Transient Multiphase Simulation and Post Processing

[kevin.kiprop]

I am running a multiphase transient simulation where slurry (mixture of water, cement, and tailings) is being discharged in a tunnel through an eight-inch diameter pipe (see the figure below). The geometry represents cut-and-fill mining panel with dimensions of approximately 60 ft (18.3 m) in length, 7.5 ft (2.3 m) in width, and 15 ft (4.6 m) in height. The model incorporates an 8-inch-diameter delivery pipeline installed along the roof with the discharge point 20 ft from the face, along with a 4-inch breather pipe near the roof.

The simulation is complete. I am in post process stage, and my goal is to see fill profile, presented in colour codes through the legend. To do this, I created a legend such that red indicates the cell is filled (100-10.1), Yellow is intermediate colour (10-0.1) and finally blue (anything between 0.1-0.0) is unfilled cells (presence of air) as seen in image 1.

I then created a second legend with the same aim, but this time around I assigned red colour a value between (100-50), yellow an intermediary of (49.9-10.1) and blue between (10 and 0.0). I am not able to view the actual expected filled cells, the bottom part is entirely yellow (see image 2). I expected to have the red colour on the bottom which shows the cells are filled, and subsequently yellow as an intermediary and blue at the top part. Please help me get a clear understanding of how to deal with the legend and colour coding so that I get the needed view.

[Rob]

In Fluent plot a contour on the z-centre plane of phase fraction (for each phase) and post those. What multiphase model did you use, was it granular and what is the packing limit? 

[kevin.kiprop]

The paste phase was modeled as a homogeneous non-Newtonian continuum rather than as a granular phase. Therefore, granular kinetic-theory parameters and a solids packing limit were not specified. The multiphase formulation consisted of air and paste phases tracked using the VOF model

[Rob]

OK. Put a phase volume fraction contour down the z-centre plane in Fluent, leave the scale as 0-1 and turn off node values. With two VOF phases it should be red/blue in the second image so I wonder if the solution has gone wrong. 

Also check the viscosity, which nonNewtonian model did you use?

[kevin.kiprop]

Transient Multiphase Simulation and Post-Processing: Visualization
Goal: To visualize simulated results during the post-processing stage

Non-Newtonian Fluid:Cemented Backfill paste

I am conducting a multiphase transient simulation in which slurry (comprising water, cement, and tailings) is discharged into a tunnel through an eight-inch-diameter pipe (see Figure 1).
The tunnel is roughly 60 feet (18.3 m) long, 7.5 feet (2.3 m) wide, and 15 feet (4.6 m) high. The pipeline is mounted along the ceiling, with the discharge point located 20 feet from the tunnel's end. A 4-inch-diameter breather pipe, also along the ceiling, allows air to escape as the tunnel fills with slurry. (Figure 4 shows material and boundary conditions set during simulation)
The simulation is now complete, and I am in the post-processing phase, aiming to visualize the fill profile. I created a volume and a corresponding legend for the isovolume: red shows filled cells (100- 10.1), yellow indicates intermediate levels (10- 0.1), and blue (0.1- 0.0) represents unfilled cells (air presence), as shown in Figure 1.

Figure 1: Red >100–10.1, yellow >10–0.1, blue > 0.1–0.0. The picture looks closer to what I want (filled at the bottom, air at the top).

Summary in post-processing
➢Created an Isovolume (Insert → Location → Volume → Method = Isovolume).
➢Colored the volume and edited the legend/colormap to match the description above.
I then created a second legend for the same purpose, but this time I assigned red a value between 100 and 50, yellow an intermediate range of 49.9 to 10.1, and blue from 10 down to
0.0. However, I can't see the actual filled cells I expected. The bottom section is entirely yellow (see Figure 2). I expected red at the bottom to indicate filled cells, followed by yellow as an
intermediary, and blue at the top. Please help me understand how to properly set up the legend and color coding so I can achieve the desired view.

Figure 2: Red = 100–50, yellow = 49.9–10.1, blue = 10–0.0. The bottom is entirely yellow. I do not see a red-filled zone except at paste_inlet.

Figure 3: Geometry settings selected
To help me get this visualization right,
➢What is the correct way to set a 3-color user map for this fill profile?
➢For VOF, should the color map use a 0–1 scale instead?

Materials and boundary conditions
The Herschel-Bulkley parameters used in the baseline CPB simulations were theoretical values selected from the literature reviewed for this study. The final Fluent implementation used a yield stress τy = 75 Pa, consistency index K = 8 Pa·sⁿ, flow index n = 0.5, and a critical shear rate of 0.001 s⁻¹. Because the Herschel-Bulkley formulation was regularized with the critical shear rate, separate minimum and maximum viscosity limits were not independently prescribed.

Figure 4: Materials and boundary conditions for the simulation

[Rob]

Please plot the contour I asked for. With two phases I'd expect a free surface at VOF=0.5  If you're using 3 phases look in Fluent for a phase specific colouring option, but I'd still use contours of each phase volume fraction. 

Volume fraction is 0 to 1.0  but with VOF you should see a distinct change from one phase to the other. If you have too much volume that's not one or other phase you've got a mesh or convergence problem. 

Check what viscosity Fluent is using: HB is fine when it's flowing, but looking at the graph when it stops it's near enough solid. That won't converge well as moving fluid is trying to displace "solid": Fluent assumes the fluid can flow, and doesn't handle solid body flow (eg toothpaste coming out of a tube) as well as CFX or Polyflow. 

[kevin.kiprop]

This is the contour I get when I plotted. Does this raise a convergence/solution problem since I do not see the two distinct colors? 

[Rob]

Yes. With VOF I'd expect most of the domain to be blue or red, the fact most of your model is green isn't good. 

Given the HB properties I suspect you didn't converge the time steps overly well, try running with a limiter at around 0.1 kg/m.s and see how it behaves. Once viscosity gets too high the fluid doesn't want to move: momentum has to go somewhere and can't. 

[kevin.kiprop]

Well, I changed timestep and made some changes from implicit to explicit sharp body. I achieved the above result.

[Rob]

With two phases it should be one or the other phase. If you leave the scale alone what are you seeing? Ie don't try and hide the mid-range values. 

[kevin.kiprop]

This is what I am able to see when I do not hide mid range values(I also turned on the contour numbers to see how it changes as filling progresses) I think I am now in the correct trajectory. One more thing I would like to do upon filling to the height I desire is to export the free surface so I can use it as the reference point surface for the next set of simulation. By doing this, I want to have the rugged nature of the fill as a true represantation on my surface for the next fill.(Will you please guide me on the best export mechanism? I will recreate the exported surface in Solid Works then run the next set of simulation) Thanks

[Rob]

Fluent uses a linear scale by default. Hence me expecting red-green-blue. I'd also trouble shoot in Fluent as CFD Post is node based so will slightly alter your result: with sufficient mesh you won't see any effect. 

For a relatively smooth surface (in Fluent) plot an iso-surface of VOF.    0.5 is traditional, but look at what's best for your model. You can export that iso-surface either via File>Export>Solution Data or by displaying the surface and File>Save Picture to VRML; that's a coloured stl for want of a better description. 

For a more faceted surface, create a register of VOF<0.5  (again check your model) and Separate cells.  You'll have two fluid zones but more importantly an interior between those zones. That interior follows the cells so cannot be smoothed. Then export that surface as above, or read the whole lot into Fluent Meshing to be left with just that surface after some work.