Hi, bashu,

Sorry, it is my mistake.

It should say that the adjustable parameter of 'specularity coefficient' can be avoided, instead of 'the particle-wall restitution coefficient'. The latter is still needed in No-Slip BC condition.

There is a paper published recently by NETL/Fluent people related to wall BC. Although the paper is focused on Simonin turbulent model, they show the effect of the BC on the profiles.

By using ktgf, it is believed that the gt values affect the density profile. It maybe the egg-chicken story though.

A good expertiment and a good mesh independent model could help in systematically identifying the three parameters (p-p and p-w restitution coefficients and specularity coefficient) needed to match the experimental values or at least figure out their effect on the maesurable variables like density, pressure and velocity.

Hi,Bashu,

I agree with you. For the grid geration,do you have any idea or principle on producing a mesh-independent model? I mean, from the viewpoint of the distribution of grid points, how to generally assign the non-uniform grids?

More specifically, taking 2-D CFB as example, I have the follwoing questions:

(1)Should the grids be uniformly-distributed along axial direction or non-uniformly-distributed specially at the inlet accerelation zone? Generally, how to treat this matter?

(2)In some literature, some people assigned the uniform grids to the radial direction. But I am uncertain how to balance the grid requirement in near wall zone (e.g in case of using turbulent model)? Do you think it is necessary to give more grid consideration over the near-wall zone? How about the grid consideration on the central zone?

Hope to get some general ideas as guidance. Thanks.

Before replying to your questions let me just remind you that a lot of research is going on in this area and all of what I mentioned should be just considered as one idea in million ideas. With that in mind, read on:

I think since in our multiphase models at least the TIME AVERAGE gradients are not very sharp near the wall (or almost anywhere in the domain) we may not need a lot of cells there.

If you have a turn or a change of flow direction (core and annulus interface, catalyst pick up from a downflow pipe) you need more mesh density.

But in the instantaneous world, we have all these boundaries (read gradients, clusters, bubbles, ...) moving. So what can we do to capture correctly these moving gradients?

I think the core-annulus interface is difficult to identify upfront so maybe for a very accurate answer you need to use the adapt mesh feature. Is it realistic for unsteady multiphase? I don't think so, since probably you end up with the whole domain refined.

I think you don't need a lot of mesh near the wall and so a unifrom radial mesh should be good and may be more mesh in the acceleration zone is not a bad idea.

I hope some other experts in this field jump into the discussion and let us know their idea.

I'm getting a Floating pointer error on my solid-gas riser model? Please what do you think i'm doing wrong?

Femi,

Can you provide more detail about your case?

I'm modelling 2DDP gas-solid flow in a riser section with dimensions 1.2m X 0.17m. I've created the mesh and trying to analyse in Fluent. particle density - 7800Kg/m3, gas density - 1.2Kg/m3, gas velocity - 1.0m/s

When iterating i get intermittent commands that, "solution is converged" but the iteration still keeps going on until it gets to the number of time steps that I set for it. How do I define the residence time so it will stop on convergence?

Also, at the end of iteration it gives me an Error: cx-dialog-done: wta[1](widget) Error Object: *the-non-printing-object*

Please what do these imply?

For a transient model, the solution needs to be converged at each time step and that's what you are seeing.

But the gas-solids flows normally do not reach a steady state and you should continue the simulation until you are in a quasi-steady state. Then you turn on the time-averaging feature of fluent and let the simulation run for some time. The time average results are what you are after.

I am not sure about the error you are getting.

Thanks alot.

How do I then make my simulation to stop on convergence?

Please also can you tell me how I can view the my data at different time sets?

For example, my volume fraction at 0.1sec, 0.5sec and 1sec?

You need to use "file, auto save" and save your data files at a certain interval. This way you can go back to different files and view the results at the desired times.

There is no convergence criteria for multiphase cases (especially dense cases) and you need to decide when to stop the model.

Thanks alot, you confirm things for me pretty well.

Okay I'm trying to evaluate how my solid holds up will change over time at different heights and with variations in the mesh and riser geometry (diameter).

How do I obtain numeric values for my X-Y plots and contours?

Can I use the K-epsilon (2eqns) or Renolds stress (5eqns) in my model and how does this affect it?

I want to create tables and graphs to analyse my data. Do you know if I can do that on the Eulerian because for some reason, I dont have the PDF table on Display. And if I was required to generate and read one how do I do that?