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.
Beginner needs a few insight on heat transfert
Posted Jun 4, 2010, 11:09 a.m. EDT 5 Replies
Please login with a confirmed email address before reporting spam
Hi,
I m atm on an internship and I was asked to model some heat transferts using COMSOL on 3D with several domains.
I have already used COMSOL but for 2D problems on a single domain.
So I have a few questions to be sure my model is correct. I tried my best with the dynamic help, but I did not really got much out of it since my problem is not how to use but what to use.
First I would like to be sure my domains are correctly coupled. I am afraid the default setting for heat transfer is total insulation on the boundaries. Althought my domains in geometric contact need to transfer heat between each others. I tried to use the tool : "model couplings > boundary similarities" in definition, but it doesn 't seem to work for boundaries already in contact, does it?
So do I have to do smthg during the modeling? Or do I have to fuse boundary meshs before resolution?
Secondly I would like to model a heater which would activate when the Temperature in one domain reach a certain value and stop when a second value is reached. I have seen I can use probes in the definition tool, but I really have no clue as to how to link it to a dependant periodic heat source.
Also I have some hesitations as to which option to use in physics to model leaks. There are a lot of options that seems to be eligible but I want to be sure my leak has impact on the heat into my domain.
That is what I have on the top of my head for now.
Thank you for your help.
I am learning, so if you have some advices, tutorials or this kind of things, I am interested.
Have a good day.
I m atm on an internship and I was asked to model some heat transferts using COMSOL on 3D with several domains.
I have already used COMSOL but for 2D problems on a single domain.
So I have a few questions to be sure my model is correct. I tried my best with the dynamic help, but I did not really got much out of it since my problem is not how to use but what to use.
First I would like to be sure my domains are correctly coupled. I am afraid the default setting for heat transfer is total insulation on the boundaries. Althought my domains in geometric contact need to transfer heat between each others. I tried to use the tool : "model couplings > boundary similarities" in definition, but it doesn 't seem to work for boundaries already in contact, does it?
So do I have to do smthg during the modeling? Or do I have to fuse boundary meshs before resolution?
Secondly I would like to model a heater which would activate when the Temperature in one domain reach a certain value and stop when a second value is reached. I have seen I can use probes in the definition tool, but I really have no clue as to how to link it to a dependant periodic heat source.
Also I have some hesitations as to which option to use in physics to model leaks. There are a lot of options that seems to be eligible but I want to be sure my leak has impact on the heat into my domain.
That is what I have on the top of my head for now.
Thank you for your help.
I am learning, so if you have some advices, tutorials or this kind of things, I am interested.
Have a good day.
5 Replies Last Post Jun 22, 2010, 4:20 p.m. EDT
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
I assume you are in V4 and "ht" solids.
a) be sure you are using the "Model tree - Geometry - Finish: form union (and not assembly mode)
then normally you should have "continuity", by default, on all interiour boundaries, and you have isolation, by default, on all external (its a closed system by default).
if you use "assembly the interiour boundaries are by default "isolated", but in normal "form union" mode its continuity
Then if you are in transient mode you can add a function depending on time in the BC value, even with a bolean or an if() based on the average temperature of another edge (define it with a "Models - Definitions - Model Couplings - Average operator")
Then normally for me, my tests work OK as expected, and as in 3.5
Hope its also so for you
Have fun Comsoling
ivar
I assume you are in V4 and "ht" solids.
a) be sure you are using the "Model tree - Geometry - Finish: form union (and not assembly mode)
then normally you should have "continuity", by default, on all interiour boundaries, and you have isolation, by default, on all external (its a closed system by default).
if you use "assembly the interiour boundaries are by default "isolated", but in normal "form union" mode its continuity
Then if you are in transient mode you can add a function depending on time in the BC value, even with a bolean or an if() based on the average temperature of another edge (define it with a "Models - Definitions - Model Couplings - Average operator")
Then normally for me, my tests work OK as expected, and as in 3.5
Hope its also so for you
Have fun Comsoling
ivar
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi Ivar,
sorry I am answering so late, I entirely thought over my model, so it took me some time.
Thank you for your answer you assumed right, except I am mixing ht solid and fluid.
I really improved since the last time I came, my models are working now.
Either you are wrong, or I missunderstood you, but I had to select some "heat flux" on boundaries between domains and both ways to model the continuity of the heat.
I am still wondering about the sub-option "heat transfert in solidx/fluidx" inside the physics. I can't really understand how it is used in the solving and modelling. On some physics I may change it, on some I cannot. So for those I can, I select "User defined" and input the initiale value of the temperature but there is already an "Initial Values" sub-option and the help won't, so it keeps me wondering. Anyway it seems to give correct solution the way I am using it, but if you have some details about it please share.
Also I tried to build an animation of the solving over time, but it always takes so much time and memory on my laptop that I can't see throught. I have a lot of frames (18000 some) that might be the problem.
I ll keep looking into it, share my solutions later if I find anything.
Have a good day.
sorry I am answering so late, I entirely thought over my model, so it took me some time.
Thank you for your answer you assumed right, except I am mixing ht solid and fluid.
I really improved since the last time I came, my models are working now.
Either you are wrong, or I missunderstood you, but I had to select some "heat flux" on boundaries between domains and both ways to model the continuity of the heat.
I am still wondering about the sub-option "heat transfert in solidx/fluidx" inside the physics. I can't really understand how it is used in the solving and modelling. On some physics I may change it, on some I cannot. So for those I can, I select "User defined" and input the initiale value of the temperature but there is already an "Initial Values" sub-option and the help won't, so it keeps me wondering. Anyway it seems to give correct solution the way I am using it, but if you have some details about it please share.
Also I tried to build an animation of the solving over time, but it always takes so much time and memory on my laptop that I can't see throught. I have a lot of frames (18000 some) that might be the problem.
I ll keep looking into it, share my solutions later if I find anything.
Have a good day.
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
I havent looked into exactly what comsol is changing, but you can check the equations by turning them on in the options.
Fluis and solids have similar behaviour (mostly today the physics are presented very differently for historical reasons) but you can do certain simplifcations in the equations (i.e. you have Poisson values for solids, mostly not for fluids etc) so I assume COMSOL is changing some of these parameters, as well as internal solver parameter certainly to better mach the equations.
Keep it simple, do you really see that great difference with 18000 steps or just 18 ? if it's linear the effect is negligible, no, but not in time consumed ;)
Have fun Comsoling
Ivar
I havent looked into exactly what comsol is changing, but you can check the equations by turning them on in the options.
Fluis and solids have similar behaviour (mostly today the physics are presented very differently for historical reasons) but you can do certain simplifcations in the equations (i.e. you have Poisson values for solids, mostly not for fluids etc) so I assume COMSOL is changing some of these parameters, as well as internal solver parameter certainly to better mach the equations.
Keep it simple, do you really see that great difference with 18000 steps or just 18 ? if it's linear the effect is negligible, no, but not in time consumed ;)
Have fun Comsoling
Ivar
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi Ivar,
again thank you for your advices.
I did look into the equations more precisely which helped a lot. Actualy the fluid model only add the velocity field variable in the equation.
So my model is working pretty good, and I coupled it with some laminar flow. But now I am working in stationary (since I figured using time was useless for my model) and I can't find how to plot and evaluate the evolution of the solution which is critical to understand it.
Also would you know if the post-processing allows to have a separated color table for one of the subdomains.
Moreover I tried to solve with time, in order to compare the results but the solving stay stuck like it entered an infinite loop. The same operations follows one another and the progress bar do not move, event after 3 hours. I know solving can sometimes take even longer but that already seems like a huge increase of my solving time compared to what I previously had.
Thanks for any feedback, I get back to it.
again thank you for your advices.
I did look into the equations more precisely which helped a lot. Actualy the fluid model only add the velocity field variable in the equation.
So my model is working pretty good, and I coupled it with some laminar flow. But now I am working in stationary (since I figured using time was useless for my model) and I can't find how to plot and evaluate the evolution of the solution which is critical to understand it.
Also would you know if the post-processing allows to have a separated color table for one of the subdomains.
Moreover I tried to solve with time, in order to compare the results but the solving stay stuck like it entered an infinite loop. The same operations follows one another and the progress bar do not move, event after 3 hours. I know solving can sometimes take even longer but that already seems like a huge increase of my solving time compared to what I previously had.
Thanks for any feedback, I get back to it.
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Alright, I found how to keep the evolution of my solution. You have to go to your solve sequence > "Variable" > expand "output" sub menu > select "variable not solved for" in the "keep solution" selection tab