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.

Please help me find how to achieve this function.

It seems to be a simple problem but I can't figure out how to make it.
Description: a 2d transient model of a square is built, the left, bottom, and right side is thermal insulation and the top side is under a convective cooling under air(300K). The domain has an initial temperature at 350K. Now I need my top side has a temperature at 400K for 15 seconds and then moved away to observe the cooling down by air. I tried two ways but none of them available.
1. In heat transfer module, I add a equation in "Temperature": 400*(t>=0[s] && t<=15[s])+(t>15[s]). After simulation, the temperature cool down rapidly to nearly 20K after 15s, and it seems that the temeprature of top side has turn to 0K after 15s.
2. In the study part, I separate the step into two steps(first is 0s-15s, and second is 15s-30s). In the second step, the "temperature" part was disabled. After simulation, the solution only provide the part of second step, but not the first step(0-15s).

Please help me find how to achieve this correctly? Thank you in advance.

4 Replies Last Post 07/12/2012, 02:41 GMT-05:00
Posted: 5 years ago 04/12/2012, 02:52 GMT-05:00
Hi

you cannot set a constant T BC and a convective cooling BC on the same boundary, these two conflict, its one or the other.

The easiest is to define the two BC, then use two time series solver sequences in series, the first one (0<t<15 sec) you disable the Convecting cooling and enable the constant temperature, then for the second (t>15sec) you enable the convective cooling and disable the constant temperature BC. This you can do in the solver Study step1 node under Physics and variable selection in v4.3a

--
Good luck
Ivar
Hi you cannot set a constant T BC and a convective cooling BC on the same boundary, these two conflict, its one or the other. The easiest is to define the two BC, then use two time series solver sequences in series, the first one (0

Posted: 5 years ago 04/12/2012, 09:39 GMT-05:00
Thank you very much Ivar!
In your description, I found it needs two "compute" of study with step 1 node. But here comes the problem. The first solution can't be stored (0-15sec) and after two computations, only the last one can be stored and shown.
So, can you tell me how to store the two separate compute results?
Thanks for helping me again.
The mph file is uploaded if you feel convenient to correct my model.
Thank you very much Ivar! In your description, I found it needs two "compute" of study with step 1 node. But here comes the problem. The first solution can't be stored (0-15sec) and after two computations, only the last one can be stored and shown. So, can you tell me how to store the two separate compute results? Thanks for helping me again. The mph file is uploaded if you feel convenient to correct my model.


Posted: 5 years ago 04/12/2012, 14:46 GMT-05:00
Hi

first some tricks ;)

When you have such nice geometries, you should rather use the layer, its much quicker to build your model like that, the layer have thickness that are cumulative (you define each layer per say) in this way you need only 1 rectangle

Your T232 cannot be defined in a Global parameter as it depends on "z" a local coordinate so it cannot be evaluated and Parameters are all evaluated once at study launch
But you can define a "model definitions variables" T232 valid on some Domain(s) or Boundaries, or all

Then you should move your Convective cooling up one level, before the Temperature BC definition. THis will make the convective cooling operate, but be disabled on the burn region while you are heating this one up with the "lower" Temperature BC case. And for your first solver case leave all BC's active.

Then enable your 2nd time study step, make it disable the Temperature BC (as you have done), but then turn ON the Initial values solved for, and point it to Study 1 Time dependent, you want the temperatures from study 1 too, not only all the other variables

When you now make a default solver, you get both steps, but indeed COMSOL does not store the first one ;)
So try to put it in yourself: right click study 1 add "Other Store solution", and move it up just after the node Time dependent solver 1 and before the Compile equations Time dependent 2, then SOLVE, you get now 2 data set, one per time solver

You have chosen a BDF free time stepping, I would suggest to use a "Intermediate" time stepping, or to use a time stepping as a power series (I often use a 2^{range(-2,1,8} to step between 0.25 and 256 sec in steps of power of 2, this gives a reasonable stepping change for diffusion problems like thermal decays)

One thing, your mesh is still rather coarse, try having at least 3 (or more) elements in the thickness, per layer, if you really want to resolve the gradient of T reasonably

--
Good luck
Ivar
Hi first some tricks ;) When you have such nice geometries, you should rather use the layer, its much quicker to build your model like that, the layer have thickness that are cumulative (you define each layer per say) in this way you need only 1 rectangle Your T232 cannot be defined in a Global parameter as it depends on "z" a local coordinate so it cannot be evaluated and Parameters are all evaluated once at study launch But you can define a "model definitions variables" T232 valid on some Domain(s) or Boundaries, or all Then you should move your Convective cooling up one level, before the Temperature BC definition. THis will make the convective cooling operate, but be disabled on the burn region while you are heating this one up with the "lower" Temperature BC case. And for your first solver case leave all BC's active. Then enable your 2nd time study step, make it disable the Temperature BC (as you have done), but then turn ON the Initial values solved for, and point it to Study 1 Time dependent, you want the temperatures from study 1 too, not only all the other variables When you now make a default solver, you get both steps, but indeed COMSOL does not store the first one ;) So try to put it in yourself: right click study 1 add "Other Store solution", and move it up just after the node Time dependent solver 1 and before the Compile equations Time dependent 2, then SOLVE, you get now 2 data set, one per time solver You have chosen a BDF free time stepping, I would suggest to use a "Intermediate" time stepping, or to use a time stepping as a power series (I often use a 2^{range(-2,1,8} to step between 0.25 and 256 sec in steps of power of 2, this gives a reasonable stepping change for diffusion problems like thermal decays) One thing, your mesh is still rather coarse, try having at least 3 (or more) elements in the thickness, per layer, if you really want to resolve the gradient of T reasonably -- Good luck Ivar

Posted: 5 years ago 07/12/2012, 02:41 GMT-05:00
Thank you so much Ivar! I have worked it out. Thanks a million!!
Thank you so much Ivar! I have worked it out. Thanks a million!!

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.