Learning Forum

[sjohn]

I have a model wherein I establish steady state condition by flowing a two-species mixture and then introduced discrete particles in transient simulation. After steady state simulation, I get concentration of both species as contour plot. Upon starting the transient simulation with very small  time steps such as 1e-07, I find that the concentration of species in the domain drops to zero except at the inlet. The inlet velocity is about 25 m/s. the length of the domain is roughly 60 mm. Is this normal? I expect the concentration of the species not to be zero after I switch from steady to transient flow. I specify the inlet flowrate, total pressure, static pressure, and total temperature at the inlet.

[Essence]

Hello,

Please check your mesh and the CFL number. The time step size should be lesser than cell pass time of the flow.

[Rob]

Did you do anything other than switch to transient in the General panel? You'd not be the first person to initialise too, which throws away the earlier solution. 

[sjohn]

 

Rob- After running the steady state model, I moved copies of the case and data files (Set 1) in a new folder. Then I added the DPM settings and saved the case and data files as new files (Set 2). When I am running the simulation, I saved the files with a new name at different time intervals – 0.1 ms (Set 3), 0.2 ms (Set 4), etc. I can plot species concentration in the domain from all three four sets of case and data files.
Set 1 – Species concentration is correct
Set 2 – Species concentration is correct
Set 3 – Species concentration is nearly zero, except near the inlet where the species enter.
Set 4 – Species concentration is nearly zero, except near the inlet where the species enter.

Hope this explanation is helpful.

 

[Rob]

Thanks. That should be fine, and also means you've not wiped data by accident. What happens if you take Set 2, adjust the set up and run 1-2 time steps? What are the particles doing, other than moving?

[sjohn]

After running Set 2 for one time step (1e-07 s), the simulation met the convergence criteria within the specified 30 iterations and the contour plots show correct species concentration.
I ran for another time step and the contour plots show correct species concentration and convergence was achieved.

Note that Set 3 was run for 1243 time steps and Set 4 was run for 1370 time steps. 

[sjohn]

In DPM, I am using a pressure swirl atomizer model, wherein I specity a mass flowrate and upstream pressure (205.3 psi). 

For the species flow, I specify a mass flowrate and supersonic/initial gauge pressure of 205.3 psi.

[sjohn]

I checked the temperature for all sets (Set 1-4) and it seems as expected. This indicates that not all variables were reset. Only the species concentrations were somehow zero (except near inlet) for Set 3 and Set 4.

[Rob]

That shouldn't be possible. Can you post some images? 

[sjohn]

 

Here is a snapshot of the species concentration – Set 2, Set 4, and Set N, which was obtained 3263 time steps of varying step size. 

Here is a snapshot of the temperature – Set 2, Set 4, and Set N. In Set 2, you can see the effect of the atomizer as a lower temperature plume. In Set N, the high tmeperature is due to the heat released from the reaction between the species (released from droplets upon evaporation) and the one of the species from the fluid flow starting at the inlet.

 

[Rob]

Did anything change in terms of mixture templates when you added the evaporating droplets? Ie it's not a case of the species got scrambled because you've changed something? 

[sjohn]

I thought I had not changed anything major but upon reviewing the log file, I discovered that the case was hybrid initialized (accidentally). I think that may have caused the species concentration to zero out. I will have to rerun the simulation.Thank you for asking the right questions.

[Rob]

You're welcome. I've benefited from watching a lot of people make mistakes over the years, and coupled with a few of my own it helps when trying to solve problems blindfolded. Commercial support (my main role) isn't any easier but at least I can look at models. 

[sjohn]

I do have question about running the model in steady state vs. transient. I am currently using small steps - 1e-06 to 1e-07 in the transient mode with URFs for product species reduced to 0.7, energy reduced to 0.9, discrete phase reduced to 0.5, and P-1 (radiation) reduced to 0.9. For such a case, would steady state run faster than transient or is that hard to tell? Since the case involves spray atomization, evaporation, and reaction, I am thinking small time steps in a pseudo transient mode would be the steady state approach. I suspect it would also have similar speed as the transient mode and hence no benefit in speed. Atleast, the transient mode would provide some intermediate values.

[Rob]

There's also steady flow and transient particles to really mess with your mind.  Difficult to say without a lot more detail, and that's likely to exceed what I can then answer....