Learning Forum

[Abdelmadjid Belmokhtar]

Momentum Residuals do not get under e-3 in a simulation of a cylone air filter with tangential inlet with sst turbulence model. What can be the problem? I tried different mesh nodes number and the results are the same.

[Rob]

What do the monitor points tell you? Note POINT not surface. 

[Abdelmadjid Belmokhtar]

Thank you Rob for your reply. I forget to mention that I am runing a steady state simulation, and for the monitor points I made one point to monitor the velocity components in terms of accumulated time step and it shows somthing like a sinusoidal function for the three components of velocity

[Rob]

OK, and if the point monitor(s) has reached a (vaguely) sine curve what does that tell you?  Think about what the flow behaves in a cyclone, focus on the vortex core. 

And, the bit I missed. Why do you think a 2-equation turbulence model is suitable for a high swirl system? 

[Abdelmadjid Belmokhtar]

Thank you very much for your reply. I think that the sine curve can tell me that the flow cannot be supposed as steady. And about choosing the SST model, I based it on the results of a work which compares the simulation of a single-phase flow in a Cyclone Gasifier using different turbulence models with the experimental results. They found that the SST model can be a good choice. Of course, LES is the preferred choice to predict phenomena, but it takes a long time

[Rob]

Correct, it's termed pseudo transient in that the results appear to be unsteady even with the steady solver. Not to be confused with the pseudo transient option in Fluent's BPCS solver.... We did explain the likely confusion when we first saw the option. 

For high swirl have a look at RSM too. If it's a proper cyclone use RSM. If it's a cyclonic filter (ie tangential inlet but filter pack in the core) k-w may be OK. There's a paper somewhere by Cokjlat, Slack & others looking at poly mesh in cyclones: I can't remember the title and don't have a copy. 

[Abdelmadjid Belmokhtar]

Thank you very much Rob, I appreciate your guidance. So there is no option to run a study simulation in my case ? And about the paper, I think this is it https://www.academia.edu/download/41468819/Cokljatetal2006.pdf

[Rob]

Simulation is a tool. We always try and run steady, but sometimes can't. But, if the steady result is good enough even with less than ideal convergence we may use that: there's a big difference between academic and commercial usage. There's also a reason why simulation engineers are sought after by employers! 

So, I can't say you won't learn from a steady solution, nor can I say you must run transient. 

[Abdelmadjid Belmokhtar]

Thank you very much for your help mister Rob, Your tips and advice have helped me a lot.

[Rob]

You're welcome.  I've not checked the link as I'm not supposed to download anything to keep within the rules. If it's Davor C then it'll either be the right one or will reference the right paper. 

[Abdelmadjid Belmokhtar]

I understand that. Thank you again.