Simulation of liquid flow through a packed bed of non-spherical particles

[youssef.badran]

Hello,

Is it possible to couple Rocky and Fluent to simulate liquid flow through a packed bed of non-spherical particles? If so, how is the porosity distribution obtained in this case?

[Rob]

Yes and no. The packing can be done in Rocky by modelling the particles being pored into the bed (assuming particle count is sensible relative to the hardware). A coupled solution will pass a porosity over to Fluent. However, we assume the cells in Fluent are larger than the particles so you won't see any tortuosity effects; semi-resolved isn't a good idea in this scenario. You can export the particle surface(s) as stl to then remesh in Fluent. 

To an extent the best approach depends on what you're trying to model. 

[youssef.badran]

Thank you very much for the information.

We are trying to model a liquid dripping onto a heap of polyhedral particles with a given particle size distribution. Different regions of the heap can exhibit varying porosity.

Could you please elaborate on how to perform the remeshing for our case, based on your advice: "You can export the particle surface(s) as STL to then remesh in Fluent"?

[Rob]

Are you after bulk flow or more localised results? For the latter you'd run a filling model in Rocky and then export the surface to mesh in Fluent Meshing to fully resolve the gaps. Check the Rocky documentation for that: it's surface extract or similar. 

[youssef.badran]

Thank you very much for the information. I would try that.

Is it possible to use continuum modeling and Darcy’s law (require the porosity distribution in the heap) ?

[youssef.badran]

Can we get an accurate porosity distribution inside the heap given that different regions of the heap exhibit varying porosity and then use this porosity distribution in an Eulerian-Eulerian simulation using Darcy’s law in Fluent ? 

[Rob]

The surface tool is Export Particle Bed,  Ansys Help > Rocky > Rocky Scripts Manual > section 6.6 

You will be passing the porosity back as an accurate value, but as your particles are then assumed to be smaller than a cell is that what you're after? In that scenario we're back to whether we need to bother with nonspherical particles etc.   For a packed bed reactor, is the goal to evenly distribute the liquid on the top of the bed, or try and model the whole bed? Why do the Chemical Engineering undergrad courses look at "thick" packing for lower diameter towers rather than very wide diameter towers with thinner beds that would have a much lower pressure loss? 

[youssef.badran]

Thank you very much for the information. I will try that.

Yes, we are using non-spherical particles.

What method is used to compute the porosity? Is it the particle centroid method?

If we can obtain the spatial distribution of porosity in the heap, we could use it to initialize an Eulerian–Eulerian simulation in Fluent. In this simulation, the particles would be assumed fixed and the liquid would pass through the heap. Then, perhaps, we could include the mineralogy of the particles (mass fraction of components) and reactions.

Would that be possible ?

 

 

[Rob]

Check the Rocky-Fluent coupling manual for any and all maths that we're explaining for how porosity is calculated and passed to Fluent. 

Now you see why I'm trying to understand whether you're after bulk effects or specifics of the particle packing relative to the liquid flow. If you use a porous approach, you don't need to consider the solids as the porous model mimics the void fraction (Fluent User's Guide). If you want to see trickle and channelling effects you need the solids modelled and then move to a separate Fluent model with DPM/VOF/wall film depending on project needs: that's going to be an expensive model as you need to mesh the gaps. 

I can give some guidance on the models, but not tell you how to run your overall project. I can also ask some of the more difficult questions to make you think about what you're trying to do. Staff input on here is covered by Export Law, hence we operate under restrictions. For info, my degree was Chemical Engineering (a long time ago) and I've spent a fair amount of time modelling packed & fluidised beds plus a range of process kit since then. 

[youssef.badran]

Thank you very much for the information.