Learning Forum

[azhao]

I am simulating a spray cooling case (no heat flux added yet) with the DPM and Eulerian wall film models in ANSYS Fluent 2021R2. Their settings are as follows:

[Rob]

Check the injection, if you've got over 9 million parcels after two DPM updates I think something's gone wrong. Did you change any of the parcel settings?

[azhao]

Hi, Rob,
Thank you for the reply. The parcel release method is "Standard" as default. I will change the number of streams injected by the nozzle to a smaller number and then have a try.
An

[Rob]

Ah, I'd missed that. 500 is a little excessive. Try 25 and see how you go. With DPM if you have a very high parcel count in a cell the source terms can become too high, that then causes the problems you're seeing. Read the DPM theory for some background on this.

[azhao]

Hi Rob I followed your suggestion by changing the particle stream number to 25 and running the case again, it went a little bit farther than before but blew off after 31 iterations, the following is the error message, I can find several parts that are suspicious:
the part marked by !!!!! looks very strange but I assume it is only an issue related to parallel computation but not convergence?
The number of parcels is heavily reduced but still quite a lot, maybe I shall decrease the particle stream further?
The direct cause seems to be the excessive CFL number, but I am not sure what does this CFL number mean. I think the coupled solution algorithm itself is implicit, so maybe the problem is because the first order explicit way selected by Eulerian wall film model? I will change it and have a try......
The two turbulence closure equations seem to be another problem as the last two columns have very big residuals. and the warning "turbulent viscosity limited to viscosity ratio of 1.000000e+05 in 719163 cells", no idea if I should do something to improve this or it will disappear after the convergence problem is solved.
Thanks for checking this long thread.
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Advancing DPM injections...

Reversed flow on 11521 faces (62.6% area) of pressure-outlet 12.

Reversed flow on 10117 faces (56.1% area) of pressure-outlet 13.

Reversed flow on 7590 faces (99.7% area) of pressure-outlet 14.
temperature limited to 1.000000e+00 in 680 cells on zone 10 in domain 1
temperature limited to 5.000000e+03 in 5 cells on zone 10 in domain 1
Film time = 1.300018e-03 with timestep = 3.1e-09, (max_cfl: 1.805477e-01)

Eulerian wall film model is enabled:

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
The use of the DPM Domain has been requested, but it is incompatible with the above
settings, and therefore the DPM Domain will not be used.
To avoid this message, disable the 'Use DPM Domain' option.
You may want to use the 'Hybrid Optimization' in the partitioning options instead.
That is, enable it, run some DPM iterations, repartition, and use the new partitions.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

Injecting 25 particle parcels with mass 1e-07
number tracked = 475615, escaped = 20, incomplete = 471972, shed = 473897
Eulerian wall film particles:
absorbed = 328, splashed = 64, stripped = 74, rebound = 38
Injecting 25 particle parcels with mass 1e-07
number tracked = 477099, escaped = 105, incomplete = 472289, shed = 1479
Eulerian wall film particles:
absorbed = 759, splashed = 312, stripped = 36, rebound = 43
Injecting 25 particle parcels with mass 1e-07
number tracked = 477701, escaped = 59, incomplete = 472274, shed = 1267
Eulerian wall film particles:
absorbed = 418, splashed = 96, stripped = 36, rebound = 28
Injecting 25 particle parcels with mass 1e-07
number tracked = 478676, escaped = 50, incomplete = 471996, shed = 1223
Eulerian wall film particles:
absorbed = 647, splashed = 180, stripped = 36, rebound = 40
Injecting 25 particle parcels with mass 1e-07
number tracked = 479681, escaped = 68, incomplete = 472161, shed = 1376
Eulerian wall film particles:
absorbed = 654, splashed = 256, stripped = 36, rebound = 26
Injecting 25 particle parcels with mass 1e-07
number tracked = 480179, escaped = 54, incomplete = 472022, shed = 955
Eulerian wall film particles:
absorbed = 630, splashed = 176, stripped = 36, rebound = 29
Injecting 25 particle parcels with mass 1e-07
number tracked = 481179, escaped = 39, incomplete = 472086, shed = 1331
Eulerian wall film particles:
absorbed = 595, splashed = 284, stripped = 36, rebound = 23
Injecting 25 particle parcels with mass 1e-07
number tracked = 482245, escaped = 36, incomplete = 472225, shed = 1192
Eulerian wall film particles:
absorbed = 887, splashed = 412, stripped = 36, rebound = 69
Injecting 25 particle parcels with mass 1e-07
number tracked = 483065, escaped = 38, incomplete = 472233, shed = 1435
Eulerian wall film particles:
absorbed = 646, splashed = 180, stripped = 36, rebound = 44
Injecting 25 particle parcels with mass 1e-07
number tracked = 484070, escaped = 37, incomplete = 472416, shed = 1363
Eulerian wall film particles:
absorbed = 756, splashed = 256, stripped = 36, rebound = 32

Warning:3.2923% of the total discrete phase mass was not tracked for the expected residence time:
7.53e-05 s less on a mass-weighted average (which is 12.5713% of their total age or 7.5307% of the last time step).


turbulent viscosity limited to viscosity ratio of 1.000000e+05 in 223317 cells
303.8262e-041.4659e-031.1315e-033.5500e-033.9687e-038.1724e-015.7855e-01 168:57:16 19970

Reversed flow on 10765 faces (55.8% area) of pressure-outlet 12.

Reversed flow on 7856 faces (48.8% area) of pressure-outlet 13.

Reversed flow on 5505 faces (69.9% area) of pressure-outlet 14.
Stabilizing pressure coupled to enhance linear solver robustness.
Stabilizing pressure coupled using GMRES to enhance linear solver robustness.

Experiencing convergence difficulties - temporarily relaxing and trying again...
Stabilizing pressure coupled to enhance linear solver robustness.
Negative k in 1044 cells after linear solve.
temperature limited to 1.000000e+00 in 59751 cells on zone 10 in domain 1
temperature limited to 5.000000e+03 in 74 cells on zone 10 in domain 1
Film time = 1.300021e-03 with timestep = 3.1e-09, (max_cfl: 3.909367e-01)
Maximum CFL reached. Reduce time step by factor of 2

turbulent viscosity limited to viscosity ratio of 1.000000e+05 in 655349 cells
311.0179e-021.7508e-152.3064e-153.7564e-153.8050e-132.1808e+021.0341e+00 143:24:54 19969

Reversed flow on 10938 faces (56.5% area) of pressure-outlet 12.

Reversed flow on 8033 faces (48.1% area) of pressure-outlet 13.

Reversed flow on 6093 faces (78.6% area) of pressure-outlet 14.
Stabilizing pressure coupled to enhance linear solver robustness.
Stabilizing pressure coupled using GMRES to enhance linear solver robustness.

Experiencing convergence difficulties - temporarily relaxing and trying again...
Stabilizing pressure coupled to enhance linear solver robustness.
Stabilizing pressure coupled using GMRES to enhance linear solver robustness.

Experiencing convergence difficulties - temporarily relaxing and trying again...

Divergence detected in AMG solver: pressure coupledStabilizing k to enhance linear solver robustness.
Negative k in 202201 cells after linear solve.
All neighbors negative on 11 cells.
temperature limited to 1.000000e+00 in 333900 cells on zone 10 in domain 1
temperature limited to 5.000000e+03 in 532 cells on zone 10 in domain 1
Film time = 1.300023e-03 with timestep = 1.5e-09, (max_cfl: 1.108611e+00)
Maximum CFL reached. Reduce time step by factor of 2

turbulent viscosity limited to viscosity ratio of 1.000000e+05 in 719163 cells

Divergence detected in AMG solver: pressure coupled

[Rob]

Given the number of incomplete particles drop the particle time step by an order of magnitude. Also turn off splashing. See how that goes. Basically, to diagnose problems we turn stuff off until it works and then slowly turn things back on until they don't. From there it's usually fairly easy to work out what the root cause of the problem is.

[azhao]

Hi Rob,
Thanks for the great suggestions.
I decreased the particle time step from 0.0001s to 0.00002s. And the splashing is turned off. Also, the DPM domain is unticked. The error message now seems simpler than before, but the solution still blows off after 25 iterations as shown in the following:
It looks like the problem is in solving for the Eulerian wall film models by checking the huge residual of it. And the problem of turbulence quantities remains. The problem of a huge DPM parcel number is eliminated, I suppose it means most parcels are generated by splashing...
----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

Advancing DPM injections ....

Reversed flow on 956 faces (8.5% area) of pressure-outlet 12.

Reversed flow on 6391 faces (24.0% area) of pressure-outlet 13.

Reversed flow on 6832 faces (86.7% area) of pressure-outlet 14.
sub-iteration: 1 residual - h: 0.000000e+00; u: 0.000000e+00; v: 0.000000e+00; t: 0.000000e+00
Film time = 2.000000e-03 with timestep = 1.0e-04, (max_cfl: 0.000000e+00)
Injecting 25 particle parcels with mass 2e-08
number tracked = 275
Injecting 25 particle parcels with mass 2e-08
number tracked = 300
Injecting 25 particle parcels with mass 2e-08
number tracked = 325
Injecting 25 particle parcels with mass 2e-08
number tracked = 350
Injecting 25 particle parcels with mass 2e-08
number tracked = 375
Injecting 25 particle parcels with mass 2e-08
number tracked = 400
Eulerian wall film particles:
absorbed = 13
Injecting 25 particle parcels with mass 2e-08
number tracked = 412
Eulerian wall film particles:
absorbed = 24
Injecting 25 particle parcels with mass 2e-08
number tracked = 413
Eulerian wall film particles:
absorbed = 26
Injecting 25 particle parcels with mass 2e-08
number tracked = 412
Eulerian wall film particles:
absorbed = 27
Injecting 25 particle parcels with mass 2e-08
number tracked = 410
Eulerian wall film particles:
absorbed = 27

...............................................................................................................................................................................................................................
turbulent viscosity limited to viscosity ratio of 1.000000e+05 in 187343 cells
244.6249e-047.3816e-021.2243e-026.3270e-021.2860e+065.6663e-011.0639e+005:11:14 19976

Reversed flow on 3235 faces (33.1% area) of pressure-outlet 12.

Reversed flow on 8237 faces (57.1% area) of pressure-outlet 13.

Reversed flow on 5250 faces (67.6% area) of pressure-outlet 14.
temperature limited to 1.000000e+00 in 513405 cells on zone 10 in domain 1
temperature limited to 5.000000e+03 in 436656 cells on zone 10 in domain 1
sub-iteration: 1 residual - h: 1.976365e+12; u: 1.926994e+13; v: 8.212897e+12; t: 1.544663e+87
sub-iteration: 2 residual - h: 1.492813e+12; u: 6.872916e+12; v: 5.264992e+12; t: 5.808624e+87
sub-iteration: 3 residual - h: 1.102153e+12; u: 1.924995e+13; v: 8.212897e+12; t: 6.931355e+87
sub-iteration: 4 residual - h: 1.103431e+12; u: 6.872916e+12; v: 5.264992e+12; t: 5.808624e+87
sub-iteration: 5 residual - h: 1.103234e+12; u: 1.924995e+13; v: 8.212897e+12; t: 6.931355e+87
sub-iteration: 6 residual - h: 1.103234e+12; u: 6.872916e+12; v: 5.264992e+12; t: 5.808624e+87
sub-iteration: 7 residual - h: 1.103234e+12; u: 1.924995e+13; v: 8.212897e+12; t: 6.931355e+87
sub-iteration: 8 residual - h: 1.103234e+12; u: 6.872916e+12; v: 5.264992e+12; t: 5.808624e+87
sub-iteration: 9 residual - h: 1.103234e+12; u: 1.924995e+13; v: 8.212897e+12; t: 6.931355e+87
sub-iteration: 10 residual - h: 1.103234e+12; u: 6.872916e+12; v: 5.264992e+12; t: 5.808624e+87
Film time = 2.193750e-03 with timestep = 6.3e-06, (max_cfl: 9.051277e+05)
Maximum CFL reached. Reduce time step by factor of 2

turbulent viscosity limited to viscosity ratio of 1.000000e+05 in 257258 cells
254.9143e+008.5547e-039.2232e-042.5725e-036.8090e+594.7955e+071.5337e+005:09:00 19975

Reversed flow on 10219 faces (47.4% area) of pressure-outlet 12.

Reversed flow on 11966 faces (73.9% area) of pressure-outlet 13.

Reversed flow on 5496 faces (71.4% area) of pressure-outlet 14.
Stabilizing temperature to enhance linear solver robustness.
Stabilizing temperature using GMRES to enhance linear solver robustness.

Divergence detected in AMG solver: temperaturesub-iteration: 1 residual - h: 1.318278e+19; u: 6.973356e+17; v: 2.637542e+17; t: 6.959193e+88
sub-iteration: 2 residual - h: 1.320348e+19; u: 2.046648e+20; v: 2.037741e+20; t: 1.007706e+95
sub-iteration: 3 residual - h: 1.320801e+19; u: 6.973356e+17; v: 2.621847e+17; t: 1.212422e+101
sub-iteration: 4 residual - h: 1.320801e+19; u: 2.046648e+20; v: 2.037741e+20; t: 1.206597e+107
sub-iteration: 5 residual - h: 1.320801e+19; u: 6.973356e+17; v: 2.621847e+17; t: 9.390307e+112
sub-iteration: 6 residual - h: 1.320801e+19; u: 2.046648e+20; v: 2.037741e+20; t: 4.265750e+111
sub-iteration: 7 residual - h: 1.320801e+19; u: 6.973356e+17; v: 2.621847e+17; t: 1.152971e+109
sub-iteration: 8 residual - h: 1.320801e+19; u: 2.046648e+20; v: 2.037741e+20; t: 9.192200e+117
sub-iteration: 9 residual - h: 1.320801e+19; u: 6.973356e+17; v: 2.621847e+17; t: 5.623902e+123
sub-iteration: 10 residual - h: 1.320801e+19; u: 2.046648e+20; v: 2.037741e+20; t: 2.564501e+131
Film time = 2.196875e-03 with timestep = 3.1e-06, (max_cfl: 4.516040e+12)
Maximum CFL reached. Reduce time step by factor of 2

turbulent viscosity limited to viscosity ratio of 1.000000e+05 in 932236 cells

Divergence detected in AMG solver: temperature
Divergence detected in AMG solver: temperature

[Rob]

That's the EWF failing. It is quite sensitive to time step, and if the droplets are warm/hot it's even more so. Ideally the film needs to have a Courant Number around 0.05 as a starting point.

[azhao]

Hi Rob,
I decreased the Maximum Courant number from 0.2 to 0.05 but the results did not change, still blow off after 25 iterations and the (max_cfl: 4.516055e+12) is almost the same as the previous result.
And the droplets are actually set to 291 K, which is the same as the initial temperature of the Eulerian wall film.
Is there any other method to improve it? I will try a nonzero initial liquid film but not sure if it will help....

[Rob]

Turn off the EWF and see what happens. What is the mesh like on the wall with the film on it?

[azhao]

I also used the same setting for the Lagrangian wall film with EWF turned off. The solution is converged fine even after 50000 iterations.
The following figure shows the near wall mesh, the blue circled part is the region where there is a film on it.

[Rob]

What happens to the film at the edge? LWF uses boundary "particles" so has a different numerical scheme. I do know that EWF really won't like the jump in cell size.

[azhao]

There is no special treatment about the film at the edge. The surrounding boundaries near this blue-circled region are set as pressure outlet. Does it mean the film flows to this part would automatically disappear? That is the reason the solution diverges?

[azhao]

[Rob]

I had to ask a grown up - my brain's had it.
The film will disappear at the edge of the wall and should be reported as mass leaving the domain. If you plot the film thickness after some time but before the model fails how does it look?

[azhao]

Hi Rob Thank you so much for your patience with this problem. :-)
Please see the following contours for the film thickness distribution (The initial liquid film thickness in the blue circled region is 1 micrometer). It looks very unphysical...
And I did not find any direct report of mass leaving the domain in the ANSYS Fluent console, shall I turn on some verbose option to monitor how much mass is leaving the domain?

[azhao]

And this is the contour for the liquid film distribution two iterations before the blow off

[Rob]

That looks like you've hit the 5mm limiter (default max thickness) and also it looks very mesh dependent. If you look at the droplet size near the surface, how big are they?

[azhao]

The droplet size near the surface is around 0.133 mm, seemingly OK......
And the following figures shows the surface mesh of the blue circled region:
The minimum quality of mesh in ICEM CFD is 0.56, the minimum angle is larger than 36 degrees, I assume it was enough for this problem:

[Rob]

I'd be looking to use Workbench or Fluent Meshing for this, ICEM CFD is overkill.
Looking at the droplets, we have relatively few 0.133mm droplets hitting the surface. If you look at the cell size in that region, or rather the surface facet size how thick will the film be? You also seem to have very few droplets near the surface given the 9.5M in the system!

[azhao]

Hi Rob I agree with you that Tmesh is already enough for such simple geometry.
Please see the following summary of the injected spray particles, after decreasing the particle streams from 500 to 25, the total number of parcels or particles have been drastically decreased to 383 and 2497.5, respectively.

(*)- Summary for Injection:
(*)-injection-0
(*)
Total number of parcels: 383
Total number of particles: 2.497500e+03
Total mass: 3.064000e-07(kg)
Overall RR Spread Parameter: 3.704653e+00
Maximum Error in RR fit: 7.290067e-02
Overall RR diameter(D_RR): 8.322037e-05(m)
Maximum RMS distance from injector: 1.029918e-02(m)
Maximum particle diameter: 1.326034e-04(m)
Minimum particle diameter: 3.668175e-05(m)
Overall mean diameter(D_10): 5.492609e-05(m)
Overall mean surface area(D_20): 5.817943e-05(m)
Overall mean volume(D_30): 6.168634e-05(m)
Overall surface diameter(D_21): 6.162547e-05(m)
Overall volume diameter(D_31): 6.537237e-05(m)
Overall Sauter diameter(D_32): 6.934709e-05(m)
Overall De Brouckere diameter (D_43): 7.720868e-05(m)


Regarding the facet size where the 0.133mm droplets hit the surface, as the length of the long side for one cell is 0.667 mm, it is around 0.15 mm^2. Hence if a droplet parcel with 2497.5 / 383 = 6.5 particles and 0.0133 mm diameter hit it, then the resulting liquid film thickness should be 1/6*pi*(0.0133)^3*6.5/0.15 = 5.34e-5 mm, it is rather small. I have no idea why some area reaches the 0.5 mm threshold after just two DPM injections....

[Rob]

I'm not either, but it does look like a very mesh related problem. We've run similar cases for isothermal VOF<>EWF testing and not seen this issue. The film really doesn't like big jumps in facet size, but shouldn't go as wrong as you're seeing. There must be something else going that's not obvious from your posts.
Turn off spreading & surface tension, let's see what that gives.

[azhao]

I agree with you that there is something suspicious other than the mesh.
After turning off the spreading & surface tension, the maximum Courant number decreases, but the solution still blows off after 23 iterations.

turbulent viscosity limited to viscosity ratio of 1.000000e+05 in 466580 cells
22 3.9660e-04 1.5672e-03 2.6568e-03 1.5392e-03 8.3966e+48 1.5275e+12 1.0119e+00 14:09:42 9978

Reversed flow on 7308 faces (14.3% area) of pressure-outlet 12.

Reversed flow on 9701 faces (42.0% area) of pressure-outlet 13.

Reversed flow on 7218 faces (94.3% area) of pressure-outlet 14.
Stabilizing pressure coupled to enhance linear solver robustness.
Stabilizing temperature to enhance linear solver robustness.
Stabilizing temperature using GMRES to enhance linear solver robustness.

Divergence detected in AMG solver: temperature sub-iteration: 1 residual - h: 4.128122e+12; u: 1.673299e+12; v: 1.005001e+12; t: 6.797438e+86
sub-iteration: 2 residual - h: 4.880305e+12; u: 1.013807e+13; v: 5.614040e+12; t: 3.893496e+89
sub-iteration: 3 residual - h: 1.700494e+13; u: 1.673299e+12; v: 1.015001e+12; t: 5.396805e+92
sub-iteration: 4 residual - h: 1.700494e+13; u: 1.013807e+13; v: 5.614040e+12; t: 3.366202e+94
sub-iteration: 5 residual - h: 1.700494e+13; u: 1.673299e+12; v: 1.015001e+12; t: 1.982033e+99
sub-iteration: 6 residual - h: 1.700494e+13; u: 1.013807e+13; v: 5.614040e+12; t: 6.886049e+104
sub-iteration: 7 residual - h: 1.700494e+13; u: 1.673299e+12; v: 1.015001e+12; t: 3.721742e+109
sub-iteration: 8 residual - h: 1.700494e+13; u: 1.013807e+13; v: 5.614040e+12; t: 1.167922e+115
sub-iteration: 9 residual - h: 1.700494e+13; u: 1.673299e+12; v: 1.015001e+12; t: 1.904285e+121
sub-iteration: 10 residual - h: 1.700494e+13; u: 1.013807e+13; v: 5.614040e+12; t: 2.468745e+129
Film time = 8.631055e-03 with timestep = 1.4e-04, (max_cfl: 3.325411e+08)
Maximum CFL reached. Reduce time step by factor of 2

turbulent viscosity limited to viscosity ratio of 1.000000e+05 in 140044 cells

Divergence detected in AMG solver: temperature

[azhao]

Here is the case file in case it is needed.

[Rob]

I'm not permitted to download or open files/attachments. /forum/discussion/23093/why-ansys-employees-dont-download-attachments
Is that the turbulence model failing first? What are the initial conditions set as?

[azhao]

Hi Rob,
I understand the company policy restriction.
Yes, the turbulence model is failing, but I think it is not before the failing of the EWF model, as can be seen in the following message, the EWF Courant number is already on the magnitude of 1e8, while the residuals for the k and w equations are still fine and there is no error message related to the turbulent viscosity. Everything blows off after the second injection of particles between iteration 19 and 20.
The continuous phase initialization was done by the Hybrid Initialization function of the Fluent.
The EWF is initialized from 1e-6 m thickness.
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Reversed flow on 5795 faces (23.4% area) of pressure-outlet 12.

Reversed flow on 3709 faces (15.3% area) of pressure-outlet 13.

Reversed flow on 5465 faces (74.3% area) of pressure-outlet 14.
sub-iteration: 1 residual - h: 1.454633e-02; u: 1.401688e-03; v: 1.794648e-03; t: 3.881760e+00
sub-iteration: 2 residual - h: 1.537926e-02; u: 1.266208e-03; v: 1.160494e-03; t: 4.136686e+00
sub-iteration: 3 residual - h: 1.563476e-02; u: 1.296829e-03; v: 1.170926e-03; t: 4.096284e+00
sub-iteration: 4 residual - h: 1.565359e-02; u: 1.300766e-03; v: 1.172663e-03; t: 4.092070e+00
sub-iteration: 5 residual - h: 1.565477e-02; u: 1.301097e-03; v: 1.173167e-03; t: 4.091915e+00
sub-iteration: 6 residual - h: 1.565487e-02; u: 1.301123e-03; v: 1.173209e-03; t: 4.091897e+00
sub-iteration: 7 residual - h: 1.565488e-02; u: 1.301126e-03; v: 1.173213e-03; t: 4.091894e+00
sub-iteration: 8 residual - h: 1.565489e-02; u: 1.301126e-03; v: 1.173213e-03; t: 4.091894e+00
sub-iteration: 9 residual - h: 1.565489e-02; u: 1.301127e-03; v: 1.173214e-03; t: 4.091894e+00
sub-iteration: 10 residual - h: 1.565489e-02; u: 1.301127e-03; v: 1.173214e-03; t: 4.091894e+00
Film time = 6.495312e-03 with timestep = 1.1e-03, (max_cfl: 4.900231e-02)
19 2.5855e-04 1.4305e-03 1.6009e-03 1.3261e-03 2.6537e-05 2.1268e-02 1.4624e-02 13:24:25 9981

Advancing DPM injections ....

Reversed flow on 4737 faces (17.1% area) of pressure-outlet 12.

Reversed flow on 4625 faces (15.4% area) of pressure-outlet 13.

Reversed flow on 5648 faces (71.5% area) of pressure-outlet 14.
sub-iteration: 1 residual - h: 2.442087e-02; u: 4.539772e-03; v: 5.469951e-03; t: 6.283665e+00
sub-iteration: 2 residual - h: 2.078571e-02; u: 3.482383e-03; v: 4.080320e-03; t: 6.284156e+00
sub-iteration: 3 residual - h: 2.285721e-02; u: 3.798136e-03; v: 4.474999e-03; t: 5.941563e+00
sub-iteration: 4 residual - h: 2.315077e-02; u: 3.857611e-03; v: 4.563861e-03; t: 5.885062e+00
sub-iteration: 5 residual - h: 2.318995e-02; u: 3.866508e-03; v: 4.578689e-03; t: 5.879676e+00
sub-iteration: 6 residual - h: 2.319425e-02; u: 3.867617e-03; v: 4.580682e-03; t: 5.878798e+00
sub-iteration: 7 residual - h: 2.319489e-02; u: 3.867763e-03; v: 4.580935e-03; t: 5.878713e+00
sub-iteration: 8 residual - h: 2.319498e-02; u: 3.867781e-03; v: 4.580963e-03; t: 5.878696e+00
sub-iteration: 9 residual - h: 2.319499e-02; u: 3.867783e-03; v: 4.580966e-03; t: 5.878694e+00
sub-iteration: 10 residual - h: 2.319500e-02; u: 3.867783e-03; v: 4.580966e-03; t: 5.878693e+00
Film time = 7.634375e-03 with timestep = 1.1e-03, (max_cfl: 5.551702e-02)
Maximum CFL reached. Reduce time step by factor of 2
Injecting 25 particle parcels with mass 2e-08
number tracked = 275
Injecting 25 particle parcels with mass 2e-08
number tracked = 300
Injecting 25 particle parcels with mass 2e-08
number tracked = 325
Injecting 25 particle parcels with mass 2e-08
number tracked = 350
Injecting 25 particle parcels with mass 2e-08
number tracked = 375
Injecting 25 particle parcels with mass 2e-08
number tracked = 400
Eulerian wall film particles:
absorbed = 13
Injecting 25 particle parcels with mass 2e-08
number tracked = 412
Eulerian wall film particles:
absorbed = 24
Injecting 25 particle parcels with mass 2e-08
number tracked = 413
Eulerian wall film particles:
absorbed = 27
Injecting 25 particle parcels with mass 2e-08
number tracked = 411
Eulerian wall film particles:
absorbed = 25
Injecting 25 particle parcels with mass 2e-08
number tracked = 411
Eulerian wall film particles:
absorbed = 28
20 2.8301e-04 1.4456e-03 1.7631e-03 1.3289e-03 2.6017e-05 1.9681e-02 1.7461e-02 12:56:32 9980

Reversed flow on 3216 faces (11.6% area) of pressure-outlet 12.

Reversed flow on 6374 faces (23.0% area) of pressure-outlet 13.

Reversed flow on 6535 faces (81.9% area) of pressure-outlet 14.
sub-iteration: 1 residual - h: 4.827287e+04; u: 1.065710e+07; v: 5.123238e+06; t: 1.889212e+09
sub-iteration: 2 residual - h: 7.221070e+07; u: 9.999922e+09; v: 9.998893e+09; t: 9.292597e+14
sub-iteration: 3 residual - h: 9.636267e+10; u: 1.998950e+10; v: 1.001359e+10; t: 7.571465e+20
sub-iteration: 4 residual - h: 1.335207e+11; u: 1.031386e+10; v: 1.003249e+10; t: 4.226672e+27
sub-iteration: 5 residual - h: 2.276113e+11; u: 2.000001e+10; v: 2.000014e+10; t: 3.586293e+34
sub-iteration: 6 residual - h: 2.842495e+11; u: 2.000037e+10; v: 2.000015e+10; t: 5.801935e+40
sub-iteration: 7 residual - h: 2.850022e+11; u: 2.001262e+10; v: 2.000004e+10; t: 2.434394e+47
sub-iteration: 8 residual - h: 4.669491e+11; u: 2.000816e+10; v: 2.001098e+10; t: 1.023309e+54
sub-iteration: 9 residual - h: 3.958626e+11; u: 2.000426e+10; v: 2.000820e+10; t: 4.295725e+60
sub-iteration: 10 residual - h: 3.462602e+11; u: 2.000963e+10; v: 2.000459e+10; t: 2.195055e+60
Film time = 8.203906e-03 with timestep = 5.7e-04, (max_cfl: 3.514216e+08)
Maximum CFL reached. Reduce time step by factor of 2
21 3.0254e-04 1.5662e-03 2.8951e-03 1.5209e-03 3.4102e-05 2.4878e-02 3.8683e-02 13:07:29 9979

Reversed flow on 7236 faces (13.3% area) of pressure-outlet 12.

Reversed flow on 8690 faces (33.9% area) of pressure-outlet 13.

Reversed flow on 6990 faces (89.4% area) of pressure-outlet 14.
Negative k in 2 cells after linear solve.
temperature limited to 1.000000e+00 in 344187 cells on zone 10 in domain 1
temperature limited to 5.000000e+03 in 605874 cells on zone 10 in domain 1
sub-iteration: 1 residual - h: 3.266015e+12; u: 2.001124e+10; v: 2.001292e+10; t: 2.373027e+63
sub-iteration: 2 residual - h: 3.188269e+12; u: 2.002072e+10; v: 2.000640e+10; t: 1.989278e+63
sub-iteration: 3 residual - h: 3.290642e+12; u: 2.000174e+10; v: 2.000627e+10; t: 2.285805e+62
sub-iteration: 4 residual - h: 3.282556e+12; u: 2.000760e+10; v: 2.000925e+10; t: 3.836486e+60
sub-iteration: 5 residual - h: 1.636357e+12; u: 2.000537e+10; v: 2.001137e+10; t: 6.722036e+65
sub-iteration: 6 residual - h: 1.672078e+12; u: 2.002215e+10; v: 2.001482e+10; t: 1.216383e+71
sub-iteration: 7 residual - h: 3.322391e+12; u: 2.000885e+10; v: 2.001572e+10; t: 1.144955e+71
sub-iteration: 8 residual - h: 3.315233e+12; u: 2.000193e+10; v: 2.000204e+10; t: 1.983045e+72
sub-iteration: 9 residual - h: 3.315119e+12; u: 2.000641e+10; v: 2.001215e+10; t: 1.700176e+72
sub-iteration: 10 residual - h: 3.286556e+12; u: 2.002153e+10; v: 2.000446e+10; t: 1.959171e+71
Film time = 8.488672e-03 with timestep = 2.8e-04, (max_cfl: 8.151928e+09)
Maximum CFL reached. Reduce time step by factor of 2

turbulent viscosity limited to viscosity ratio of 1.000000e+05 in 529711 cells
22 3.9665e-04 1.5730e-03 2.6496e-03 1.5462e-03 1.4345e+40 7.7525e+12 1.0242e+00 15:29:16 9978

[Rob]

Can you try in 2022R1 as the EWF had some extra stability functions added in.

[azhao]

Hi Rob I tried 2022R1 as shown in the following figure. The same problem remains...

[Rob]

Turn off energy. I'm not permitted to get the model so it's a bit harder to diagnose.

[azhao]

Hi Rob After turning off the energy equation in the EWF model, everything went well except a huge residual rise at the beginning as shown in the following figures.
119 4.8185e-04 3.1840e-03 3.9198e-03 3.0896e-03 7.1001e-06 5.0463e-03 1.7154e-02 26:34:53 9881

Advancing DPM injections ....

Reversed flow on 3040 faces (48.1% area) of pressure-outlet 12.

Reversed flow on 10405 faces (88.1% area) of pressure-outlet 13.

Reversed flow on 7273 faces (98.1% area) of pressure-outlet 14.
sub-iteration: 1 residual - h: 1.050327e+04; u: 5.754825e+04; v: 2.915104e+04
sub-iteration: 2 residual - h: 6.504198e+01; u: 3.419180e+03; v: 2.812191e+03
sub-iteration: 3 residual - h: 6.306149e+00; u: 3.647290e+02; v: 1.177123e+02
sub-iteration: 4 residual - h: 5.503609e-01; u: 2.380574e+01; v: 8.346133e+00
sub-iteration: 5 residual - h: 3.450068e-02; u: 7.064277e-01; v: 2.178804e-01
sub-iteration: 6 residual - h: 1.283698e-03; u: 3.649173e-02; v: 5.285062e-02
sub-iteration: 7 residual - h: 6.184168e-05; u: 8.121571e-03; v: 6.495069e-03
sub-iteration: 8 residual - h: 1.006310e-05; u: 8.526671e-04; v: 4.520582e-04
sub-iteration: 9 residual - h: 1.267574e-06; u: 6.637187e-05; v: 1.855256e-05
sub-iteration: 10 residual - h: 1.114968e-07; u: 3.628433e-06; v: 4.135351e-06
Film time = 7.761697e-03 with timestep = 7.0e-08, (max_cfl: 4.647466e-02)
Injecting 25 particle parcels with mass 2e-08
number tracked = 408
Eulerian wall film particles:
absorbed = 25
Injecting 25 particle parcels with mass 2e-08
number tracked = 408
Eulerian wall film particles:
absorbed = 27
Injecting 25 particle parcels with mass 2e-08
number tracked = 406
Eulerian wall film particles:
absorbed = 25
Injecting 25 particle parcels with mass 2e-08
number tracked = 406
Eulerian wall film particles:
absorbed = 25
Injecting 25 particle parcels with mass 2e-08
number tracked = 406
Eulerian wall film particles:
absorbed = 24
Injecting 25 particle parcels with mass 2e-08
number tracked = 407
Eulerian wall film particles:
absorbed = 24
Injecting 25 particle parcels with mass 2e-08
number tracked = 408
Eulerian wall film particles:
absorbed = 23
Injecting 25 particle parcels with mass 2e-08
number tracked = 410
Eulerian wall film particles:
absorbed = 26
Injecting 25 particle parcels with mass 2e-08
number tracked = 409
Eulerian wall film particles:
absorbed = 25
Injecting 25 particle parcels with mass 2e-08
number tracked = 409
Eulerian wall film particles:
absorbed = 26
120 4.8607e-04 3.3282e-03 4.0813e-03 3.1492e-03 6.8589e-06 4.8522e-03 1.7238e-02 26:12:11 9880

[Rob]

Good. What does the film look like now?

[azhao]

It looks much more reasonable than before as shown in the following figure but still not right. It shows an unphysical pattern in the circumferential direction while it should be statistically uniform. And it accumulates to a high value instead of spreading radially. (This might be due to the spread term in the momentum equation of EWF model being turned off or the simulation time is not long enough?)

[Rob]

Run for longer. It may be the spray hasn't been hitting the surface long enough to form a proper film. I don't like the streaking, but it's hard to say if that's due to lack of solved time or something else.
You're right to expect a more uniform ring, but that's also a function of mesh and droplet count.

[azhao]

Unfortunately it still diverged after 1250 iterations due to the same too large Maximum CFL number in the EWF model...
Also, an error window reports floating point exception.
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

turbulent viscosity limited to viscosity ratio of 1.000000e+05 in 843571 cells
12509.8709e+043.2876e-012.0323e-018.0409e-015.1685e-061.0080e+147.2821e-01 81:51:45 18750

Reversed flow on 13370 faces (92.6% area) of pressure-outlet 12.

Reversed flow on 9653 faces (75.7% area) of pressure-outlet 13.

Reversed flow on 3668 faces (53.8% area) of pressure-outlet 14.
Stabilizing pressure coupled to enhance linear solver robustness.
Stabilizing pressure coupled using GMRES to enhance linear solver robustness.

Experiencing convergence difficulties - temporarily relaxing and trying again...

Experiencing convergence difficulties - temporarily relaxing and trying again...

Experiencing convergence difficulties - temporarily relaxing and trying again...

Experiencing convergence difficulties - temporarily relaxing and trying again...
Stabilizing pressure coupled to enhance linear solver robustness.
Stabilizing pressure coupled using GMRES to enhance linear solver robustness.

Experiencing convergence difficulties - temporarily relaxing and trying again...
Divergence detected in AMG solver: pressure coupled sub-iteration: 1 residual - h: 3.091311e+33; u: 2.805212e+33; v: 1.333100e+32
sub-iteration: 2 residual - h: 3.091311e+33; u: 1.633315e+32; v: 2.900875e+31
sub-iteration: 3 residual - h: 2.871481e+27; u: 2.805212e+33; v: 1.333100e+32
sub-iteration: 4 residual - h: 2.871481e+27; u: 1.633315e+32; v: 2.900875e+31
sub-iteration: 5 residual - h: 2.871481e+27; u: 2.805212e+33; v: 1.333100e+32
sub-iteration: 6 residual - h: 2.871481e+27; u: 1.633315e+32; v: 2.900875e+31
sub-iteration: 7 residual - h: 2.871481e+27; u: 2.805212e+33; v: 1.333100e+32
sub-iteration: 8 residual - h: 2.871481e+27; u: 1.633315e+32; v: 2.900875e+31
sub-iteration: 9 residual - h: 2.871481e+27; u: 2.805212e+33; v: 1.333100e+32
sub-iteration: 10 residual - h: 2.871481e+27; u: 1.633315e+32; v: 2.900875e+31
Film time = 7.807621e-03 with timestep = 7.6e-11, (max_cfl: 1.894628e+15)
Maximum CFL reached. Reduce time step by factor of 2

turbulent viscosity limited to viscosity ratio of 1.000000e+05 in 840131 cells

Error at host: floating point exception

Error: floating point exception
Error Object: #f

[azhao]

Unfortunately it still diverged after 1250 iterations due to the same too large Maximum CFL number in the EWF model...

+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

Advancing DPM injections ....

Reversed flow on 4239 faces (46.7% area) of pressure-outlet 12.

Reversed flow on 6042 faces (48.4% area) of pressure-outlet 13.

Reversed flow on 3073 faces (42.4% area) of pressure-outlet 14.
sub-iteration: 1 residual - h: 3.456029e+16; u: 2.676280e+21; v: 2.748499e+20
sub-iteration: 2 residual - h: 6.626032e+15; u: 1.741486e+19; v: 7.348845e+17
sub-iteration: 3 residual - h: 6.626032e+15; u: 2.676280e+21; v: 2.748499e+20
sub-iteration: 4 residual - h: 6.626032e+15; u: 1.741486e+19; v: 7.348845e+17
sub-iteration: 5 residual - h: 6.626032e+15; u: 2.676280e+21; v: 2.748499e+20
sub-iteration: 6 residual - h: 6.626032e+15; u: 1.741486e+19; v: 7.348845e+17
sub-iteration: 7 residual - h: 6.626032e+15; u: 2.676280e+21; v: 2.748499e+20
sub-iteration: 8 residual - h: 6.626032e+15; u: 1.741486e+19; v: 7.348845e+17
sub-iteration: 9 residual - h: 6.626032e+15; u: 2.676280e+21; v: 2.748499e+20
sub-iteration: 10 residual - h: 6.626032e+15; u: 1.741486e+19; v: 7.348845e+17
Film time = 7.807621e-03 with timestep = 1.5e-10, (max_cfl: 1.941754e+03)
Maximum CFL reached. Reduce time step by factor of 2
Injecting 25 particle parcels with mass 2e-08
number tracked = 1217, incomplete = 1207, shed = 810
Eulerian wall film particles:
absorbed = 10

Warning: injection: injection-0: zero particle mass, cannot conserve injection mass flow rate
Injecting 25 particle parcels with mass 0
number tracked = 776, escaped = 70, incomplete = 696
Eulerian wall film particles:
absorbed = 10, rebound = 2

Injection injection-0: Out of 25 locations, 25 lost outside the domain in axial staggering; reducing staggering factor to 80% for them.
Injection injection-0: Out of 25 locations, 25 still lost outside the domain in axial staggering; reducing staggering factor to 60% for them.
Injection injection-0: Out of 25 locations, 25 still lost outside the domain in axial staggering; reducing staggering factor to 40% for them.
Injection injection-0: Out of 25 locations, 25 still lost outside the domain in axial staggering; reducing staggering factor to 20% for them.
Injection injection-0: Out of 25 locations, 25 still lost outside the domain in axial staggering; reducing staggering factor to 0% for them.

Warning: injection: injection-0: zero particle mass, cannot conserve injection mass flow rate
Injecting 25 particle parcels with mass 0
number tracked = 740, escaped = 42, incomplete = 672, shed = 44
Eulerian wall film particles:
absorbed = 26

Injection injection-0: Out of 25 locations, 25 lost outside the domain in axial staggering; reducing staggering factor to 80% for them.
Injection injection-0: Out of 25 locations, 25 still lost outside the domain in axial staggering; reducing staggering factor to 60% for them.
Injection injection-0: Out of 25 locations, 25 still lost outside the domain in axial staggering; reducing staggering factor to 40% for them.
Injection injection-0: Out of 25 locations, 25 still lost outside the domain in axial staggering; reducing staggering factor to 20% for them.
Injection injection-0: Out of 25 locations, 25 still lost outside the domain in axial staggering; reducing staggering factor to 0% for them.

Warning: injection: injection-0: zero particle mass, cannot conserve injection mass flow rate
Injecting 25 particle parcels with mass 0
number tracked = 651, escaped = 10, incomplete = 639, shed = 12
Eulerian wall film particles:
absorbed = 2

Warning:100.0000% of the total discrete phase mass was not tracked for the expected residence time:
0.0002 s less on a mass-weighted average (which is 74.4535% of their total age or 100.0000% of the last time step).
turbulent viscosity limited to viscosity ratio of 1.000000e+05 in 843571 cells
12509.8709e+043.2876e-012.0323e-018.0409e-015.1685e-061.0080e+147.2821e-01 81:51:45 18750

Reversed flow on 13370 faces (92.6% area) of pressure-outlet 12.

Reversed flow on 9653 faces (75.7% area) of pressure-outlet 13.

Reversed flow on 3668 faces (53.8% area) of pressure-outlet 14.
Stabilizing pressure coupled to enhance linear solver robustness.
Stabilizing pressure coupled using GMRES to enhance linear solver robustness.

Experiencing convergence difficulties - temporarily relaxing and trying again...

Experiencing convergence difficulties - temporarily relaxing and trying again...

Experiencing convergence difficulties - temporarily relaxing and trying again...

Experiencing convergence difficulties - temporarily relaxing and trying again...
Stabilizing pressure coupled to enhance linear solver robustness.
Stabilizing pressure coupled using GMRES to enhance linear solver robustness.

Experiencing convergence difficulties - temporarily relaxing and trying again...
Divergence detected in AMG solver: pressure coupled sub-iteration: 1 residual - h: 3.091311e+33; u: 2.805212e+33; v: 1.333100e+32
sub-iteration: 2 residual - h: 3.091311e+33; u: 1.633315e+32; v: 2.900875e+31
sub-iteration: 3 residual - h: 2.871481e+27; u: 2.805212e+33; v: 1.333100e+32
sub-iteration: 4 residual - h: 2.871481e+27; u: 1.633315e+32; v: 2.900875e+31
sub-iteration: 5 residual - h: 2.871481e+27; u: 2.805212e+33; v: 1.333100e+32
sub-iteration: 6 residual - h: 2.871481e+27; u: 1.633315e+32; v: 2.900875e+31
sub-iteration: 7 residual - h: 2.871481e+27; u: 2.805212e+33; v: 1.333100e+32
sub-iteration: 8 residual - h: 2.871481e+27; u: 1.633315e+32; v: 2.900875e+31
sub-iteration: 9 residual - h: 2.871481e+27; u: 2.805212e+33; v: 1.333100e+32
sub-iteration: 10 residual - h: 2.871481e+27; u: 1.633315e+32; v: 2.900875e+31
Film time = 7.807621e-03 with timestep = 7.6e-11, (max_cfl: 1.894628e+15)
Maximum CFL reached. Reduce time step by factor of 2

turbulent viscosity limited to viscosity ratio of 1.000000e+05 in 840131 cells

Error at host: floating point exception

Error: floating point exception
Error Object: #f

[Rob]

The incompletes suggest something has failed in the flow solver too. We know the mesh isn't ideal given the jump in cell size. We also know it's a symmetrical domain. Chop the domain down to a quarter & remesh with a more uniform size. Use tets if necessary, but avoid the jump in size. This will allow you to either refine the mesh (same cell count but smaller volume) or speed up the simulation (same-ish resolution as it's a smaller domain). The injection will report several lost parcels but that can be ignored as they're just being released in the 3/4 that's not there.
And use either Workbench (Ansys) Meshing or Fluent Meshing. Or tets in ICEM CFD, avoid the O-grid as that's possibly what is causing the problem as the cell size jumps moving away from the central block.

[azhao]

Hi Rob The above figure shows the fluid zone which is cut into a quarter and the corresponding uniform-sized tetrahedral meshes. As you said Fluent reported some missing mass of injected particles. The amount of missing mass changes for every injection. May I assume the statistical average of the injected mass would be 25%?

----------------------------------------------------------------------------------------------------------------------------
Injection injection-0: Out of 7 locations, 4 lost outside the domain in axial staggering; reducing staggering factor to 80% for them.
Injection injection-0: Out of 7 locations, 2 still lost outside the domain in axial staggering; reducing staggering factor to 60% for them.
Injection injection-0: Out of 7 locations, 1 still lost outside the domain in axial staggering; reducing staggering factor to 40% for them.
Injection injection-0: Out of 7 locations, 1 still lost outside the domain in axial staggering; reducing staggering factor to 20% for them.

Warning: Injection injection-0: LOST 18 out of 25 injection locations, probably outside the domain. This will inject 68% LESS MASS.

Pressure-swirl atomizer injector:
Pressure 320000 less than down-stream pressure 4.16935e+71, or
Invalid downstream density 1.225
Skipping injection
number tracked = 0

turbulent viscosity limited to viscosity ratio of 1.000000e+05 in 1488 cells
3130 4.7840e+21 1.8051e+23 6.8509e+12 2.0474e+22 3.1447e-06 1.1489e+00 7.1855e+41 17:35:15 16870

Reversed flow on 999 faces (56.9% area) of pressure-outlet 5.

Reversed flow on 598 faces (35.8% area) of pressure-outlet 6.

Reversed flow on 1235 faces (47.1% area) of pressure-outlet 8.
Stabilizing pressure coupled to enhance linear solver robustness.
Stabilizing pressure coupled using GMRES to enhance linear solver robustness.

Experiencing convergence difficulties - temporarily relaxing and trying again...

Experiencing convergence difficulties - temporarily relaxing and trying again...

Experiencing convergence difficulties - temporarily relaxing and trying again...

Experiencing convergence difficulties - temporarily relaxing and trying again...
Stabilizing pressure coupled to enhance linear solver robustness.
Stabilizing pressure coupled using GMRES to enhance linear solver robustness.

Experiencing convergence difficulties - temporarily relaxing and trying again...

Divergence detected in AMG solver: pressure coupled Stabilizing k to enhance linear solver robustness.
Stabilizing k using GMRES to enhance linear solver robustness.

Divergence detected in AMG solver: k Stabilizing omega to enhance linear solver robustness.
Stabilizing omega using GMRES to enhance linear solver robustness.

Divergence detected in AMG solver: omega Stabilizing temperature to enhance linear solver robustness.
Stabilizing temperature using GMRES to enhance linear solver robustness.

Divergence detected in AMG solver: temperature sub-iteration: 1 residual - h: 3.529965e+86; u: 1.592715e+94; v: 1.630529e+94
sub-iteration: 2 residual - h: 5.865961e+85; u: 3.570367e+91; v: 1.260537e+93
sub-iteration: 3 residual - h: 6.186259e+85; u: 1.592715e+94; v: 1.630529e+94
sub-iteration: 4 residual - h: 6.186259e+85; u: 3.570367e+91; v: 1.260537e+93
sub-iteration: 5 residual - h: 6.186259e+85; u: 1.592715e+94; v: 1.630529e+94
sub-iteration: 6 residual - h: 6.186259e+85; u: 3.570367e+91; v: 1.260537e+93
sub-iteration: 7 residual - h: 6.186259e+85; u: 1.592715e+94; v: 1.630529e+94
sub-iteration: 8 residual - h: 6.186259e+85; u: 3.570367e+91; v: 1.260537e+93
sub-iteration: 9 residual - h: 6.186259e+85; u: 1.592715e+94; v: 1.630529e+94
sub-iteration: 10 residual - h: 6.186259e+85; u: 3.570367e+91; v: 1.260537e+93
Film time = 5.329656e-03 with timestep = 1.3e-12, (max_cfl: 5.582687e+71)
Maximum CFL reached. Reduce time step by factor of 2

[Rob]

OK, I'd hoped this would help. Next is to fix the droplet size and try again. Use a hollow cone and 100 streams (it's divisible by 4).
As an aside, if your post doesn't show up please only try again once. I think the filters saw the multi-post and put them all in the spam box.

[azhao]

Hi Rob I increased the particle steam to 100 and changed the nozzle type to hollow cone with its parameters as follows. It is noted that the spray velocity is calculated by setting discharge coefficient to 0.7.
Now the missed spray mass is exactly 75%, is it because the Stagger positions option is turned off?
Unfortunately, the solution blows off again after 6000+ iterations......
-----------------------------------------------------------------------------------------------------------------------------------------------------------------
turbulent viscosity limited to viscosity ratio of 1.000000e+05 in 73551 cells

itercontinuityx-velocityy-velocityz-velocityenergykomegatime/iter
64471.3401e+306.8450e+151.1586e-015.5702e+153.6026e-024.1768e+303.0208e+19 52:28:32 13553
Reversed flow on 647 faces (36.5% area) of pressure-outlet 5.
Reversed flow on 393 faces (23.6% area) of pressure-outlet 6.
Negative k in 14775 cells after linear solve.
All neighbors negative on 592 cells.
Stabilizing omega to enhance linear solver robustness.
Stabilizing omega using GMRES to enhance linear solver robustness.

Divergence detected in AMG solver: omega temperature limited to 1.000000e+00 in 18 cells on zone 3 in domain 1
temperature limited to 5.000000e+03 in 713 cells on zone 3 in domain 1
sub-iteration: 1 residual - h: 6.417020e+107; u: 7.316087e+113; v: 4.855059e+113
sub-iteration: 2 residual - h: 6.417023e+107; u: 9.033871e+124; v: 9.642971e+124
sub-iteration: 3 residual - h: 2.791816e+116; u: 7.316087e+113; v: 4.855059e+113
sub-iteration: 4 residual - h: 2.791816e+116; u: 9.033871e+124; v: 9.642971e+124
sub-iteration: 5 residual - h: 2.791816e+116; u: 7.316087e+113; v: 4.855059e+113
sub-iteration: 6 residual - h: 2.791816e+116; u: 9.033871e+124; v: 9.642971e+124
sub-iteration: 7 residual - h: 2.791816e+116; u: 7.316087e+113; v: 4.855059e+113
sub-iteration: 8 residual - h: 2.791816e+116; u: 9.033871e+124; v: 9.642971e+124
sub-iteration: 9 residual - h: 2.791816e+116; u: 7.316087e+113; v: 4.855059e+113
sub-iteration: 10 residual - h: 2.791816e+116; u: 9.033871e+124; v: 9.642971e+124
Film time = 2.748134e-02 with timestep = 1.0e-12, (max_cfl: 4.534227e+103)
Maximum CFL reached. Reduce time step by factor of 2
------------------------------------------------------------------------------------------------------------------------------------------------------------------
Then yesterday I tried if changing the turbulence model (from the default K-omega SST to realizable k-epsilon) would eliminate the error message of "turbulent viscosity limited to viscosity ratio of 1.000000e+05......". Finally, the divergence problem disappears! It is still calculating after 10000+ iterations and the residuals seem all fine. But I am not sure it is because of the turbulence model or because the discretization accuracy has been changed from second-order upwind (default for K-omega) to first-order upwind ((default for K-epsilon).

[Rob]

I usually expect k-w and k-e to behave in much the same way: the w bit is near the wall and both use much the same maths in the free stream. High order for turbulence isn't something I tend to use, so it's possible that the setting makes the solver too stiff. Difficult to diagnose without the files, and we're not permitted to do this via the Community. Good to hear it's working.

[azhao]

Hi Rob Thank you for your patience in this long discussion. Now I will try to put other factors back and see if it still works.
Besides, I did try to run the case with k-omega model and first order upwind spatial discretization scheme, the solution still blows off after 6000+ iterations, while the realizalbe k-epsilon model works fine. This is so strange to me.....
Injecting 25 particle parcels with mass 5e-09
number tracked = 269, escaped = 58, incomplete = 50, shed = 25

Warning:100.0000% of the total discrete phase mass was not tracked for the expected residence time:
3.02e-05 s less on a mass-weighted average (which is 11.1146% of their total age or 15.0930% of the last time step).


turbulent viscosity limited to viscosity ratio of 1.000000e+05 in 97185 cells
6010 1.2332e-02 1.1923e+00 4.5793e-01 8.5680e-01 2.1687e-06 8.1384e-01 1.2615e+00 19:33:26 13990

Reversed flow on 1353 faces (77.8% area) of pressure-outlet 5.

Reversed flow on 604 faces (36.0% area) of pressure-outlet 6.

Reversed flow on 1644 faces (61.5% area) of pressure-outlet 8.
sub-iteration: 1 residual - h: 3.844885e+04; u: 1.066348e+09; v: 1.945965e+08
sub-iteration: 2 residual - h: 1.514765e+03; u: 3.161290e+08; v: 3.636856e+07
sub-iteration: 3 residual - h: 1.552265e+02; u: 3.208125e+07; v: 2.969201e+06
sub-iteration: 4 residual - h: 1.577075e+01; u: 3.288545e+06; v: 2.975264e+05
sub-iteration: 5 residual - h: 1.603691e+00; u: 3.345456e+05; v: 3.009608e+04
sub-iteration: 6 residual - h: 1.630612e-01; u: 3.402588e+04; v: 3.056388e+03
sub-iteration: 7 residual - h: 1.657999e-02; u: 3.459926e+03; v: 3.106720e+02
sub-iteration: 8 residual - h: 1.685844e-03; u: 3.518090e+02; v: 3.158639e+01
sub-iteration: 9 residual - h: 1.714158e-04; u: 3.577189e+01; v: 3.211620e+00
sub-iteration: 10 residual - h: 1.742946e-05; u: 3.637270e+00; v: 3.265548e-01
Film time = 2.755759e-02 with timestep = 3.6e-10, (max_cfl: 1.258643e-01)
Maximum CFL reached. Reduce time step by factor of 2

turbulent viscosity limited to viscosity ratio of 1.000000e+05 in 139663 cells
6011 1.2623e+00 3.7001e-02 3.1007e-02 4.2146e-02 4.1384e-07 1.3674e+05 1.1393e+00 18:45:12 13989

Reversed flow on 1277 faces (73.0% area) of pressure-outlet 5.

Reversed flow on 737 faces (44.6% area) of pressure-outlet 6.

Reversed flow on 2000 faces (74.4% area) of pressure-outlet 8.
Negative k in 177 cells after linear solve.
All neighbors negative on 2 cells.
sub-iteration: 1 residual - h: 5.741291e+09; u: 4.921588e+08; v: 6.672962e+07
sub-iteration: 2 residual - h: 1.281817e+02; u: 7.600074e+07; v: 6.669881e+06
sub-iteration: 3 residual - h: 1.480683e+01; u: 2.432676e+06; v: 8.216693e+03
sub-iteration: 4 residual - h: 6.582004e-01; u: 7.004886e+04; v: 1.384374e+03
sub-iteration: 5 residual - h: 2.224449e-02; u: 1.955233e+03; v: 5.987396e+01
sub-iteration: 6 residual - h: 6.755315e-04; u: 5.391183e+01; v: 1.957430e+00
sub-iteration: 7 residual - h: 1.950298e-05; u: 1.476821e+00; v: 5.854702e-02
sub-iteration: 8 residual - h: 5.483125e-07; u: 4.030657e-02; v: 1.675844e-03
sub-iteration: 9 residual - h: 1.519220e-08; u: 1.100063e-03; v: 4.649162e-05
sub-iteration: 10 residual - h: 4.147296e-10; u: 3.194809e-05; v: 2.145767e-06
Film time = 2.755759e-02 with timestep = 1.8e-10, (max_cfl: 1.203342e-01)
Maximum CFL reached. Reduce time step by factor of 2
turbulent viscosity limited to viscosity ratio of 1.000000e+05 in 162856 cells
6012 1.1067e+00 1.3190e+03 7.2258e-01 9.5467e+02 5.6622e-06 6.9358e-01 3.6442e+00 18:53:14 13988

Reversed flow on 1723 faces (98.2% area) of pressure-outlet 5.

Reversed flow on 1017 faces (60.5% area) of pressure-outlet 6.

Reversed flow on 2666 faces (99.2% area) of pressure-outlet 8.
sub-iteration: 1 residual - h: 1.382905e+09; u: 1.042138e+12; v: 3.138733e+12
sub-iteration: 2 residual - h: 3.853988e+06; u: 5.112315e+12; v: 2.192171e+12
sub-iteration: 3 residual - h: 9.521054e+06; u: 2.905077e+13; v: 2.138299e+12
sub-iteration: 4 residual - h: 9.608164e+06; u: 1.007068e+13; v: 4.947367e+12
sub-iteration: 5 residual - h: 9.608164e+06; u: 2.905077e+13; v: 2.202171e+12
sub-iteration: 6 residual - h: 9.608164e+06; u: 1.007068e+13; v: 4.947367e+12
sub-iteration: 7 residual - h: 9.608164e+06; u: 2.905077e+13; v: 2.202171e+12
sub-iteration: 8 residual - h: 9.608164e+06; u: 1.007068e+13; v: 4.947367e+12
sub-iteration: 9 residual - h: 9.608164e+06; u: 2.905077e+13; v: 2.202171e+12
sub-iteration: 10 residual - h: 9.608164e+06; u: 1.007068e+13; v: 4.947367e+12
Film time = 2.755759e-02 with timestep = 9.1e-11, (max_cfl: 2.711508e+00)
Maximum CFL reached. Reduce time step by factor of 2

turbulent viscosity limited to viscosity ratio of 1.000000e+05 in 182368 cells
6013 1.3607e+03 1.0403e+00 1.7850e-01 8.8734e-01 1.1785e-06 3.7896e+08 2.2363e+00 18:13:01 13987

Reversed flow on 682 faces (38.6% area) of pressure-outlet 5.

Reversed flow on 945 faces (56.7% area) of pressure-outlet 6.

Reversed flow on 1260 faces (46.2% area) of pressure-outlet 8.
sub-iteration: 1 residual - h: 5.083086e+20; u: 2.797341e+19; v: 2.944314e+19
sub-iteration: 2 residual - h: 5.083086e+20; u: 3.545303e+27; v: 2.201262e+27
sub-iteration: 3 residual - h: 5.083086e+20; u: 3.964509e+22; v: 2.700608e+22
sub-iteration: 4 residual - h: 5.083086e+20; u: 3.545303e+27; v: 2.201262e+27
sub-iteration: 5 residual - h: 5.083086e+20; u: 3.964509e+22; v: 2.700608e+22
sub-iteration: 6 residual - h: 5.083086e+20; u: 3.545303e+27; v: 2.201262e+27
sub-iteration: 7 residual - h: 5.083086e+20; u: 3.964509e+22; v: 2.700608e+22
sub-iteration: 8 residual - h: 5.083086e+20; u: 3.545303e+27; v: 2.201262e+27
sub-iteration: 9 residual - h: 5.083086e+20; u: 3.964509e+22; v: 2.700608e+22
sub-iteration: 10 residual - h: 5.083086e+20; u: 3.545303e+27; v: 2.201262e+27
Film time = 2.755759e-02 with timestep = 4.5e-11, (max_cfl: 3.720834e+09)
Maximum CFL reached. Reduce time step by factor of 2

[Rob]

Which suggests the k-w model is picking up something odd on the wall. If you run both with no DPM or film how do the results differ?

[azhao]

Hi Rob,
After checking the liquid film thickness distribution after 13000+ iterations, it turns out the result is still incorrect (it is not even axisymmetric!) yet, although it does converge

[azhao]

And I increased the injected particles from 100 to 400, aiming to check if the aforementioned nonphysical distribution is due to the non enough particle parcels. The solution blows off after 6000+ iterations again, this is for the realizable k-epsilon model case....

[Rob]

I think the result is correct, but that the streams are clustered so the spray isn't even. Set 3 injections, D10, D50 & D90 (or D25, D50 & D75 - we're after a sample) with suitable mass flow rates. The solver is failing for a reason, so it's now a case of learning why. When plotting the film thickness do so with node values off as that'll highlight any mesh related problems.

[azhao]

May I ask what do you mean by D10, D50, D90, does the number after D denote the particle stream numbers?

[Rob]

Diameter - 10%, 50% and 90% as you've got a size distribution.

[azhao]

Sorry I did not get which setting you are referring to. Do you mean the parcel definition of the injection nozzle, that I should set the parcel diameter as 10%, 50%, 90% of the injection diameters?

[Rob]

Not the parcels, leave that as default. In the injection point properties set up three injections and use the diameters rather than the atomiser model. You'll need to use a cone.

[azhao]

Hi, Rob I did the settings for with hollow cone models just as you said, with 3 injectors injecting droplets whose sizes are 90%, 50% 10% of the nozzle diameter. The mass flow rate is just divided by 3 for each injector. It is quite strange that the solution diverges again after 6000+ iterations. The EWF temperature blows off while the wall thickness is still fine. This is quite strange so I started again from the converged case with the new settings.. But the solution diverges again. As you said there must be an unfound reason causing this...

Divergence detected in AMG solver: temperature sub-iteration: 1 residual - h: 2.414831e+07; u: 2.124927e+10; v: 5.058492e+10; t: 5.676491e+76
sub-iteration: 2 residual - h: 5.733142e+04; u: 6.915366e+08; v: 2.500983e+08; t: 7.915565e+36
sub-iteration: 3 residual - h: 7.467264e+02; u: 1.474016e+08; v: 4.440876e+07; t: 1.866327e+38
sub-iteration: 4 residual - h: 1.704392e+02; u: 1.753650e+07; v: 5.015200e+06; t: 4.400412e+39
sub-iteration: 5 residual - h: 2.187857e+01; u: 2.030204e+05; v: 3.245214e+05; t: 1.037526e+41
sub-iteration: 6 residual - h: 7.022633e-01; u: 4.577671e+05; v: 1.530065e+05; t: 2.446272e+42
sub-iteration: 7 residual - h: 4.712636e-01; u: 1.229956e+05; v: 3.827056e+04; t: 5.767805e+43
sub-iteration: 8 residual - h: 1.392537e-01; u: 1.815837e+04; v: 5.342540e+03; t: 1.359929e+45
sub-iteration: 9 residual - h: 2.192493e-02; u: 1.070329e+03; v: 3.925675e+02; t: 3.206432e+46
sub-iteration: 10 residual - h: 1.561311e-03; u: 2.539414e+02; v: 9.019746e+01; t: 7.560103e+47
Film time = 5.376527e-02 with timestep = 4.2e-10, (max_cfl: 1.193756e-01)
Maximum CFL reached. Reduce time step by factor of 2

turbulent viscosity limited to viscosity ratio of 1.000000e+05 in 52 cells

Error at host: floating point exception

Error: floating point exception
Error Object: #f

Message: The previous active graphics window 3 contains a 2D plot, which is incompatible with injections display.
Injections are displayed in a new graphics window 52.

[Rob]

The only clue is that the film thickness is growing. How thick is the first cell? Ie the film is getting on for 1mm, how does this compare to the mesh? Note, the film can be 1m thick in a 1mm deep domain so it's not a solver check, it's more likely flow related.

[azhao]

Hi, Rob As shown in the following figure, most of the first cells are near 0.1mm, which is much smaller than the liquid film thickness, which is around 0.004 mm in the spray impact wall region. (Though the solution diverges, the liquid film thickness distribution seems more reasonable than before) And from the static pressure contour, we could see it diverged quite a lot. And the wall temperature is not even real number, changes to NAN. anymore. I guess that is the reason the floating number occurs.
By the way, I checked the theory guide on EWF, but I did not understand how ANSYS Fluent handles the turbulence boundary condition when the EWF model is turned on. In this case, the EWF thickness is larger than the first cell. So how the wall function is actually applied. Is it still applied on this first cell near the wall or it is actually applied on the first cell that is located outside the EWF?

[Rob]

EWF uses a virtual cell, so the wall functions etc are still calculated on the near wall cell regardless of how thick the film is. Can you replot the thickness with node values off?

[azhao]

[Rob]

OK, refine the mesh: having all of the film in one cell isn't going to assist solver stability.

[azhao]

I see. So EWF and Lagrangian wall film model are both modeling virtual on wall liquid films. So, if the liquid wall film thickness is actually larger than a threshold, like in the above case, several millimeters, even bigger than the boundary layer thickness, then it seems to me using the realizable k-epsilon model with standard wall function is less accurate than using the k-omega model with wall treatment in obtaining the near wall air velocity field. Am I correct on this?

[Rob]

You're conflating accuracy with validity. Depending on the y+ the k-w will be better, if y+ is high then k-e isn't going to be any worse. However, both models assume the film is "thin" as that's what the model is intended for. If the film is "thick" you need the VOF model.

[azhao]

Hi Rob Thank you for the explanation. About the threshold value to determine if the film is "thin" or "thick", shall I compare it to the near wall grid size? If the EWF thickness is larger than the first cell, than it is deemed to be too thick and should be switched to the VOF model? I found that 2022R1 version has introduced the DPM VOF wall film transforming and coupling function.
And I have tried refining the mesh to do the simulation, now the total size is 3 million tetrahedral cells. The computation now takes quite a long time, namely, several days... The following figure is the liquid film thickness result at the 1000 iteration step. It becomes circumferentially more ununiform. Unfortunately, the splashing solution diverged at around 1200+ iterations and the nosplashing solution diverged at around 3000+ iterations due to the following errors.
-------------------------------------------------------------------------------------------------------------------------------------------
turbulent viscosity limited to viscosity ratio of 1.000000e+05 in 912707 cells
12881.8061e+072.3892e-032.6434e-032.1075e-032.5719e+475.6911e-019.9946e-01 41:19:44 49754

Reversed flow on 2654 faces (23.7% area) of pressure-inlet 5.

Reversed flow on 6601 faces (61.4% area) of pressure-outlet 6.

Reversed flow on 13017 faces (79.1% area) of pressure-outlet 8.
Stabilizing temperature to enhance linear solver robustness.
Stabilizing temperature using GMRES to enhance linear solver robustness.

Divergence detected in AMG solver: temperaturesub-iteration: 1 residual - h: 6.496940e+09; u: 5.613071e+12; v: 5.380132e+12; t: 1.304682e+96
sub-iteration: 2 residual - h: 6.727023e+09; u: 4.310764e+16; v: 2.120615e+16; t: 6.735130e+102
sub-iteration: 3 residual - h: 6.726912e+09; u: 3.095855e+12; v: 2.967379e+12; t: 3.476727e+109
sub-iteration: 4 residual - h: 6.726914e+09; u: 4.310764e+16; v: 2.120615e+16; t: 1.794653e+116
sub-iteration: 5 residual - h: 6.726914e+09; u: 3.097524e+12; v: 2.968979e+12; t: 9.263599e+122
sub-iteration: 6 residual - h: 6.726914e+09; u: 4.310764e+16; v: 2.120615e+16; t: 4.781583e+129
sub-iteration: 7 residual - h: 6.726914e+09; u: 3.097524e+12; v: 2.968979e+12; t: 2.468078e+136
sub-iteration: 8 residual - h: 6.726914e+09; u: 4.310764e+16; v: 2.120615e+16; t: 1.273922e+143
sub-iteration: 9 residual - h: 6.726914e+09; u: 3.097524e+12; v: 2.968979e+12; t: 6.575442e+149
sub-iteration: 10 residual - h: 6.726914e+09; u: 4.310764e+16; v: 2.120615e+16; t: 3.393952e+156
Film time = 4.109502e-03 with timestep = 3.0e-11, (max_cfl: 5.777522e+00)
Maximum CFL reached. Reduce time step by factor of 2
turbulent viscosity limited to viscosity ratio of 1.000000e+05 in 790015 cells

Divergence detected in AMG solver: temperature
Error at host: floating point exception

===============Message from the Cortex Process================================

Compute processes interrupted. Processing can be resumed.

==============================================================================