- **FLUENT**
(*http://www.cfd-online.com/Forums/fluent/*)

- - **Modelling Sand erosion using Mixture model and DPM**
(*http://www.cfd-online.com/Forums/fluent/103812-modelling-sand-erosion-using-mixture-model-dpm.html*)

Modelling Sand erosion using Mixture model and DPMHello, I am currently modeling a relatively tricky problem concerning the effects of sand erosion in the desert upon the surfaces of a high speed train.
I would like to model the effects of an atmospheric boundary layer physically lifting the sand into the air flow and modeling the erosion effects on the train surfaces. So far I have read all Fluent tutorials available on this subject and none seem to be relevant. I have also read many papers on the subject and some have given me ideas but none are specifically relevant. The model I have so far is a very basic 2D mesh where I have modeled an atmospheric boundary layer using logarithmic profiles for velocity and turbulence. I have then through reading much literature found that using the mixture model and modeling the slip velocity for a granular flow seems to be the most physical approach. I have solved this model with good results however I obviously only achieve the free surface representation of the sand /air phases. What I would like to do from this stage is to model the particle tracking and erosion, I need to somehow use the DPM model to calculate the sand / air mixture from this stage and hence the erosion. I guess my ultimate question is can a free surface Volume fraction be turned into particles and therefore tracked using DPM? Possible a UDF to turn Volume fractions to particles? Any help would be greatly apreciated. Many Regards Karl |

Hi Karl,
If i understand the mixture model correctly, it is a hybrid between the eulerian model and Lagrangian model. As far as I know, the individual particle trajectories can not be taken from the solids volume fraction. The nature of the Eulerian method is to average over the solids phase, therefore you can't get the individual particle properties back. If you are talking about eroded sand affecting a train, I would expect individual particle properties wouldn't matter very much. You would gain much efficiency switching to the Eulerian method. You can use the solids volume fraction, solids velocity and mass flux (solids volume fraction * solids velocity) and look at the momentum transfer from the sand phase to the surface of the train using the eulerian method. |

Thanks very much for your reply. I am currently running the simulation switching to Eulerian - Granular flow which I belive is the most relevant approach. However at present I am not seeing much change in interphase shape.
Another idea I have is to use the Dense decrete phase model, however I am not entirely sure if I need to use an injection of particles or with a patch of the phase it can calculate particle trajectories from there. The reason I am thinking of using this model is that this model will caculate particle saltation and suspension. It can also calculate the erosion effects of the 2nd phase on a surface. A third possibility I have been thinking of is to run a mixure/granular or eulerian/granular flow calculate the interface and hence height of sand mixing from the sand bed to the air. From here extracting a number of profiles of phases mixture. I was then thinking of converting this to a concentration of phase using a UDF and injecting this using the DPM model. The UDF would change the injection concentration in accordance to the profiles I have extracting and hopefully simulate a periodic sand /air mixture. From here it is easy to calculate erosion rate etc. Please if you have time could comment on these ideas I have. Thank you very much for you help on this subject. Regards Karl |

Quote:
trust Madrid is suiting you well. You need to use the DPM for this case. In CFX if your flow is steady state and only one-way coupling is assumed between gas phase and the particles then you can run a single iteration using your converged single phase case results as the initial solution and the particle tracks will be calculated almost like a post-processing operation. Regards Jon |

All times are GMT -4. The time now is 13:11. |