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.

Solving non-dimensional NS equations!

Please login with a confirmed email address before reporting spam

Hi Everybody,
Last week I attended a seminar organised by COMSOL in Melbourne. I asked a question to the presenter but he could not answer the question! My question was very simple. How can we use unit = none in COMSOL 4.0. Somebody told that it is available in the COMSOL 3.4 and 3.5. Can anybody help me? I would like to test how COMSOL deals with solving non-dimensional NS equations.

Many Thanks
Subrata

5 Replies Last Post 2010年6月7日 GMT-4 12:37
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 2010年6月4日 GMT-4 02:50
Hi

There is no problem just ignoring the units in COMSOL (at least in 3.5a), the "red" flag are there as warnings, but they appear even when COMSOl is not correctly porting the units (as with coupling variables)

So it is just to go a head, there is no obligation to use them

On the other side, when applying physics the units are an essential tool to check your dimensional coherence, and for me is a must
I agree if you do plain math you can often forget them, but do expect some problems to communicate with the physicists.
I remeber a nice example from my studies: our theoretical physics colleague was preapring the theory for an experiment to test some of Einsteins theory, and we others were there to design the test set up.
For him to put C=1 speed of light simplified his writing, nice, until we others suddenly found out that wee needed an experimental set-up of a size of some million kilometers to test the theory, he gote a nice note for the theory, but we had trouble with the CAD design, it didnt really scale to stay within one screen :)

So for me UNITS remain essentially usefull (to stay down on earth)

Have fun Comsoling
Hi There is no problem just ignoring the units in COMSOL (at least in 3.5a), the "red" flag are there as warnings, but they appear even when COMSOl is not correctly porting the units (as with coupling variables) So it is just to go a head, there is no obligation to use them On the other side, when applying physics the units are an essential tool to check your dimensional coherence, and for me is a must I agree if you do plain math you can often forget them, but do expect some problems to communicate with the physicists. I remeber a nice example from my studies: our theoretical physics colleague was preapring the theory for an experiment to test some of Einsteins theory, and we others were there to design the test set up. For him to put C=1 speed of light simplified his writing, nice, until we others suddenly found out that wee needed an experimental set-up of a size of some million kilometers to test the theory, he gote a nice note for the theory, but we had trouble with the CAD design, it didnt really scale to stay within one screen :) So for me UNITS remain essentially usefull (to stay down on earth) Have fun Comsoling

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2010年6月5日 GMT-4 00:50
Hi Ivar,
Thanks for your response. People who do reseach on fundamental CFD problem usually like to show non-dimensional solutions based on non-dimensional parameters (Rayleigh number, Prandtl number, aspect ratio....). They also normalise the NS equation with a set non-dimensiona variable (e.g. u=u*/U_0, t=t*/(h^2/eta) ....). Therefore, the NS equations take the non-dimension form the with Ra, Re, Pr ... instead of rho, eta, alpha.... Then if we choose in COMSOL rho=1, eta = Pr/Ra^1/2...... then we can get the non-dimensional solutions. But in COMSOL 3.4 somebody told that we can use unit = none in Physics->modelling settings->Base unit system. But I could not find this in COMSOL 4.0.

I love COMSOL as it shows all equations and we can alter everything we wish. It is not like other blackbox package. If I get this information then I will be able to convivce my BOSS to get this software in our lab.

Cheers
Subrata
Hi Ivar, Thanks for your response. People who do reseach on fundamental CFD problem usually like to show non-dimensional solutions based on non-dimensional parameters (Rayleigh number, Prandtl number, aspect ratio....). They also normalise the NS equation with a set non-dimensiona variable (e.g. u=u*/U_0, t=t*/(h^2/eta) ....). Therefore, the NS equations take the non-dimension form the with Ra, Re, Pr ... instead of rho, eta, alpha.... Then if we choose in COMSOL rho=1, eta = Pr/Ra^1/2...... then we can get the non-dimensional solutions. But in COMSOL 3.4 somebody told that we can use unit = none in Physics->modelling settings->Base unit system. But I could not find this in COMSOL 4.0. I love COMSOL as it shows all equations and we can alter everything we wish. It is not like other blackbox package. If I get this information then I will be able to convivce my BOSS to get this software in our lab. Cheers Subrata

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 2010年6月5日 GMT-4 05:36
Hi

Well I suppose we feel the same way w.r.t. Comsol. finally a programme that allows us rapidly to study the physics, without having to spend 80% of the time doing maths. It's not that I do not like the maths, it's just that my Boss/clients wants to see results for the physics we are studying, not pages of equations.

Still, I do not see what holds you back. I have not heard of the the units "none", but its a long time since 3.4 (only a year and a half but still ....) However, so what if there is no "none"? I do not see the reason why you cannot just continue to use te programme. As I said before its only a question of normalisation, unitless is one way, you can get the basics COMSOL and write the rest "unitless", or you get the additional toolboxes, by default normalised to SI units, but then up to you to change all formulas to get them unitless. I beleive you can wrie out your constants/parameters ti use to transfor your different phyiscal values to your unitless system, and then just multiply i.e. a density by your "rho_unit", or you Prandt number, ort why notspeed of light c=1, etc, idem for the postprocessing, to convert back to your unitless system.

You can make this as a load-in file that you do once when you set-up your new models, no ?

For me units "red flags" are warning of potential errors, not necesarily errors, and does not stop Comsol to solve, as it has no way to tell if you are right or not, it just solves what you put in.

It's slightl more typing, but personally I use the units of COMSOL to check my equations, its so easy to do a typo, that an independent check is always good to have. Its the same withthe two other tools I like to use to check my results: Maple and MapleSim, bot can be systematic with the units, provided you type out a little more ;)

With the possibilites offered by COMSOL, for a price I beleive is very copetitive with respect to any other high level FEM programme, you can only be a winner. (PS it took me some years to convince my boss, and I have still problems with two major clients not accepting it as they are married to the competition, so just be patient, and defed your cas)

Have fun COMSOLING
Ivar
Hi Well I suppose we feel the same way w.r.t. Comsol. finally a programme that allows us rapidly to study the physics, without having to spend 80% of the time doing maths. It's not that I do not like the maths, it's just that my Boss/clients wants to see results for the physics we are studying, not pages of equations. Still, I do not see what holds you back. I have not heard of the the units "none", but its a long time since 3.4 (only a year and a half but still ....) However, so what if there is no "none"? I do not see the reason why you cannot just continue to use te programme. As I said before its only a question of normalisation, unitless is one way, you can get the basics COMSOL and write the rest "unitless", or you get the additional toolboxes, by default normalised to SI units, but then up to you to change all formulas to get them unitless. I beleive you can wrie out your constants/parameters ti use to transfor your different phyiscal values to your unitless system, and then just multiply i.e. a density by your "rho_unit", or you Prandt number, ort why notspeed of light c=1, etc, idem for the postprocessing, to convert back to your unitless system. You can make this as a load-in file that you do once when you set-up your new models, no ? For me units "red flags" are warning of potential errors, not necesarily errors, and does not stop Comsol to solve, as it has no way to tell if you are right or not, it just solves what you put in. It's slightl more typing, but personally I use the units of COMSOL to check my equations, its so easy to do a typo, that an independent check is always good to have. Its the same withthe two other tools I like to use to check my results: Maple and MapleSim, bot can be systematic with the units, provided you type out a little more ;) With the possibilites offered by COMSOL, for a price I beleive is very copetitive with respect to any other high level FEM programme, you can only be a winner. (PS it took me some years to convince my boss, and I have still problems with two major clients not accepting it as they are married to the competition, so just be patient, and defed your cas) Have fun COMSOLING Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2010年6月7日 GMT-4 12:27
You can simply cheat by associating "dimensional" parameters in the COMSOL implementation with the dimensionless coefficients in your dimensionless formulation. For example for standard NS equation, type in 1 for viscosity and type in the value of your Reynolds number for density etc. and ignore the unit warnings, if any. Of course understand that the results will also be really dimensionless although COMSOL will tell you it is in the units of whatever base system you implemented the cheat in (i.e. it may give you velocity field in m/s when in fact it is really dimensionless with whatever scale you used to non-dimensionalize your problem to begin with).
Hope this helps.
Ozgur
You can simply cheat by associating "dimensional" parameters in the COMSOL implementation with the dimensionless coefficients in your dimensionless formulation. For example for standard NS equation, type in 1 for viscosity and type in the value of your Reynolds number for density etc. and ignore the unit warnings, if any. Of course understand that the results will also be really dimensionless although COMSOL will tell you it is in the units of whatever base system you implemented the cheat in (i.e. it may give you velocity field in m/s when in fact it is really dimensionless with whatever scale you used to non-dimensionalize your problem to begin with). Hope this helps. Ozgur

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2010年6月7日 GMT-4 12:37
Thanks both of you for your suggestions. I will follow it. I don't know why they ignored this important item (Physics->modelling settings->Base unit system as it was in COMSOL 3.4) in this version. Then everything would be fine. Especially who deals with the non-dimensional equations. I know I will face problem when I will do the post processing (e.g. If I want to show the legend of a figure it will show values with unit !!).

Thanks again.

Cheers
Subrata
Thanks both of you for your suggestions. I will follow it. I don't know why they ignored this important item (Physics->modelling settings->Base unit system as it was in COMSOL 3.4) in this version. Then everything would be fine. Especially who deals with the non-dimensional equations. I know I will face problem when I will do the post processing (e.g. If I want to show the legend of a figure it will show values with unit !!). Thanks again. Cheers Subrata

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.