Discussion Forum

Hi

why do you have a GND + 10V electric potential (I would rather use a Terminal but OK) and in addition impose a contant current density ? That is one too much, let COMSOl resolve the current from the sections and the materials provided a total voltage drop. Just disable the last node (External Current density)
--
Good luck
Ivar

thanks for your suggestion. But I need a source that can provide a low-voltage and high current in the model.
Since I need high temperature at the center to heat up the sample (the smallest cylindrical section at the center) which would be heated by high current only. With this much less voltage alone, i dont think it would attain a temperature of 1200 deg. celcius.
So i think i need to define an external current source.

This is what i come up to... if you have any other suggestion to combat the problem i would gladly appreciate it.

And thanks for your reply.

Looking forward to your help.

hey?
now i used terminal node and i defined current as 2000 A as I needed for my experiment.
But I have few confusions regarding it:
1) which boundary should i select, i mean both the lower and upper ends or just one of them?
2) I also need to apply a voltage of 10 V.
How can I get both of them to be implemented at the same time across the ends? or in other words a power source that can maintain a potential difference 0f 10 V across the terminal and also can produce a current of 2000A.

Looking forward eagerly to your help.

_regards,
Nitesh

Hi

no probelm to define a current source and deduce the voltage at the end ;)
BUT I have a problem if you define a constant current density ina varying material and shape geoemtry, how will you enforce the total current to change like that ?

Use rather a current density on the inlet and outlet, or add avaraible and adaptthe voltage to a given inlet current density (inlet section
at the end where you put our terminal or GND. But you can also, use a total current and let COMSOl arrange the get the correct current DENSITY dependeing on the local sections

--
Good luck
Ivar

Hey I am trying it by applying terminal current of 2000A and nothing else.. i mean no voltage, no external current density.
But with the above situation, comsol has stopped working. I mean its taking to comsol years to render. It didnt take this much time to solve, even with the much high current density which i had applied earlier. So why with only the terminal current it is unable to solve.

I applied external current density to0 at the ends.. but it is not giving any result..

There are hundreds of types of current sources, which one to use. I tried everyone, but nothing is working out. Please help me out, I would be grateful to you..

I am attaching the model, please tell me where i am missing?

Hi

have you tried to:
1) enable your GND on boundary 7
2) Disable all your current sources
3) Add a boundary electric Terminal on boundary 10 (but NOT on 7) and set it to 2000[A] (total current through the boundary)

The terminal sets a current leaving the boundary, your two boundaries are opposed, if you use the same current direction on both boundaries it will not work, but with a GND and another current source it's OK, you can integrate the current density over your GND are to check you get your final current

Then to see what is happening turn on the Study Time dependent - Plot while solving FOR all solver steps

Then I get almost to 2000°C for the hottest even with your radiative exchange, you should perhaps also add some convective cooling (BC) but I'm not sure my standard "h=5-10W/m^2/K" parameter is really valid at those temperatures
--
Good luck
Ivar

Hi

have you tried to:
1) enable your GND on boundary 7
2) Disable all your current sources
3) Add a (or enable your) boundary electric Terminal on boundary 10 (but NOT on 7) and set it to 2000[A] (total current through the boundary)

The terminal sets a current leaving the boundary, your two boundaries are opposed, if you use the same current direction on both boundaries it will not work, but with a GND and another current source it's OK, you can integrate the current density over your GND are to check you get your final current

Then to see what is happening turn on the Study Time dependent - Plot while solving FOR all solver steps

Then I get almost to 2000°C for the hottest even with your radiative exchange, you should perhaps also add some convective cooling (BC) but I'm not sure my standard "h=5-10W/m^2/K" parameter is really valid at those temperatures
--
Good luck
Ivar

hey,

thank you for your kind help. I got my result!! :) :)

May i ask, what was the problem with the current density? I mean it was quite a huge deviation? why it was showing negative temperature.?
was it because the radiation heat transfer was higher than the heat produced by the current in the body?

And again an honest thanks for the help!!! :) :)

Hi

when you do time domain studies of heat propagation, you must ensure that your mesh density and time stepping is coherent with the heat diffusivity of your material, otherwise the meshing, which is a discretization of your functions will not catch the diffusion phenomena.
Think of sound sampling, if you want to catch the high frequency & pitch of a soprano, you ned to sample at high speed, else you get aliasing and frequency overlapping, the same happens in FEM with the meshing, if you cannot resolve the gradients and their changes in time, you will not get the correct results, or even funny negative temperatues (even below 0[K]) which is not very physical
--
Have fun COMSOLing
Ivar

thanks for the great help.. The example really cleared my concept(a bit) :) :)

but i need to flow nitrogen gas.. actually this whole set up is kept in an atmosphere where nitrogen is blowing from bottom for convecting cooling.

In comsol, shall i cover the setup in a block and make that block purely of nitrogen?
So for that do I need to add one more physics (i mean non-isothermal flow)? or that alone joule heating would be sufficient for it?? but if i am working in joule heating alone then i couldn't apply surface radiation to the walls.
Please help

Waiting eagerly for your reply.

Hi

indeed if you have N2 flow, it's rather conjugated heat flow you should use as you also have heat transport in addition to conduction and convection, and at which pressure for the N2, low a few Pa or high atmospheric ?)


then you need indeed a block around and quite some RAM and time to get results out ;)

Except if you have the material parameters for N2 and can use the natural convective cooling physics for the boundary cooling BC, perhaps easier for you without the true Fluid modelling, and quite as precise for open free convection cooling
--
Good luck
Ivar

so you mean, the easiest way to do it is convective cooling node--->heat transfer coefficient-->external natural convection on vertical walls, but the software is asking for the wall height.. what does it mean.. i entered it as 118mm as the height of my model.. is it correct?
and took air as an external fluid( i can take it as an approximation to nitrogen, right?) and selected all the outer boundaries for convective cooling and simulated it for 5 sec., 60 seconds.

Actually I performed a practical today for the same model but might be there are slight difference in properties, and practically the temperature came out to be around 200 deg. celcius.

But with the modeling the temperature is touching 475+ or i should say 700 C !!!
^^the above one i got while following the above steps given in first paragraph.

I am attaching the file..
Please......... please.. tell me where i am going wrong??

and yeah more thing i should have added in the previous message

actually we set the temperature 200 deg C at at the center. and from there the current follows the T(t) graph so as to keep the temperature close to 200 and thereby heat up the material.

How can i solve the situaion like this... do i need to make current as a function of temperature?
How can i do this?

Thanks in advance.

Hi

I have never used these advanced BC conditions, they are there to calculate the Hc W/m^2/K equivalent based on standard formulas that have been tested, pls read carefully the doc, and perhaps one or two of the references. normally you need to have different wall conditions for vertical, oblique and horizontal boundaries, furthermore, to get your current here, you have two large cables, probably Cu brand, these will also exchange quite a lot of heat with the device so you need a more reasonable heat flow condition at your GND and Terminal boundary. the effect from the cables is certainly greater than from the convective cooling.

pls use the correct BC for natural, or forced (blowing) conditions too

--
Good luck
Ivar

okay, i did what you suggested me... i followed the description given in the reference guide,( the problem given in the reference guide seems to be similar to mine.)

but now my comsol has stopped working.. its taking years to render it.

<can there be any easy way/trick by which i can run it in a quickest way?
and please check the model, this is the final model i have come up with
since i couldnt say anything about how close i am until it computes..
so can you please check it and tell me if i am wrong at some steps?

I would be grateful to you! :)

Hi

when you start to mix several physics its better to start one by one, and for example map the temperature of the joule heating to the internal boundary and of your object and run a "simple" flow case, get that one up running first then start to mix.

For example I would suggest to leave outlet pressure no viscous stress p=0 and set an ambient pressure of 7[atm] if that is the case (now you have 1[atm] see main Fluid node). Then check your velocity and the Reynolds cell number, your mesh is far too coarse for such speeds, I'm not even sure you are in laminar flow case at all, then you must use turbulent flow which is far more intense on settings and computer power.

not to say with fluidics in one side and joule heating on the other, you do heat your fluid but you have only transport phenomena no fluid heat conduction. I believe you need the conjugated heat flow module to mix that. but first get a flow case running, then make a new model with the complex conjugated HT.

Anyhow you fluid model will be wrong if you do not include the support and wires bringing the current in and out, and holding yuor object in free space

another question is how much the flow case would change, if you have such a steady flow velocity, it might be worth only calculating the flow in stationary case and use that for the fluid velocity, any case its worth to start with a steady state flow simulation (with other t=0 conditions) to establish good initial conditions for your fluid flow

--
Good luck
Ivar