[sadeghiroholah]

Dear Colleague

With greetings, we are writing this letter of request as we inquire help in the following simulation which we have encountered in our research about solvent extraction settler.
A 3-D simulation of industrial gravity settler with 11m×11m×0.87m dimensions has been carried out. k-ε model for turbulence and MUSIG model for droplet size distributions were considered. For MUSIG model minimum size of droplet was 100µm and maximum size was 2000µm. Experimental data for droplet size distribution were used for inlet by a mean diameter of 180µm. In this study, number of size groups were (5, 10, 15) and turbulence coalescence coefficient were (0.5, 0.3, 0.1, 0.01, 0.001). Bouncy coalescence coefficient were (0.5, 0.3, 0.1,0.01,0.001) and break up coefficient were (1, 0.5,0.3) for various simulation runs. In this simulation continuous phase density was 800 12kg/m3' type="#_x0000_t75">, poly dispersed phase density was 1000 12kg/m3' type="#_x0000_t75"> and volume fraction of continuous phase was 0.53. This simulation was carried out with various numbers of tetrahedral elements (300000, 400000, 500000, 600000). Upwind and high resolution schemes were used for various runs in simulations. Fluid time scale was controlled by physical time scaling. At the beginning, time scale was 1 sec. and gradually decreased to 0.1 sec. Mass flow rate for inlet, average static pressure and mass flow rate for outlet as well as free slip wall for free surface were used as boundary conditions.
Generally in all runs the following problems exist:
1- Maximum droplet size is reached after convergence, and there is this droplet size in the whole of calculation domain and phase separation is not normal.
2- Computation time is long (2-3 days) and convergence rate is slow

Your consideration is highly appreciated. I look forward to receive your positive response.

[sadeghiroholah]

Originally Posted by sadeghiroholah

Dear Colleague

With greetings, we are writing this letter of request as we inquire help in the following simulation which we have encountered in our research about solvent extraction settler.
A 3-D simulation of industrial gravity settler with 11m×11m×0.87m dimensions has been carried out. k-ε model for turbulence and MUSIG model for droplet size distributions were considered. For MUSIG model minimum size of droplet was 100µm and maximum size was 2000µm. Experimental data for droplet size distribution were used for inlet by a mean diameter of 180µm. In this study, number of size groups were (5, 10, 15) and turbulence coalescence coefficient were (0.5, 0.3, 0.1, 0.01, 0.001). Bouncy coalescence coefficient were (0.5, 0.3, 0.1,0.01,0.001) and break up coefficient were (1, 0.5,0.3) for various simulation runs. In this simulation continuous phase density was 800 12kg/m3' type="#_x0000_t75">, poly dispersed phase density was 1000 12kg/m3' type="#_x0000_t75"> and volume fraction of continuous phase was 0.53. This simulation was carried out with various numbers of tetrahedral elements (300000, 400000, 500000, 600000). Upwind and high resolution schemes were used for various runs in simulations. Fluid time scale was controlled by physical time scaling. At the beginning, time scale was 1 sec. and gradually decreased to 0.1 sec. Mass flow rate for inlet, average static pressure and mass flow rate for outlet as well as free slip wall for free surface were used as boundary conditions.
Generally in all runs the following problems exist:
1- Maximum droplet size is reached after convergence, and there is this droplet size in the whole of calculation domain and phase separation is not normal.
2- Computation time is long (2-3 days) and convergence rate is slow

Your consideration is highly appreciated. I look forward to receive your positive response.

thank you for your attention

[ghorrocks]

Hi,

This is quite a complicated simulation and there is no way we can give detailed help on the forum. However, the following general comments apply (I am going to have to write a FAQ on this, I can't count the number of times I have said this....)

1) Have you established you have adequately converged?
2) Have you established your timestep is accurate (ie small) enough?
3) Have you established your mesh is fine enough? The range of meshes you have checked sounds inadequate. Have a read of this for an outline (or even better read Roache's text on CFD accuracy) http://journaltool.asme.org/Template...umAccuracy.pdf
4) With multiphase and MUSIG simulations you have a heap of extra models to validate. Have you done a systematic validation of all the key parameters in these models?

Glenn Horrocks