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 The information provided may be out of date.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.

HT-block sliding with radiation

Please login with a confirmed email address before reporting spam

Hi Ivar

I wanted to once again thanks for your help. so I did the HT-block sliding in 3D and I got what I wanted now I need to add radiation to it. I could add radiation and detect heat flux radiosity simply by adding surface-to-surface-radiation commend when I was using parametric sweep but I do the same things but it does not works now that we have ALE in model. I got the error saying "segregated solver steps do not include all components"!.

I really do not know what does this mean I have included every thing! I think there is something with the ALE, could you please take a look at model and give me your comments. I appreciate.

thanks Ivar




22 Replies Last Post Apr 2, 2013, 9:17 p.m. EDT

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Mar 29, 2011, 3:48 p.m. EDT
Hi Ivar

I worked on the problem and still could not get any way for the error :(.
where does that come from ?!

Thanks
Hi Ivar I worked on the problem and still could not get any way for the error :(. where does that come from ?! Thanks

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 Mar 29, 2011, 4:42 p.m. EDT
Hi

first of all when you have added the radiation exchange, the depepndetn variable radiousity "J" should be added, it seems that COMSOL has not added it by default since last you updated your solver nodes, so this needs to be done.

Best is to have COMSOL doing it (ie.e. delete the sovler nodes and remake a new set, via the COMSOL GUI but this will also delete your results section

Then you need to try a gain ;)

I'm not sure how COMSOl handles radiation exchange and such moving mesh,and if the "shadow of the domains really exchange. To be trested.

my suggestion, forget 4.0, and build and use only 4.1, jumping from one to the others is not making the model better, and 4.0 was an early version, the updated 4.1 is really good, and I hear that more is comong in v4.2 soon ;)

--
Good luck
Ivar
Hi first of all when you have added the radiation exchange, the depepndetn variable radiousity "J" should be added, it seems that COMSOL has not added it by default since last you updated your solver nodes, so this needs to be done. Best is to have COMSOL doing it (ie.e. delete the sovler nodes and remake a new set, via the COMSOL GUI but this will also delete your results section Then you need to try a gain ;) I'm not sure how COMSOl handles radiation exchange and such moving mesh,and if the "shadow of the domains really exchange. To be trested. my suggestion, forget 4.0, and build and use only 4.1, jumping from one to the others is not making the model better, and 4.0 was an early version, the updated 4.1 is really good, and I hear that more is comong in v4.2 soon ;) -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Mar 29, 2011, 5:44 p.m. EDT
Hi Ivar

so I just deleted the solver and again compute the results for time dependent and it worked.
yes I am doing all simulations in V4.1. so I did what you said and again you were right :) and now another problem:(
the error says there is a singularity in the model! the model works awesome without radiation and radiation alone works well too but the combination is problem :(

I want to simulate the surfaces have radiation in the model and also add the hole wall and the above box to have radiation as they get heated and also reflect radiation coming form hot sliding box too. it was so easy with parametric sweep so what is here now :(
please see attached file, appreciate it. in time about 0.034 it says there is probably singularity where it is not even touched the block.

thanks a lot Ivar
Hi Ivar so I just deleted the solver and again compute the results for time dependent and it worked. yes I am doing all simulations in V4.1. so I did what you said and again you were right :) and now another problem:( the error says there is a singularity in the model! the model works awesome without radiation and radiation alone works well too but the combination is problem :( I want to simulate the surfaces have radiation in the model and also add the hole wall and the above box to have radiation as they get heated and also reflect radiation coming form hot sliding box too. it was so easy with parametric sweep so what is here now :( please see attached file, appreciate it. in time about 0.034 it says there is probably singularity where it is not even touched the block. thanks a lot Ivar


Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Mar 30, 2011, 12:12 p.m. EDT
Hi

it does not look to wanted to be solved :( whatever I do this kind of singularity comes up !! it would not be correct answers for me if I do it in septate modeling using parametric sweep. I want the blocks to also have radiation as they are being heated by the hot box during the moving mesh :( what is the problem of the attached modeling Ivar could you help please. I know I ask a lot but I really do not know what to do and I can not find similar example :( could take a loook at the model? I appreciate.
thanks
Hi it does not look to wanted to be solved :( whatever I do this kind of singularity comes up !! it would not be correct answers for me if I do it in septate modeling using parametric sweep. I want the blocks to also have radiation as they are being heated by the hot box during the moving mesh :( what is the problem of the attached modeling Ivar could you help please. I know I ask a lot but I really do not know what to do and I can not find similar example :( could take a loook at the model? I appreciate. thanks

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Mar 30, 2011, 12:49 p.m. EDT
Is view factor caching switched off? Could be important, since they are changing due to ALE.
Is view factor caching switched off? Could be important, since they are changing due to ALE.

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Mar 30, 2011, 1:03 p.m. EDT
Hi
I could not find that where is the view factor, radiation direction is positive normal direction and view factor option is not in the options!
thanks
Hi I could not find that where is the view factor, radiation direction is positive normal direction and view factor option is not in the options! thanks

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 Mar 30, 2011, 3:17 p.m. EDT
Hi

as the view factor is the relative area of the other surface seen from the first, it is changing with the ALE, but I have no idea how and if COMSOL is taking this correctly into account (with ALE).

This must be carefully tested (separately), just as what happens when one surface (the cylindrical part of the hole) is hiding part of the emitter surface ?

--
Good luck
Ivar
Hi as the view factor is the relative area of the other surface seen from the first, it is changing with the ALE, but I have no idea how and if COMSOL is taking this correctly into account (with ALE). This must be carefully tested (separately), just as what happens when one surface (the cylindrical part of the hole) is hiding part of the emitter surface ? -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Mar 30, 2011, 6:17 p.m. EDT
Hi Ivar

thanks for comment should not it be just showing zero in this case?! because there is no radiating thing. I did it using parametric sweep and it worked. so here is another suggestion:
is there the possibility in COMSOL to do this manually, I mean suppose first I do the heat transferee simulation for several steps as the hot block goes between boxes and no that we have temperature distribution in all model then somehow from results to add surface radiation to surfaces (which already have heat transfered and have temperature distribution) and then ask COMSOL to solve the surface radiating problem to get radiative heat flux.

Is it doable in COMSOL? I do not know what else can I do because as I said it works using parametric sweep.

thanks
Hi Ivar thanks for comment should not it be just showing zero in this case?! because there is no radiating thing. I did it using parametric sweep and it worked. so here is another suggestion: is there the possibility in COMSOL to do this manually, I mean suppose first I do the heat transferee simulation for several steps as the hot block goes between boxes and no that we have temperature distribution in all model then somehow from results to add surface radiation to surfaces (which already have heat transfered and have temperature distribution) and then ask COMSOL to solve the surface radiating problem to get radiative heat flux. Is it doable in COMSOL? I do not know what else can I do because as I said it works using parametric sweep. thanks

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 Mar 31, 2011, 10:03 a.m. EDT
Hi

study the segregated solver approach(es) in the doc

--
Good luck
Ivar
Hi study the segregated solver approach(es) in the doc -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Mar 31, 2011, 5:29 p.m. EDT
Hi Ivar

thanks for your advise, this vast field of COMSOL problems make it very hard to find good and relative examples, could you please let me know some useful examples for this topic or have you any realted to this to send? I appreciate your helps alot

thanks again
Hi Ivar thanks for your advise, this vast field of COMSOL problems make it very hard to find good and relative examples, could you please let me know some useful examples for this topic or have you any realted to this to send? I appreciate your helps alot thanks again

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Apr 1, 2011, 3:37 p.m. EDT
Hi Ivar

could you please take a look at this model :( I have defined two surfaces to radiate and also used the moving mesh so as the hot block comes over the hole the cylinder should absorb radiation and take effect but it does not runs :( when I do the 'compute' it says 'repeated error failure, may have reached a singularity' :( could you please let me know what is the problem now! I appreciate it totally.


thanks
Hi Ivar could you please take a look at this model :( I have defined two surfaces to radiate and also used the moving mesh so as the hot block comes over the hole the cylinder should absorb radiation and take effect but it does not runs :( when I do the 'compute' it says 'repeated error failure, may have reached a singularity' :( could you please let me know what is the problem now! I appreciate it totally. thanks


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 Apr 2, 2011, 11:22 a.m. EDT
Hi

I'm not sure there are any "problems" ;)

Have you checked the heat load you expect between the two parts ? for radiations and for contact ?.

Radiative exchange is low compared to contact, so I would adwise that you turn off (disable the continuity pairs) to leave those "heat isolated" and look only at the heat exchange in radiative mode. Then probe the integrated heat arriving on the small cylindrical part over time, I see a change in sign of the heat flux arriving after some 0.35 sec hence about when I see the part sliding in. Then compare this to your rule of thumb hand calculations ... (always verify and validate your models, step by step)

Furthermore, to be more relaistic, you should add a reasonnable heat loss on the contact surface, not use "just" identity pair continuity, as these heat losses, due to surface rugosity, poor contact etc, are rather high, if not to say VERY high, compared to bulk conduction (in the real world)

--
Good luck
Ivar
Hi I'm not sure there are any "problems" ;) Have you checked the heat load you expect between the two parts ? for radiations and for contact ?. Radiative exchange is low compared to contact, so I would adwise that you turn off (disable the continuity pairs) to leave those "heat isolated" and look only at the heat exchange in radiative mode. Then probe the integrated heat arriving on the small cylindrical part over time, I see a change in sign of the heat flux arriving after some 0.35 sec hence about when I see the part sliding in. Then compare this to your rule of thumb hand calculations ... (always verify and validate your models, step by step) Furthermore, to be more relaistic, you should add a reasonnable heat loss on the contact surface, not use "just" identity pair continuity, as these heat losses, due to surface rugosity, poor contact etc, are rather high, if not to say VERY high, compared to bulk conduction (in the real world) -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Apr 4, 2011, 9:44 a.m. EDT
Hi

thanks for your information Ivar. I see there are lots of parameters in the problem, but what is important is just the radiation flux absorbed by the cylinder so I do not want the cold block to be effected by the radiation which I know does not have much effect because the cylinder is supposed to just detect the radiation of the hot block and not from anywhere else. so all I need is to simulate the cylinder to detects radiation from the hot block as ht.gm and temperature change because of the radiation next I will make it in contact with cold block with an isolator and find its temperature.

could you please send me or introduce me some examples please :( I am confused now again

thanks
Hi thanks for your information Ivar. I see there are lots of parameters in the problem, but what is important is just the radiation flux absorbed by the cylinder so I do not want the cold block to be effected by the radiation which I know does not have much effect because the cylinder is supposed to just detect the radiation of the hot block and not from anywhere else. so all I need is to simulate the cylinder to detects radiation from the hot block as ht.gm and temperature change because of the radiation next I will make it in contact with cold block with an isolator and find its temperature. could you please send me or introduce me some examples please :( I am confused now again thanks

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 Apr 4, 2011, 11:15 a.m. EDT
Hi

I now, I say/write too many things in the same line ...

Start to disable the identity pair continuation HT, and look what is happening only with your radiative exchange.

When you have checked that the values you observe are what you expect, then you can add the conduction again (which currently are hiding your radiative effects, because of the large difference in power flux)

--
Good luck
Ivar
Hi I now, I say/write too many things in the same line ... Start to disable the identity pair continuation HT, and look what is happening only with your radiative exchange. When you have checked that the values you observe are what you expect, then you can add the conduction again (which currently are hiding your radiative effects, because of the large difference in power flux) -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Apr 4, 2011, 12:14 p.m. EDT
Hi Ivar

thanks for your comments, I understand it, in my simulation it not important how much is the effect of the radiation on temperature difference the cylinder is supposed to be a Infrared detector so I just need to get the surface irradiation from that (ht.Gm) so when I disable the identity part it works, but the point is the hot block is in its initial value because no conduction is defined between hot and cold block and when I do that the error of repeated error or singularity comes up :(

please take a look at the model it also looks like it does not consider the walls of the hole in the radiation because the emitting surface should be like part of the circle (part of hot block which is over the hole) but here it looks like hole walls are not considered :(

the attached picture is what I did with parametric sweep which has good results but just for completing modeling I need to define this new model.
Hi Ivar thanks for your comments, I understand it, in my simulation it not important how much is the effect of the radiation on temperature difference the cylinder is supposed to be a Infrared detector so I just need to get the surface irradiation from that (ht.Gm) so when I disable the identity part it works, but the point is the hot block is in its initial value because no conduction is defined between hot and cold block and when I do that the error of repeated error or singularity comes up :( please take a look at the model it also looks like it does not consider the walls of the hole in the radiation because the emitting surface should be like part of the circle (part of hot block which is over the hole) but here it looks like hole walls are not considered :( the attached picture is what I did with parametric sweep which has good results but just for completing modeling I need to define this new model.


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 Apr 5, 2011, 2:49 a.m. EDT
Hi

indeed that is what I feared, you could try to tell Comsol that the cylindrical wall is opaque, hence to shield your sensor. Radiation is a tricky part, as view factors are difficult to handle fully and correctly.

My first quick try did not seem to work. A little trick, to see only the radiation on a given boundary plot

ht.Gm*(dom==21)

where dom==21 is the boundary id number 21 for your receiver circular surface

--
Good luck
Ivar
Hi indeed that is what I feared, you could try to tell Comsol that the cylindrical wall is opaque, hence to shield your sensor. Radiation is a tricky part, as view factors are difficult to handle fully and correctly. My first quick try did not seem to work. A little trick, to see only the radiation on a given boundary plot ht.Gm*(dom==21) where dom==21 is the boundary id number 21 for your receiver circular surface -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Apr 5, 2011, 12:02 p.m. EDT
Hi Ivar

it really looks like COMSOL can not do the radiation and moving mesh together. please look at the attached file.
indeed when I cancel moving mesh it works even without defining a opaque material because the surface above the circular object is radiating, anyhow even by defining the opaque mesh it did not work again with moving mesh but it is good without that.

please let me know if you have comments for this. it looks like I have to first run a pure conduction simulation and then find the temperature of hot block as it gets over the hole and then in a separate model do the radiation without moving mesh, what do you think? (can I ask you to send me the file if you could solve it please, whereas i am still new it sometimes take much time to go for command you mention) thanks a lot.

Bests,
Hi Ivar it really looks like COMSOL can not do the radiation and moving mesh together. please look at the attached file. indeed when I cancel moving mesh it works even without defining a opaque material because the surface above the circular object is radiating, anyhow even by defining the opaque mesh it did not work again with moving mesh but it is good without that. please let me know if you have comments for this. it looks like I have to first run a pure conduction simulation and then find the temperature of hot block as it gets over the hole and then in a separate model do the radiation without moving mesh, what do you think? (can I ask you to send me the file if you could solve it please, whereas i am still new it sometimes take much time to go for command you mention) thanks a lot. Bests,


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 Apr 6, 2011, 1:42 a.m. EDT
Hi

to get the best reply, send the request to support, and pls report back for us others out here
it's an interesting remark, but I do not see why its should be any different.

What I know is that heat radiation and view factor calculations in complex geometries is a programme of itself (that you normally buy at a price tag higher than for the full COMSOL multiphysics package, so I not that astonished ;)

--
Good luck
Ivar
Hi to get the best reply, send the request to support, and pls report back for us others out here it's an interesting remark, but I do not see why its should be any different. What I know is that heat radiation and view factor calculations in complex geometries is a programme of itself (that you normally buy at a price tag higher than for the full COMSOL multiphysics package, so I not that astonished ;) -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Apr 6, 2011, 9:52 a.m. EDT
yes Ivar

I did that already and here is the response :


Dear Navid,

Thank you for contacting COMSOL Support.

In your model, when either of the thermal radiation or the heat continuity
across the moving block and the stationary blocks is switched off, the
model works fine. When both of them are present, the model works fine till
the moving block gets in touch with the stationary block and then the
solver fails to converge. It seems that the solver is not able to handle
the combined effect of the surface to surface radiation and the heat
continuity for the bottom surface of the moving block. We are currently
investigating this issue and will get back to you soon.

Best Regards,

COMSOL Support


I am running out of time so I think I have to do the model just considering heat conduction with ALE and then take the temperatures of each step from the result and do a radiation model according to the temperature manually and get the radiation power flux absorbed by the cylinder.

what do you think Ivar?

Bests,
yes Ivar I did that already and here is the response : Dear Navid, Thank you for contacting COMSOL Support. In your model, when either of the thermal radiation or the heat continuity across the moving block and the stationary blocks is switched off, the model works fine. When both of them are present, the model works fine till the moving block gets in touch with the stationary block and then the solver fails to converge. It seems that the solver is not able to handle the combined effect of the surface to surface radiation and the heat continuity for the bottom surface of the moving block. We are currently investigating this issue and will get back to you soon. Best Regards, COMSOL Support I am running out of time so I think I have to do the model just considering heat conduction with ALE and then take the temperatures of each step from the result and do a radiation model according to the temperature manually and get the radiation power flux absorbed by the cylinder. what do you think Ivar? Bests,

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 Apr 7, 2011, 1:24 a.m. EDT
Hi

if the solver is OK in plain stationary, you could perhaps envisage to use a full parametric sweep, and a sequence of stationary solver steps instead of the ALE, it might do the job

--
Good luck
Ivar
Hi if the solver is OK in plain stationary, you could perhaps envisage to use a full parametric sweep, and a sequence of stationary solver steps instead of the ALE, it might do the job -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Apr 7, 2011, 10:43 a.m. EDT
Hi Ivar

I got what you said, but I have not seen any example about 'a sequence of stationary solver steps instead of the ALE'
could you please introduce or send me one related to this. I really appreciate it.
is this what you said is related to the segregated solver?! I appreciate your help to tell me how to use this option specially for my problem.

Thanks,


Hi Ivar I got what you said, but I have not seen any example about 'a sequence of stationary solver steps instead of the ALE' could you please introduce or send me one related to this. I really appreciate it. is this what you said is related to the segregated solver?! I appreciate your help to tell me how to use this option specially for my problem. Thanks,

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Apr 2, 2013, 9:17 p.m. EDT
Hello Navid

I speculate what Ivar implies is that instead of ALE, you should use Parametric Sweep (right click on study and choose parametric sweep) by which you can solve your model for different geometries in an stationary mode. In other words, it solves your model for different positions of the moving part but the problem is not time-dependent. It sweeps the part and solves the model for each position separately!

Good Luck
Hello Navid I speculate what Ivar implies is that instead of ALE, you should use Parametric Sweep (right click on study and choose parametric sweep) by which you can solve your model for different geometries in an stationary mode. In other words, it solves your model for different positions of the moving part but the problem is not time-dependent. It sweeps the part and solves the model for each position separately! Good Luck

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.