Discussion Forum

Posted: 1 decade ago 2 apr 2013, 14:52 GMT-4

Hi those two conditions are slightly in contradiction. As when you decide to fix a boundary temperature, then Comsol will locally adapt the power flux such to get the boundary to stabilise at the desired temperature, but this will most proabably not always be 1 [W] (as average). When you fix boundary temperature you are in fact saying you can extract enough enerxy to enforce the desired value (without any limitation). And if you fix he hea flux over the boundary, the emperature will adapt to respect the heat gradient across your domain, hence will mostly not be a fixed temperature. In your case you have a sort of controller, with a min temp of 4.2[K] for max 1[W] flux, so you need to define rather a heat flux that saturates at 1W for T boundary > 4.2[K] and reduces to 0W at T=4.2[K] -- Good luck Ivar

Posted: 1 decade ago 2 apr 2013, 20:35 GMT-4

[QUOTE] Hi those two conditions are slightly in contradiction. As when you decide to fix a boundary temperature, then Comsol will locally adapt the power flux such to get the boundary to stabilise at the desired temperature, but this will most proabably not always be 1 [W] (as average). When you fix boundary temperature you are in fact saying you can extract enough enerxy to enforce the desired value (without any limitation). And if you fix he hea flux over the boundary, the emperature will adapt to respect the heat gradient across your domain, hence will mostly not be a fixed temperature. In your case you have a sort of controller, with a min temp of 4.2[K] for max 1[W] flux, so you need to define rather a heat flux that saturates at 1W for T boundary > 4.2[K] and reduces to 0W at T=4.2[K] -- Good luck Ivar [/QUOTE] Thank you for your reply. Actually, you may know about cryocooler, they have cooling capacity of 1 W at 4.2 K. It means that when we connect our cryocooler at particular part of the system and heat conduction to that part from other part of the system say at 300 K or 77K is less than 1W ( Conduction heat loss) then we can keep the temperature of that part at 4.2 K. To be very honest, I have design my system and have calculated anyalytically that the heat load ( conductive, radiative) at that part is less thatn 1 W. Now, I want to put boundary conditions to my model and want to confirm that whether with 1 W cooling capacity, I am able to keep temperature of particular part of the sytem at 4.2 K or not? OR is there any gradient of the Temp. is there or not? Here, I am attaching my MODEL to you, hope you may help. I need to give 1W cooling power to domin 47 ( Actually, it is copper plate and I will connect the 2nd Stage of my Cryocooler with it). There are three chambers of the system, 1. outer is cryostat which will stay at Room Temperaure ( made up to SS) 2. Middal is Copper shield for radiation guard ( made up of Cooper) remain at 43 K. 3. Inner one is solid nitrogen shield ( made up of SS) and will contain solid nitrogen of 4.2 K. Some pipes are connected from top of the cryostat for current leads, cryogen transfer and voltages tapes feed through. Model will experience conductive heat load from 300 K to 4.2 K with some intruption at 43 K. If you will give or suggest appropriate boundary condition to be then it will be better for me. Also, if you can suggest me how Can I incorporate radiation in my model then it will really help me. I could not attached my MODEL becasue it size is 16 MB, is there any other way to send model to you. Waiting for your reply.

Posted: 1 decade ago 3 apr 2013, 02:18 GMT-4

Hi One conditions that Physics is imposing => to have a heat flux you need a temperature gradient and/or the other way(check your formula and play with simple HT models) so it's not physical to fix a temperature AND to set a boundary heat flux imposed on that same boundary, only one of the two might be driving, the other is driven For any heater / cooler specification we have limit values as max or min T, and max or min Q0 but these are in fact functions of each other, this needs to be untangled to get a model correctly expressed Setting up a model often learns us a lot on how the physics is really controlling our case. -- Good luck Ivar

Posted: 1 decade ago 3 apr 2013, 09:19 GMT-4

Thank you for your reply. Actually, today I have tried with Heat Source and I have given total power P tot = -1 W and it is giving something expected result. What do you think, can I give my cooling power at particular location with Heat Source ( P tot) = (negative) 1 W.

Posted: 1 decade ago 3 apr 2013, 11:12 GMT-4

Hi Yes if you say 1W heat source and play with the signs such that it goes "out" of the domain, you will extract that power, and the temperature you get out is the final one, but to stop at 4.2K you must also have 1W coming in on your model from external convection to any residual gas, conduction or radiation. To be in steady state -- Good luck Ivar

Posted: 1 decade ago 3 apr 2013, 19:54 GMT-4

I think this will come and my problem may have solved. Still if I will have problem then I will let yoiu know.

Posted: 1 decade ago 4 apr 2013, 02:27 GMT-4

I have checked by giving -1 W it is working. Good Output of Discussion. Now, I want to incorporte radiation effect in my model. I am using surface to surface radiation, actually there are three surfaces. Outer one, which is at 300 K radiating on the 77 K surface and 77 K surface radiating on the 4.2 K surface. I have selected my 77 K surface ( only outer surface rather then entire domain, means when I click on that part, on first click it is selecting the outer surface, when I click second time on it, it is selecting the next surface. Means if I have solid hollow cylinder, when I click on the outerside, it select outer surface but when I click second time it is selecting inner surface, so you understand? ) and give temp 77 K in model input and 300 K at T amb. Similary, I have select 4.2 K surface ( only outer surface rather then entire domain, same as above) and give temp 4.2 K in model input and 77 K at T amb. When I tried to run for commute, it is showing that you need at least one adjecent opaque surface. Also, there was comment that you go to surface where you had given temperature and made surface opaque. I went to opaque node and select all the surfaces where I had given model temperaure but still it is giving same error. In short, I want to know that how radiaton effect in 3D, I can incorporate.

Posted: 1 decade ago 18 apr 2013, 07:46 GMT-4

[QUOTE] Hi those two conditions are slightly in contradiction. As when you decide to fix a boundary temperature, then Comsol will locally adapt the power flux such to get the boundary to stabilise at the desired temperature, but this will most proabably not always be 1 [W] (as average). When you fix boundary temperature you are in fact saying you can extract enough enerxy to enforce the desired value (without any limitation). And if you fix he hea flux over the boundary, the emperature will adapt to respect the heat gradient across your domain, hence will mostly not be a fixed temperature. In your case you have a sort of controller, with a min temp of 4.2[K] for max 1[W] flux, so you need to define rather a heat flux that saturates at 1W for T boundary > 4.2[K] and reduces to 0W at T=4.2[K] -- Good luck Ivar [/QUOTE] I have a question concerning the controller, how can this be modelled smoothly in a time dependent study. If you are doing such kind of controller when reaching the 4.2[K] the solver will take very small time steps, because it must continously check wheter or not the T is above or under 4.2[K] Kind regards Hans Parmentier

Posted: 1 decade ago 19 apr 2013, 11:02 GMT-4

Hi I'm not sure I fully understand your question. COMSOL solves this as a integral equation, hence more or less as a least square way -- Good luck Ivar

Posted: 1 decade ago 22 apr 2013, 01:57 GMT-4

Ivar, that's right but when doing such kind of studies it takes a very long time to calculate a time dependent study, because at a certain moment the system can not longer predict the next solution, because the temperature is just under or above the set temperature, at that moment the solver is taking very small times steps leading to very long calculation times. So a temperature range could be helpfull but then you have the same problem just at the end of the range.

Posted: 1 decade ago 22 apr 2013, 03:20 GMT-4

Hi may be, but for my cases I haven't noticed any unusual looong time to solve, perhaps you have a specific case -- Good luck Ivar

Posted: 1 decade ago 25 apr 2013, 03:11 GMT-4

It is indeed a special case, it is a larger system (with symmetries still 8 sensors looking if a temperature is above or under a certain temperature) depending on the position of the temperature an action must be done. So we need a long time to heat the system completely but the sensors are fast in the area were they switch off and on. Changing the calculation method to linear is improving a lot, but it still takes two-three days to calculate one simulation. So I was wondering if I could improve the temperature controller without losing accuracy (a minimum time step could maybe a solution?)