Discussion Forum

Hi

It sounds astonishing that you see no change, and 120 sec of 4kW/cm^2 = 120*4E3/1e-4 = 1.2GJ/m^2 you shold be able to do a simple hand calculation to see about how much you could expect as T rise, given the area, density heat capacity and conduction you seem to know

--
Good luck
Ivar

I think it has something to do with the 'temperature' node inside the 'heat transfer in solids' physics node. When I'm enabling the 'temperature' node, does it define the initial temperature? Or does it mean that the surface I've selected has to remain at a constant, predefined temperature? Also, if I want to make the heat transfer purely 1 dimensional (say through thickness only), do I have to insulate the other boundaries/surfaces? I have applied the "General Inward Heat Flux" node to specify the heat flux that is being applied, what is the difference between this and the 'Inflow Heat Flux" node?

So I'm applying high heat flux to one side of the brick while insulating the sides, however there's no (or very negligible) increase in temperature within the break. I'm attaching the .mph file here so that you can comment on it.

One final (and somewhat silly) question, does COMSOL 4.3 automatically convert any unit to it's basic unit? For example, if I define a density parameter as 1 gm/cm^3, will COMSOL do the calculations by converting this to kg/m^3 ?

Thanks.

Hi

You have a specific node for the initial conditions (on the domain, boundaries gets their temperature from an average of the boundaries (if they are sheared, else from the domain itself)
The temperature node imposes a flux field Q_bc(x,y,z,t) on this boundary such that the temperature remains constant to your predefined value T0 (or field T0(x,y,z,t))
COMSOl uses Si units by default, so if write [kW/cm^2] as units for a flux then COMSOl will look after the conversion to W/m^2 for you (very handy and allows constant unit check for you.
Note: a red field means typo error, this will not solve, an orange field means unit errors, this will solve but the user must ensure that the unit error still and hence the result makes sens, and a balck field means everything is OK (to solve ;)
In 1D you have only "point" boundaries, the domain or "line" has no "lateral sides" so no extra isolation is required. In your case I would expect one point attached to a temperature BC and one to a heat flux BC
(PS I do not yet have 4.3 so I cannot open your model)


--
Good luck
Ivar

Thank you so much Ivar. Let me get this straight, so you are saying I should always convert my units to their corresponding S.I units to generate the correct results? Does that mean I can't use U.S units to define things like parameters or material properties?

Also, my problem geometry is in 3-D (not 1-D as my previous conversation might wrongly imply). However, I want to simulate a 1-D (only through the thickness of the block) heat conduction scenario through the brick and measure the thermal stresses or other physical changes to the block. In my case, I have a 4x4x2 inch block of refractory brick that is being exposed to a high heat flux (*flame) on one side for around 120sec. Then the flame is taken out and the heat is allowed to propagate through the brick for around 10 more minutes. This is when most of the physical and mechanical changes (and I guess some thermal changes as well) happens to the brick. So I need a way to simulate applying the flame for 120 sec and then take out the flame and let it cool for 10 more minutes, all in the same simulation. How can I do that? Do I need a step function to define two different time steps with two different types of definition for the heat source (the first 120 sec with a flame and the rest of the time for free propagation)

Thanks for all your cooperation (and trouble :) )

Hi

first of all you can redefine the SI to other default units, check the top model node somewhere (might have changed location in 4.3), but you can also mix units COMSOL will convert into the default ones (personally I prefer SI, but that's another story about "taste and colors" as we say in French ;)

if you model in 3D but its a 1D problem you can also try to do it in 1D which implies a constant section "A"

Well I understand you are in a transient solver case, then I would say, take care w.r.t. your mesh and the heat diffusivity, be sure you resolve the gradient correctly with the link of mesh density, time stepping and heat diffusivity (check the older threads on that)

addding a time dependent heat source is rater trivial, use a step() function to define a smooth transition from 1 to 0 at your time t = 120 sec and multiply your Q0*step1(t[1/s]). then let your solver run for the full 10 minutes or more

--
Good luck
Ivar

Ok, so for the transition I've defined a step function which goes from a value of 1 to 0 at a location of 120 seconds and let the transient study run for 600 secs. Is that the right way to do what you told me to do?

I'm still not understanding what kind of boundary conditions to apply to my model. Like I said previously, the heat flux is being concentrated on one side and is expected to be conducted through the material and reach the opposite side. I've insulated the other four boundaries on the side of the block. Should I define an 'outflow' condition at the opposite side of the flame and an 'inflow heat flux' (rather than just 'heat flux') at the side of the flame? I'm still not getting a rise in temperature inside the block. Note that, the material has a very very low thermal conductivity. But I've run the simulation for t=20 minutes as well and still there's no change in the temperature of the block.

Also, this is the very first (and a highly simplified) step for my research work. At the end I'd have to model a way that simulates material removal from top of the brick due to melting by an impinging flame.

Hi

I would say apply a heat source with your time dependence for Q on the flame side, and apply a T0 constant temperature on the opposite side, and isolation on the 4 lateral sides.

You could also consider a heat cooling by convection on the opposed side with a T ambient and a h[W/m^2/K] of perhaps 2-5. As you solve for T you need to have a constant reference temperature somewhere to balance the flux, otherwise your T will always increase with no limit (still possible to study with a limited time time-series)

--
Good luck
Ivar

Thanks Ivar. I'm still puzzled as to why the temperature at the back side of the brick is not increasing, even after 600 seconds and being exposed to such high temperatueres! This simple model available at the COMSOL website also shows the same thing (the temperature not changing) : www.comsol.com/showroom/gallery/453/ . It's basically heat conduction through a cylindrical model where a very high heat flux is being applied on one boundary.

Also, I need to model a flame being impinged upon the surface of the brick and calculate (or perhaps even map) the pressure and temperature profile on the surface on the brick. Any suggestion how I should approach this? Should I draw like a air-box/flow-box around the brick for this? Or can I just simply specify the impingement surface and impingement velocity? How can I model a 'flame' that is a mixture of two different combustible gases? Is there any option for that in the heat transfer module (I'm still looking through the manual, that is why I'm bugging experts like you :D )?

Thanks again!

Hi

heat diffusion is an interesting phenomeny, not trivial I find.
Have you cheked the heat diffusivity of your material
alpha = k/rho/Cp in m^2/s ?
It gives you an indication of the "speed of propagation" of a heat front in a material, take a look at a few materials, air, water, aluminium or copper, steel and a brick or ceramics / glass for these alpha is greatly different.

This explains also why you can pick up a hot ceramic plate with youre fingers, while you will burn yourself if you take a copper or silver sppon at even just half the temperature !

And if you go to MEMS or nano deveices, take care, your habits have to be adapted, as the volume to surface effects changes drastically the thermal responses, not to say at these scales a system is often highly anisotropic from its size ratios

If its the fluidics of the flame you are looking for, that is another story, I would start with a thermal and fluidics approach, probably a conjugated heat model in turbulent (?) mode, but I'm no specialist of flames ;) and normally flames imply more even a plasma if you really go into details, not to say radiative exchanges. Anyhow I would have to run a litterature search and read a few articles to find a good first simple approach, unfortunately I do not have time for that, but probably you have some info about flames already.

--
Good luck
Ivar

Hello,

I´m a researcher of the Microelectronics Institute of Seville (IMSE), and I would like to make a something similar simulation to what you were talking about . ¿Could you help me a little bit with the modeling process, please?
I would like to make a simulation of a nano-bar of aluminium on a substrate (modeling a nanoantenna), receiving heat radiation. In principle, this could be a heat point source (located 1m away from the antenna on a normal line coming up from the middle point of the antenna), but we could be also interested in an external source coming from the sun, as modeled in the Heat Transfer module. I have here attached the model I built.

Thank you anyway,

kind regards,
Alfredo