PowerPoint-presentasjon

Comparison of Mørtsell’s and Cepuritis’ results
with Computational Fluid Dynamics simulations
Flow-3D software
Mørtsell vs Flow-3D
Comparison of flow rates for distilled water at 20.5 𝒐𝑪
1.8
•
Flow rates in general show similar
pattern
•
Accurate FlowCyl geometry is critical
•
Steeper curve can be a result of the
computer model set-up (e.g. coarse vs
fine mesh)
•
What about matrix modeling?
1.6
1.4
MASS LOSS (KG)
1.2
1
0.8
0.6
0.4
0.2
0
0
2
4
6
8
10
12
14
16
TIME (SEC)
Mørtsell
2
Flow-3D - New geometry
Flow-3D - Old geometry
18
Cepuritis vs Flow-3D
Comparison of λQ for four mixes
3
Flow resistance λQ
ρ
τ
Mix no.
2.5
Cepuritis Flow-3D Deviation [kg/dm3] [Pa]
MASS LOSS (KG)
2
1.5
1
0.5
0
0
10
20
30
40
50
60
70
TIME (SEC)
Mixno 1-2 - Coarse mesh
3
µ
Mixno 2
Mixno 3
Mixno 4
Mixno 5
80
1-2
2
3
4
5
5-2
5-3
5-4
5-5
6
7
8
9
10
0.535
0.562
0.475
0.653
0.524
0.688
0.699
0.613
0.744
0.68
22 %
20 %
22 %
12 %
23 %
1.957
1.945
1.936
1.992
10.1
10.7
7.0
10.7
[Pas]
0.28
0.28
0.19
0.38
Coarse vs Fine mesh in Flow-3D
Influence on mix no. 1-2
0.45
0.3
MASS LOSS (KG)
Fine mesh
0
97.7
203.3
306.3
406.9
0
1
2
3
4
0.35
0.25
Coarse mesh
0
103.3
207.4
309
407.8
0.2
0.15
Mix no.
Type of mesh
0.1
1-2
1-2
Coarse
Fine
0.05
0
1
2
3
TIME (SEC)
Mixno 1-2 - Coarse mesh
Mixno 1-2 - Fine mesh
Flow resistance λQ
Cepuritis
Flow-3D
Deviation
0.535
0.688
22 %
0.535
?
?
•
Finer mesh should indicate possibility for
more accurate results
•
However, this will result in a higher
demand of computer-hours
0
4
Mass loss (g)
Time (s)
0.4
4
Bingham vs Herschel-Bulkley models in Flow-3D
Influence on mix no. 1-2
3
2.5
Flow resistance λQ
Mix no.
MASS LOSS (KG)
2
Model type
1-2
Bingham
1-2
Herschel-Bulkley
1.5
Cepuritis
Flow-3D
Deviation
0.535
0.688
22 %
0.535
0.620
14 %
1
Mixno 1-2 - Coarse mesh - Bingham
0.5
Mixno 1-2 - Fine mesh - Bingham
Mixno 1-2 - Herschel-Bulkley
0
0
10
20
30
TIME (SEC)
5
40
50
60
Conclusion
6
•
Raw input parameters (instead of 𝜌, 𝜇 and 𝜏) should be used for the computer model
(Herschel-Bulkley instead of Bingham material parameters)
•
Accurate FlowCyl geometry is critical
•
Deviation of 12-22% from physical experiments
•
Possibility for more accurate results when using finer mesh in computer model
•
More simulations needs to be finalized in order to evaluate the performance of the
computer model
Q&A
7
FlowCyl geometry comparison
Mørtsell’s PhD vs ’95 Paper
8
Interpreting Mørtsell’s plots
WebPlotDigitizer
9
1.6
1.4
1.2
µ [Pas]
1.0
0.8
µ vs τ0 (below 20Pa)
Linear (µ vs τ0 (below 20Pa))
0.6
0.4
0.2
0.0
0
2
4
6
8
10
τ0 [Pa]
10
12
14
16
18