I have performed a simple 2D simulation of a fluidized bed with heat transfer. I have initialized the solid and air with their respective volume fractions and temperatures as given in the file attached.
As initialized, at t=0s, the air temperature throughout the bed is 300K. However, at 0.01s, the half bed which was filled with solids at 350K and air at 300K, now has air at 349.7K and the rest of the upper bed at 300K.
How is this possible for air to attain ~350K from 300K in 0.01s? Why is this happening?
Appears to me that at initialization itself, the air and solids are being initialized at the same temperature.
Would appreciate some guidance here from MFix team.
It does not seem to be an initialization issue, when you look at a higher resolution time scale.
I set a constant DT=1e-6 seconds and set up monitors at that same time resolution. I set up monitors for min and max solid and fluid temperature for the bed region.
The gas in the bed region is rapidly heating up from 300K to 350K while the solids phase stays at a fixed 350K. Interesting. There should be some 300K air entering at the bottom but that does not seem to be cooling things off either.
I’m not completely sure what’s going on here but I suggest experimenting with the thermal conductivity model and parameters.
Which is unphysical right? I see in the MFix documentation that there is a Nusselt number based correlation to determine the interphase heat transfer. Also, how exactly do you play around with the models? I assume these are hard coded into MFix solver?
The only work around I see here is too keep everything at the same initial temperature and let the gas flow into the domain at either a higher/lower temperature and study the transient response.
Hey I just did a simple hand calc with the Nusselt number correlation in MFix. It yielded an interphase htc of 25000 W/m^2K when the particle Re is zero at t=0s. No wonder the air is heating up so fast lol.
This correlation is based on Gunn et al and several other experimental investigations and has its limitations. ANSYS does not use this correlation. But nevertheless, I strongly feel MFix team look into this problem in the forthcoming versions of MFix!
You can choose from several different thermal conductivity models, and you can adjust tbe parameters. See the Fluid and Solids materials setup panes. That’s what I meant by experimenting.
What is the correlation that Ansys uses, and what htc do you get with it?
The other question is whether it is physically possible to start with a packed bed with particles and gas (between these packed particles) having a 50 degree difference. My guess is the temperature difference should quickly evolve to an equilibrium, and due to the several order of magnitude difference in thermal inertia between gas and solids, the gas will heat up to nearly the solids temperature and the solids temperature will barely drop.
It is possible that the Gunn correlation is not suitable here, but it is pretty standard for fluidized beds.
Why isn’t it possible? I heat particles upto 350K, pour them into a FB having air at 300K and start the gas in-flow at 300K. If air with a thermal conductivity of 0.025W/mK can reach 350K from 300K in 0.01s, then we have solved half the problems of heat transfer on earth.
ANSYS does indeed derive the htc based on Nu only. I misspoke there. But there are several other models such as Hughmark, two-resistance, etc in addition to Gunn model and these models seem to give a reasonable htc of about 50-100 W/m^2K. MFix does not have any option for that in that sense right? Unless I were to put a subroutine for the same…
Ohh okay. That is interesting. Thanks!
But can I write a subroutine to put these models into MFix to compare and contrast their predictive capabilities in TFM?
Yes, you have the source code, you could implement this either as a UDF or a modification to the main code. Please let us know if you find anything interesting!
