[tehache]

Dear Foamers,


I am trying to simulate a process where several gases are mixing at a junction of several tubes using rhoReactingBuoyantFoam. Reactions are switched off. Temperatures are varying, prescribed on the outer walls. I thought I was all fine and set, when on a closer look I realized I have a region where temperature has local maxima. After a lot of experimenting with a lot of parameters and meshes I am pretty sure it is not a mesh, convergence criteria, or discretization scheme issue.

I have then tried thermos based on sensibleEnthalpy and on sensibleInternalEnergy, the result was that I got temperature maxima for both, but with different distributions.
Pressure is close to 1 atmosphere, temperature around 1300K, those maxima are some 20K above their surroundings (not limited to a cell, or few cells, its extended).
Mach number is small, around 0.003.


I do not have much experience with thermodynamics and gases and was, probably naively, expecting the same results from both approaches.


So my actual questions are:


1.) Those inner temperature maxima stay there even averaging over time, in a region where changes in kinetic energy are not big enough to account for this. Am I correct assuming that this should thermodynamically be impossible?


2.) Why can the use of sensibleEnthaly or sensibleInternalEnergy give different results,
and/or which of these formulations should be used in what cases?


thx a lot for any hints!

[Gerry Kan]

Dear Thomas:

If your flow does not involve large pressure change (from the sound of your set up so far it should not) then using sensible internal energy and sensible enthalpy should not make a big difference.

However, if you are going to activate chemical reactions, and these reactions will alter the thermaldynamic state of the flow significantly, then you should stick with enthalpy.

The description of your problem sounds numerical in nature. Two things you could look at (if you have not done so already):

1) check the thermodynamic properties of your mixture components. If you are using polynomial representation, they should be consistent.

2) check the convergence history of your hotspot as well as other key locations. I have a feeling it is jumping around (I.e., failed to converge).

Hope that helps, Gerry.

[tehache]

Thank you, Gerry, for taking the time for a reply!


But alas, I did already check my polynomials several times for correctness and consistency.
Concerning convergence: I am doing time dependent calculations, tried both Pimple and Piso, and did experiments with extremely reduced convergence criteria, bringing the time step continuity error down to below 1e-14, to no avail.


I am now testing and debugging using a simplified test case, where for now I have seen hot and cold regions in some cases, too, in others, not. I will post it if I manage to find out what is going on there.