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 The information provided may be out of date.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.

different channel thickness

Giovanni Battista Accetta

Please login with a confirmed email address before reporting spam

Hi everybody,
In changing from version 3.5a to version 4.0a
I encountered a small problem:
Using the Navier Stokes model in the old version I could setup two different channel thickness for two different subdomain. In Version 4 this option is in the main tab of the physic interface and what I choose there will obviously apply to all the elements in that geometry. How can I solve the problem? Should I use two different physic interfaces?
thanks everybody,
Giovanni ACCETTA

8 Replies Last Post 2011年3月30日 GMT-4 10:15
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9月4日 GMT-4 03:10
Hi
you add another tab of the same physiscs(at same node level) and you define the new channel , then you will notice that the domain from the previous will have a remarque like (superseeded). You should note the small green/red triangle arrows on top of the icon, these are linked with the way certain BC settings take preceedence onthe next. If you add 3 times the same BC one after the other they are analysed in that order, so if you include the same domain in all three it's only the one of the last (lowest) line that is used (except if you "disable it") then you have another BC settings, as the 2nd will put its values on the (shaded) ones

--
Good luck
Ivar
Hi you add another tab of the same physiscs(at same node level) and you define the new channel , then you will notice that the domain from the previous will have a remarque like (superseeded). You should note the small green/red triangle arrows on top of the icon, these are linked with the way certain BC settings take preceedence onthe next. If you add 3 times the same BC one after the other they are analysed in that order, so if you include the same domain in all three it's only the one of the last (lowest) line that is used (except if you "disable it") then you have another BC settings, as the 2nd will put its values on the (shaded) ones -- Good luck Ivar

Giovanni Battista Accetta

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2010年9月4日 GMT-4 11:38
Thank you for your answer Ivar.
The problem is a little more complicate, because the option I want to modify is not a boundary condition, nor a subdomain condition but a setting of the physics tab itself for all the domains included in that physics. For instance you can't select it with a "rose" tab. It seems to me strange that in previously versions it was a subdomain setting and now not anymore.
thank you again,
Giovanni ACCETTA
Thank you for your answer Ivar. The problem is a little more complicate, because the option I want to modify is not a boundary condition, nor a subdomain condition but a setting of the physics tab itself for all the domains included in that physics. For instance you can't select it with a "rose" tab. It seems to me strange that in previously versions it was a subdomain setting and now not anymore. thank you again, Giovanni ACCETTA

Giovanni Battista Accetta

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2010年9月4日 GMT-4 12:33
I found one manual solution that could work.
The shallow channel approximation in which I am interested pratically just add a volume force member to the equation. So I can simulate the different thickness by manually adding these volume forces as subdomain condition respecting the same equation which is used and given in the help.
The results seem to be pratically the same using both approaches, so I think this solution could work properly.
I found one manual solution that could work. The shallow channel approximation in which I am interested pratically just add a volume force member to the equation. So I can simulate the different thickness by manually adding these volume forces as subdomain condition respecting the same equation which is used and given in the help. The results seem to be pratically the same using both approaches, so I think this solution could work properly.

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9月5日 GMT-4 12:31
Hi

I was also proposing something not possible in V4, in fluidics CFD the thickness variable in V4 does not seem to be defined, which for me means it's by default 1[m] thick.

I.e. volume forces such as gravity are still defined as N/m^3 so you do not need then to use the thickness to get your "g_cnst*rho"

It is in 2D ST (structural) that you have explicitely the thickness variable d.

For the fluidic there is an exception: the thin film laminar flow has a initial film thickness variable,

Obviously I need to read the new 4.0a doc again :)

Do not forget that in V4 there are many physics nodes that accept additional sub level addons, so always try a right click on each node, to see what is awailable

Hope I'm closer to your initial question now ;)
--
Good luck
Ivar
Hi I was also proposing something not possible in V4, in fluidics CFD the thickness variable in V4 does not seem to be defined, which for me means it's by default 1[m] thick. I.e. volume forces such as gravity are still defined as N/m^3 so you do not need then to use the thickness to get your "g_cnst*rho" It is in 2D ST (structural) that you have explicitely the thickness variable d. For the fluidic there is an exception: the thin film laminar flow has a initial film thickness variable, Obviously I need to read the new 4.0a doc again :) Do not forget that in V4 there are many physics nodes that accept additional sub level addons, so always try a right click on each node, to see what is awailable Hope I'm closer to your initial question now ;) -- Good luck Ivar

Giovanni Battista Accetta

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2010年9月5日 GMT-4 14:32
Hi Ivar,
Yes that's correct what you are saying. Even if much less manual sometimes version 4.0 still disappoints me for some lack with respect to the previous one. But I think it's normal, I think it still needs to develop. I only hope an improvement in the documentation.

Anyway the thickness variable, in the laminar flow physics (2d geometry), is taken into account only when you add the shallow channel approximation. In this case the thickness field appears to be specified. Anyway according to the doc, this thickness will just add a term F(volume force) to the Navier Stokes equation of the form:
12*velocity*dynamic viscosity/thickness. This should consider the effect of the close walls of the channel making the hypothesis that the section is rectangular.

I tried both simulation with the shallow channel approximation and without by manually adding a volume force equally the formula I just wrote obtaining the same results.

I didn't know that sometimes you could sub level add-ons. Good to know.

Thank you again,
Giovanni
Hi Ivar, Yes that's correct what you are saying. Even if much less manual sometimes version 4.0 still disappoints me for some lack with respect to the previous one. But I think it's normal, I think it still needs to develop. I only hope an improvement in the documentation. Anyway the thickness variable, in the laminar flow physics (2d geometry), is taken into account only when you add the shallow channel approximation. In this case the thickness field appears to be specified. Anyway according to the doc, this thickness will just add a term F(volume force) to the Navier Stokes equation of the form: 12*velocity*dynamic viscosity/thickness. This should consider the effect of the close walls of the channel making the hypothesis that the section is rectangular. I tried both simulation with the shallow channel approximation and without by manually adding a volume force equally the formula I just wrote obtaining the same results. I didn't know that sometimes you could sub level add-ons. Good to know. Thank you again, Giovanni

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年3月22日 GMT-4 13:26
Hi Ivar and Giovanni,

I'm new to COMSOL 4.0, i've been used to the "shallow channel approximation" option in subdomain menu in Comsol 3.4.
Currently, i'm trying to get the same results by including the "volume forces" in the 4.0 version. I have not worked with variables in Comsol and having difficulties figuring out how to include all the right components.

I have my model (mod1), my laminar flow physics (spf) and my material (water). The force component is F=-12*dynamic viscosity*velocity/height^2.

I'm putting in the following equation in the x: -12*mod1.eta(T)*mod1.spf.u/((3.5e-6)^2)
and y: -12*mod1.eta(T)*mod1.spf.v/((3.5e-6)^2)
my channel height is 3.5um.

Am i doing something wrong?

Thank you,
Ivan
Hi Ivar and Giovanni, I'm new to COMSOL 4.0, i've been used to the "shallow channel approximation" option in subdomain menu in Comsol 3.4. Currently, i'm trying to get the same results by including the "volume forces" in the 4.0 version. I have not worked with variables in Comsol and having difficulties figuring out how to include all the right components. I have my model (mod1), my laminar flow physics (spf) and my material (water). The force component is F=-12*dynamic viscosity*velocity/height^2. I'm putting in the following equation in the x: -12*mod1.eta(T)*mod1.spf.u/((3.5e-6)^2) and y: -12*mod1.eta(T)*mod1.spf.v/((3.5e-6)^2) my channel height is 3.5um. Am i doing something wrong? Thank you, Ivan

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 2011年3月23日 GMT-4 02:37
Hi

it's not 100% easy to reply, as we do not really have enough info on your model, but what I believe is:

you do not need to repeat "mod1.spf" for dependent variables just "v" should do.

The same for eta, in fact just "eta" (if this is the material property of a defined material) is enough as it implicitly includes the reference to the temperature "T" (that must be defined, at least in a Parameter, but normally in spf its already looked into). Then if you divide by a length add the units, in your case v/(3.5[um])^2 should do.

COMSOL v4 flags unit errors in orange, and typos in red, in most places (still not 100% there in v4.0)

--
Good luck
Ivar
Hi it's not 100% easy to reply, as we do not really have enough info on your model, but what I believe is: you do not need to repeat "mod1.spf" for dependent variables just "v" should do. The same for eta, in fact just "eta" (if this is the material property of a defined material) is enough as it implicitly includes the reference to the temperature "T" (that must be defined, at least in a Parameter, but normally in spf its already looked into). Then if you divide by a length add the units, in your case v/(3.5[um])^2 should do. COMSOL v4 flags unit errors in orange, and typos in red, in most places (still not 100% there in v4.0) -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年3月30日 GMT-4 10:15
Hi,

it is important to remember that a flow is only two dimensional if it is either axisymmetric or if the flow domain has an infinite extension in the third direction (i.e. an infinite thickness). That is the reason to why there is no "thickness" available in the general 2D case. In practice, the simulated device must be "thick enough" so that the influence of the side walls is negligible.

When the "Shallow channel approximation" is activated, it is assumed that the side walls are instead close enough for the flow to have a fully developed laminar flow profile in the third direction. As you have observed, this effect is accounted for by an additional drag term in the momentum equations. The thickness does however not enter the continuity equation, and a change in thickness would hence result in a mass loss or mass source. The thickness input in version 3.5a did not respect this restriction. That is the reason to why the thickness now is an interface parameter.

Niklas
Hi, it is important to remember that a flow is only two dimensional if it is either axisymmetric or if the flow domain has an infinite extension in the third direction (i.e. an infinite thickness). That is the reason to why there is no "thickness" available in the general 2D case. In practice, the simulated device must be "thick enough" so that the influence of the side walls is negligible. When the "Shallow channel approximation" is activated, it is assumed that the side walls are instead close enough for the flow to have a fully developed laminar flow profile in the third direction. As you have observed, this effect is accounted for by an additional drag term in the momentum equations. The thickness does however not enter the continuity equation, and a change in thickness would hence result in a mass loss or mass source. The thickness input in version 3.5a did not respect this restriction. That is the reason to why the thickness now is an interface parameter. Niklas

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.