Discussion Forum

Hi

I had a very quick look at your model. I have a few things to comment (but I'm not sure it will help that much as I do not know wher you are finally going ;)

1) as you have these thin steel plates, why not simulate them as thin boundary plates, and leave out fully the thin side domains, this will lighten your model (Note: if you define steel as a material for a domain, you cannot use it for a boundary, you must redefine a second time the material and specify "boundary" in the material settings page, it took me some months to really understand this ;)

2) to debug your model, check the energy fluxes in & out of your model (integrate over the boundaries) you might well be loosing energy where you do no expect it

3) last you have the pressure defined at several places but I lack some coherence, I would use ather ambant pressure for the initial condition, then perhaps also at outlet

--
Good luck
Ivar

Hi Ivar,

Thank you for your answer. Will try and see if your advice works. Especially number one looks promissing!

Hi,

I would just like to add that the "out" messages in the solver log are normal and indicate that the solver outputs solution data to the model at the specified output times. In addition, the log constains all intermediate steps that the time-dependent solver takes to fulfill the tolerance criteria.

Regarding the model I'm not that familiar with the application, but I'm not sure that you need the highly conductive layer between the steel and the air domains.

Best regards,

Magnus Ringh, COMSOL

Hi

I agree both conductive layer and the thin steel plate is onetoo much, and I would try with only the conductive layer (assuming that that means perfect thermal isolation behind (external side) of this boundary layer ?)

--
Good luck
Ivar

Hi all,
With the highly conductive boundary layer, the simulation is way faster! But something puzzles me in this model. The temperature of the air becomes cooler in time......it drops way below zero (to -200 degC even) even though my inititial temperature is 25 degC..... has anybody an idea why this is?

Hi

that happens typically when one has lost/forgotten a BC of the ambient temperature, and COMSOL tries to use deault 0[K]. It represents a good heat drain, you will have no fire like that ;)

--
Good luck
Ivar

Hi Ivar, Thanks for the quick response. I see a great heat flux in my model through the axial symmetry boundary....there is nothing I can define at that boundary right?

Hi

are you sure you have "axial symmetry" for all (in each) physics then ? I do not have access to my COMSOL from here, and its time to listen to the COMSOL Webinar

--
Good luck
Ivar

Hi again

I have just had some time to look at your 2D axi model, I seee a few things:

I might have mislead you with the pressure, as there are two representation (absolute and gauge representatons), check carefully the doc too, if you use the Gauge pressure I seem to remember you start at p=0 and have pref = 1[atm]. Then for the outlet I believe it's an absolute pressure (1[atm]) this is something I would need to test to be sure I catch it correctly myself, so pls verify carefully)

Then for your forces/body loads in 2D axi, A assume the gravity is in -Fz and not radilly along "r", or are you rotating your pipe there ? Anyhow then you have the centrifugal forces too, so I suppose that the body force is slightly mixed up.

Finally I believe you need a denser mesh, I would say minimum 10-20 elements in the R direction, you could try a structured mesh, but there are some restriction on the mesh touching the symmetry axis, to avoid a too dens mesh you should be able to have a mesh shape that is longer (along the axis) than wide but do not overdo it lets say 1:5 ratio to avoid a too heavy model.

Then I reset your solver and startd from fres again and I did not get the tendensy to pull the temperature towards 0[K] any more. often its worth to rebuild fully the solver when you are chnging many things note that you loose your results in the same time, except if you make a new study and then manually connect all previous Results to te new study before you delete the old study

A last thing: update to the latest patch of 4.1 (see the main comsol site) there are a few typos corrected in the latest verision.

--
Good luck
Ivar

Hi Ivar,

Thanks again for your response. I will have a look at your comments. The pipe is indeed rotated.....it should represent a horizontal pipe. Is it possible to rotate/adjust the rotational axis?
I'll see if I can update to the latest patches.
KR, Rembrandt

Hi

carefull, if you have a horizontal axis (COMSOL is blocked verticaly in 2D axi) it will not change if you define the gravity along "r" it is RADIAL not along "-y" that is not the same thing !

For such cases I believe you MUST use a 3D simulation as the constant gravity field perpendicular to an rotation axis is destryoing the overall 2D-axi symmetry

--
Good luck
Ivar