[Zahoor]

Hi Everyone,

i am trying to simulate solid particles in a conveyor using ANSYS CFX. it's a transient simulation.
i am using in-homogeneous multiphase model.
defined two materials like air as continuous fluid and another material as dispersed solid.
the particle model as interface drag model in fluid pair model TAB.interface momentum transfer Schiller Naumann model is used.
in fluid specific model TAB density difference for air while KINETIC THEORY OF GRANULAR FLOW is activated for dispersed solid.
Three domain created first named as hopper which contain hopper 2nd named as fluid its the fluid domain in which screw will rotate its a pipe like geometry and 3rd named as screw and it is an immersed solid domain.
time step used is 1e-4. RMS Co # less than 1 while MAX RMS goes upto 10.
now after running for almost few time steps it show an error floating point exception: overflow.
i have asked this question in another thread earlier but someone refer me to FAQ. i studied all those things. it says about checking some things. i have checked all those but unfortunately could not locate the problem. what should i do now.
Thanks.

[ghorrocks]

Please post your CCL and an image of your geometry and a example of your mesh.

[Zahoor]

[Zahoor]

The out file is attached.

[ghorrocks]

This model appears to have immersed solids, moving mesh and three phases including a free surface. This is a complex model so you should expect some difficulties in getting it started.

You will have to get it working bit by bit. I recommend you first run single phase immersed solid to make sure that works. And then run the multiphase model by itself as well. Only when those models work by themselves should you consider combining them.

Two question:
* Have you check immersed solids is compatible with the other physics you are using?
* Why are you using a moving mesh model? What mesh is moving?

[Zahoor]

Thanks for quick reply.
i have done a tutorial of immersed solid with two phases, air and water. tutorial is given on https://youtu.be/qJyJhz4bEQI.
after that i repeat the same tutorial on this geometry it works.
you said it has three phases. i didn't get this point i defined only two fluids air as continuous and another as dispersed solid, although i have generated three domain as Hopper, Fluid & Screw. where hopper and fluid are fluid domains.
the screw is defined as immersed solid it has to rotate in the fluid domain. i have to check the flow pattern only. i don't have to study heat transfer.

[ghorrocks]

Yes, you are correct. You have solid particles in air. There is no liquid phase.

Please note that the particles are likely to pack into a bed against the screw. CFX has no model for this. So if the particles end up as a packed bed you will not be able to model this with CFX - at least without adding something like a DEM model. You have defined a maximum packing fraction but I suspect this will not be sufficient to model a packed bed.

Why do you have a moving mesh model in this? All the boundaries I see are defined as being stationary so nothing will move (I might have missed something though).

[Zahoor]

immersed solid i.e. Screw is rotating at about 30 rpm in the fluid domain. simulation runs for a while then the error came up. same setup with coarse mesh and 0.01 sec timestep proceeds. but as i refined the mesh error came up.

max packing is in the hopper where powder is stored. it falls in the conveyor under gravity. so i used opening type inlet and outlet. if pressure outlet is used it says that a wall has been placed at 70 % area.

as it falls from hopper to the conveyor do you think at that time packing will be max?

you said about using DEM model with this. i don't know how to do that. if you can guide me about that.

[ghorrocks]

I suspect the immersed solid model will place a restriction on time step size. It would not surprise me if you have to scale the time step down with smaller mesh. For instance having the immersed solid move more than about half an element per time step sounds like a bad idea to me.

What I am saying about max packing is that it is not a good model for particles packing to form a bed.

A DEM model is discrete element model. This models each particle as a discrete particle and has collision models, proper particle to particle and particle to wall interaction and similar physics. If you want particles to form beds and be pushed around or scoured you should consider this. There are several DEM software around like EDEM and Rocky, and I understand they can both be coupled to CFX. But note they are commercial software which will require an additional license.

[Zahoor]

i don't want to form a bed. particle will move along the screw. correct me if i am wrong, so as they move i don't think a bed of max packing will be formed.

2nd what if i reduce the max packing? will it work?

i don't have access to those commercial software. so i have only ANSYS.

[ghorrocks]

If the particles are pushed along by the screw as a large mass of contacting solid particles then you definitely need a DEM model. CFX cannot handle this type of model.

[Zahoor]

it's my final year project. and i have only ANSYS. any recommendation what should i do now?

[ghorrocks]

Fluent has a very simple particle interaction model which might be good enough. Have a look at it.

Otherwise discuss this with your supervisor. It appears you have been given a project which is not possible to complete with the tools you have available.

[hcs129]

Originally Posted by ghorrocks

I suspect the immersed solid model will place a restriction on time step size. It would not surprise me if you have to scale the time step down with smaller mesh. For instance having the immersed solid move more than about half an element per time step sounds like a bad idea to me.

What I am saying about max packing is that it is not a good model for particles packing to form a bed.

A DEM model is discrete element model. This models each particle as a discrete particle and has collision models, proper particle to particle and particle to wall interaction and similar physics. If you want particles to form beds and be pushed around or scoured you should consider this. There are several DEM software around like EDEM and Rocky, and I understand they can both be coupled to CFX. But note they are commercial software which will require an additional license.

I can't find tutorial about how to couple CFX with DEM softwares. Do you have any suggestion?

[ghorrocks]

You should ask the DEM software about that.