase the iterations, but am at 2000 and it has not helped.
I have created a 2d model, which is working better, although it still wont solve due to not enough netwon iterations. Ivan i found one of your old posts on how to increase the iterations, but am at 2000 and it has not helped.


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 11 gen 2011, 16:54 GMT-5
Hi

first I suppose its not crystalline Si (silicon) but perhaps "silicone" you wanted for the fresbee, no ?, anyhow you do not need to mesh it as you are first of all interested in the fluid flow and you can start with assuming the frisbee is solid (not flexing). So for that subtract the fresbee volume frm the block

Then in your case you are not interested in the flow profile along the walls of your box as I assume these are "infinite", so you could use "slip" conditions on the large box (but not on the fresbee wall), and for the outlet use simply "0" pressure , no stress

I notice you have not patched your 4.1 to the latest update, pls check on the main COSMOL update site, there are still a few error in 4.1.0.88

for me you can leave COMSOL set up the mesh it will propose a finer mesh that converges better

sometimes adding a p=0 point constraint on one of the outlet vall points help to fix the gauge pressure

at least like that it convrged for me in some 39 steps and using somewhat less than 3Gb RAM, I had to relaunch the solver (continuation) after the first 25 iteation as it did not converge with default settings

furhermore, to make a thighter mesh around and underneath the fresbee, extending somewhat behind it , you could box it with a box just slightly larger than the fresbee, and mesh this small box finer (comsol will do it also partially automatically), I believe your mesh is far too coarse

--
Good luck
Ivar
Hi first I suppose its not crystalline Si (silicon) but perhaps "silicone" you wanted for the fresbee, no ?, anyhow you do not need to mesh it as you are first of all interested in the fluid flow and you can start with assuming the frisbee is solid (not flexing). So for that subtract the fresbee volume frm the block Then in your case you are not interested in the flow profile along the walls of your box as I assume these are "infinite", so you could use "slip" conditions on the large box (but not on the fresbee wall), and for the outlet use simply "0" pressure , no stress I notice you have not patched your 4.1 to the latest update, pls check on the main COSMOL update site, there are still a few error in 4.1.0.88 for me you can leave COMSOL set up the mesh it will propose a finer mesh that converges better sometimes adding a p=0 point constraint on one of the outlet vall points help to fix the gauge pressure at least like that it convrged for me in some 39 steps and using somewhat less than 3Gb RAM, I had to relaunch the solver (continuation) after the first 25 iteation as it did not converge with default settings furhermore, to make a thighter mesh around and underneath the fresbee, extending somewhat behind it , you could box it with a box just slightly larger than the fresbee, and mesh this small box finer (comsol will do it also partially automatically), I believe your mesh is far too coarse -- 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.