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 The information provided may be out of date.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.
Thermocapillary recirculation
Posted 19 sept. 2014, 02:37 UTC−4 Fluid & Heat, Heat Transfer & Phase Change, Studies & Solvers Version 4.4 5 Replies
Please login with a confirmed email address before reporting spam
EDITED
Hello,
I am a beginning user of COMSOL.
Currently, I am trying to create a preliminary model of Marangoni convection, like this one available on the COMSOL site: www.comsol.com/model/marangoni-convection-110
The main difference is that mine is axisymmetric, rather than two-dimenional, and instead of a temperature difference, the fluid is heated with a point heat source on the upper, free surface.
I was planning on doing both a stationary and time dependent study, but both are running into errors.
EDIT: The problem with the strange temperature distribution has been solved. The heat conductivity was improperly defined. However, now it reports errors in user defined functions.
The error message is as follows:
Could someone help me get this right?
I've attached my model to this post.
With regards,
Aaron Ayal
Hello,
I am a beginning user of COMSOL.
Currently, I am trying to create a preliminary model of Marangoni convection, like this one available on the COMSOL site: www.comsol.com/model/marangoni-convection-110
The main difference is that mine is axisymmetric, rather than two-dimenional, and instead of a temperature difference, the fluid is heated with a point heat source on the upper, free surface.
I was planning on doing both a stationary and time dependent study, but both are running into errors.
EDIT: The problem with the strange temperature distribution has been solved. The heat conductivity was improperly defined. However, now it reports errors in user defined functions.
The error message is as follows:
Error in user-defined function.
- Function: rho
Failed to evaluate variable.
- Variable: comp1.ht.rho
- Defined as: ((unit_kg_cf*rho((comp1.T/unit_K_cf)))/(unit_m_cf^3))
Failed to evaluate expression.
- Expression: d((-2*comp1.ht.rho*comp1.ht.Cp*(comp1.ht.ur*comp1.Tr+comp1.ht.uz*comp1.Tz)*test(comp1.T)*pi*r)*(dvol),{test@15})
- Feature: Stationary Solver 1 (sol2/s1)
Could someone help me get this right?
I've attached my model to this post.
With regards,
Aaron Ayal
Attachments:
5 Replies Last Post 15 déc. 2014, 06:42 UTC−5
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hello Aaron,
perhaps you want to check out your conductivity, which is negative in your model, that explains why you get cooling instead of heating.
Moreover, in 2D axi, you need to multiply by 2*pi*r the weak contribution.
At last, 1e5 W is a large value. Perhaps you want to estimate your Marangoni number and your Reynolds number so that you have an estimate of the thickness of the boundary layers. Your mesh should be adjusted accordingly. You can guess as well the regime of the flow (laminar or turbulent).
I hope this helps, good luck,
Eric
perhaps you want to check out your conductivity, which is negative in your model, that explains why you get cooling instead of heating.
Moreover, in 2D axi, you need to multiply by 2*pi*r the weak contribution.
At last, 1e5 W is a large value. Perhaps you want to estimate your Marangoni number and your Reynolds number so that you have an estimate of the thickness of the boundary layers. Your mesh should be adjusted accordingly. You can guess as well the regime of the flow (laminar or turbulent).
I hope this helps, good luck,
Eric
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
I forgot to add that the Microfluidics Module contains the physics needed to ease the equations part in the "Laminar Two-Phase Flow, Moving Mesh Interface". You can use it without the Moving Mesh of course. The Marangoni condition is expressed by entering the temperature dependent expression of the surface tension. You can add a Heat Transfer equation to get the full coupling.
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Dear Eric,
Thank you for your reply.
I just found out abut the error in the conduction.
Also, thank you for solving the weak contribution problem.
With regards,
Aaron
Thank you for your reply.
I just found out abut the error in the conduction.
Also, thank you for solving the weak contribution problem.
With regards,
Aaron
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi Eric,
I am bit confused by your remark regarding negative conductivity. I plotted the value of "k" as defined in the program for the specified temperature range. I did not get any negative value.
Regards,
Susant
I am bit confused by your remark regarding negative conductivity. I plotted the value of "k" as defined in the program for the specified temperature range. I did not get any negative value.
Regards,
Susant
Please login with a confirmed email address before reporting spam
Posted:
10 years ago
The file has been edited and corrected since the mention of the wrong conductivity.
Here the heating is pretty strong : the Marangoni number should be evaluated that will probably show that the regime gets turbulent, or at least time-dependent.
It's a good idea to use the time-dep. solver then, because the flow is very likely to be unstationnary.
The mesh shown in the model is not consistent with the boundary layers size, that you can estimate with your Marangoni number.
So the error in your model is a consequence of the divergence of the non-linear solver. rho is actually correctly defined.
I recommend strongly ramping up the heat source in time (with a smoothed step function), and avoid using a point heat source in space, since the temperature will be mesh dependent at this point. If you use the point heat source, have a look to the option to specify a radius for the heat source, it spreads over a small location the heat source, without adding extra dofs and keeping the problem well-posed (it adds directly the right contributions to the matrix system).
There is a nice tutorial about point heat sources in the Heat Transfer Module : localized Heat Source that you can find here :
www.comsol.fr/model/localized-heat-source-17307
(this example is new in COMSOL V5, it makes use of the special operators diskavg() or the like).
For fluid flow problems, reading the documentation of the models in COMSOL and trying not to introduce discontinuities in time or in space (that lead to shocks that are in some circumstances hard to damp out) is probably a good advice. Estimating the size of the boundary layers with a quick non dimensional analysis should be done at some point. This is hard to automate. You can read a nice similar analysis in terms of natural convection in :
www.comsol.fr/model/buoyancy-flow-in-free-fluids-665
I hope this helps,
Eric
COMSOL France
Here the heating is pretty strong : the Marangoni number should be evaluated that will probably show that the regime gets turbulent, or at least time-dependent.
It's a good idea to use the time-dep. solver then, because the flow is very likely to be unstationnary.
The mesh shown in the model is not consistent with the boundary layers size, that you can estimate with your Marangoni number.
So the error in your model is a consequence of the divergence of the non-linear solver. rho is actually correctly defined.
I recommend strongly ramping up the heat source in time (with a smoothed step function), and avoid using a point heat source in space, since the temperature will be mesh dependent at this point. If you use the point heat source, have a look to the option to specify a radius for the heat source, it spreads over a small location the heat source, without adding extra dofs and keeping the problem well-posed (it adds directly the right contributions to the matrix system).
There is a nice tutorial about point heat sources in the Heat Transfer Module : localized Heat Source that you can find here :
www.comsol.fr/model/localized-heat-source-17307
(this example is new in COMSOL V5, it makes use of the special operators diskavg() or the like).
For fluid flow problems, reading the documentation of the models in COMSOL and trying not to introduce discontinuities in time or in space (that lead to shocks that are in some circumstances hard to damp out) is probably a good advice. Estimating the size of the boundary layers with a quick non dimensional analysis should be done at some point. This is hard to automate. You can read a nice similar analysis in terms of natural convection in :
www.comsol.fr/model/buoyancy-flow-in-free-fluids-665
I hope this helps,
Eric
COMSOL France