Discussion Forum

Posted: 1 decade ago 4 mars 2011, 09:17 UTC−5

Hi I believe you have forgotten something in your expatiations there you have energy entering, but nothing leaving so you can isolate all you want all will inevitably end up at 1400 [degC], no ? Once you have the total energy loss you can live with (assuming its taken from the inside of the Mo tube, you can enforce a fixed temperature on this surface at the desired value, and you integrate over the tube the total flux needed to get to this temperature, you compare it to your accepted heat flux and you continue to add layers until you get far enough down (if you manage) because even if you have 2 Ta sheets, how do they (both) loose energy/heat power ?, if they are isolated, they would also heat up and the shielding will loose efficiency Your model as you explain is easiest in 2D-axi, but probably its rather 1D-axi even So making a tube is simply a rectangle to the right of the r=0 line but then the fun start: 1) you cannot add geometries without restarting your model (topology change) manually, this should be OK for a few restarts 2) you need to explain/model how your Ta sheet exchange with the surrounding, on both sides 2a) air, contact with resistance, or radiative exchange with the Mo tube ? 2b) air, contact with resistance, or radiative exchange between the Ta sheets ? 2c) do you also have convection ? probably you could start to ignore that 3) if you add radiation on the external surface(s) be aware that you need inv4.1 to duplicate the material properties of your Ta, once defined for the domain and the second time defined for the boundaries (as radiative exchange asks for surface emissivity, by default from the material db 4) I get an error using the default T[1/K] varying surface emissivity when linked to radiative BC (funny even if its highly non linear it should be able to solve), so you might ignore 3) and put a user defined value directly Nice project, simple but not trivial, interested to hear what you get to -- Good luck Ivar

Posted: 1 decade ago 4 mars 2011, 10:14 UTC−5

Hello, and thank you for these helpful comments! [QUOTE] I believe you have forgotten something in your expatiations there you have energy entering, but nothing leaving so you can isolate all you want all will inevitably end up at 1400 [degC], no ? [/QUOTE] Yes you are right and that is the main reason I want to simulate the whole thing. Our system will be in vacuum (no convection, no heat conduction), with a current driven through Ta wires all along the Mo tube to heat it up (that's pretty similar to a filament lamp). The whole thing will be controlled using a s-type thermocouple and a PID circuit. I am still doing the math, but with an emissivity of 0.2 (i.e. reflectivity of 0.8), most of the heat should keep within the tube. But I need some feeling as to the cooling rate of the Ta sheet wrt to the reaction time of the whole system. [QUOTE] 1) you cannot add geometries without restarting your model (topology change) manually, this should be OK for a few restarts [/QUOTE] Do you mean I need a new model for each extra Ta sheet I need? Good point! No trouble though. [QUOTE] 2) you need to explain/model how your Ta sheet exchange with the surrounding, on both sides 2a) air, contact with resistance, or radiative exchange with the Mo tube ? 2b) air, contact with resistance, or radiative exchange between the Ta sheets ? 2c) do you also have convection ? probably you could start to ignore that [/QUOTE] All radiative. Can I define the rectangle (i.e. the cross-sections of the Mo tube, and Ta sheets) as being made of Mo and Ta? Doesn't Comsol Multiphysics have the db data for their emissivity? [QUOTE] 3) if you add radiation on the external surface(s) be aware that you need inv4.1 to duplicate the material properties of your Ta, once defined for the domain and the second time defined for the boundaries (as radiative exchange asks for surface emissivity, by default from the material db [/QUOTE] Sorry, what is inv4.1? I have not understood this point. [QUOTE] 4) I get an error using the default T[1/K] varying surface emissivity when linked to radiative BC (funny even if its highly non linear it should be able to solve), so you might ignore 3) and put a user defined value directly [/QUOTE] BC? Boundary condition?

Posted: 1 decade ago 4 mars 2011, 10:48 UTC−5

Hi OK its easy to add the Ta layers one by one, but you need to close your cavity with the temperature and the emissivity of the surrounding external from the Ta tubes, as these filaments will loose energy outwards too. I misunderstood slightly your model I see. Indeed you draw 2 rectangles (in 2D axi) for both material, surrounded by a 1/2 circle for the external loss, if you want to consider it as a closed cavity, and you should take a look at the example of Thermal radiative exchange in the model library Yes COMSOL uses DB data, if you ask it to do it, but the radiation exchange is a boundary effect, and today if you define a Domain with a material it applies to the DOMAIN but not to its boundaries, so you need to redefine the same material a second time, and tell COMSOL to apply it to the boundaries, that's all (easy to forget and not "logical"). But, then I noticed that the material data of Ta is not covering your temperature range, and the extrapolation is set by default to "constant", while "linear" would have been better in your case, you should change, and finally I got an error message that the temperature dependence was not accepted by COMSOL ? could it be that a temperature dependence on the emissivity for a T^4 relation is too much for COMSOL, I'm astonished ,but something is not ready, yet, for that you should check and if really so, ask "support" The v4.1 is the COMSOL version, you should use V4.1 with latest patch (4.1.0.154 or newer) to avoid errors And by the way do not forget to add "surface to surface radiation" in the main HT node -- Good luck Ivar