You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: _tutorials/incompressible_flow/Inc_Turbulent_Bend/Inc_Turbulent_Bend.md
+28-22Lines changed: 28 additions & 22 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -12,9 +12,9 @@ complexity: intermediate
12
12
follows: Inc_Turbulent_Flat_Plate
13
13
---
14
14
15
-
16
-
15
+
17
16
17
+
Figure (1): impression of the 90 degree bend with velocity contours.
18
18
19
19
## Goals
20
20
@@ -46,9 +46,9 @@ The configuration is a 90 degree bend with a diameter of D=0.104 m and a radius
46
46
47
47
The mesh consists of a structured mesh with 70k cells and 75k points. The mesh was created using Gmsh. The configuration file to create the mesh is ([sudo.geo](https://github.com/su2code/Tutorials/tree/master/incompressible_flow/Inc_Turbulent_Bend_Wallfunctions/sudo.geo)) and the only thing you need to do to create a mesh from the geometry is start Gmsh, and then load the .geo file. You will then see the geometry in the Gmsh visualization window. If you click on *Mesh->3D* the 3D mesh will be generated. You can then export the mesh as a .su2 file by choosing *File->Export*. The mesh will automatically be saved in su2 format when the filename has the extension .su2. In general, you should not choose *save all elements* because this will also save additional points that were used to construct the geometry but are not part of the final mesh, like for example the center of a circle.
Figure 2: Mesh generated by Gmsh, showing the cross-sectional block structured mesh.
51
+
Figure (2): Mesh generated by Gmsh, showing the cross-sectional block structured mesh.
52
52
53
53
### Configuration File Options
54
54
@@ -108,8 +108,8 @@ Your results should converge in around 1000 iterations.
108
108
109
109
### Results
110
110
111
-
112
-
Figure 3: Velocity contour with location of the slices where experimental values were obtained.
111
+
112
+
Figure (3): Velocity contour with location of the slices where experimental values were obtained.
113
113
114
114
The paraview multiblock file can now be visualized, and the result is shown in Figure (3). Shown is the velocity magnitude on the horizontal symmetry plane together with a number of contour plots at several vertical locations in the bend.
115
115
@@ -118,12 +118,20 @@ The paraview multiblock file can now be visualized, and the result is shown in F
118
118
In the experiment, the pipe bend was preceded by a very long straight pipe section with a length of 100D, ensuring that the turbulent flow is fully developed by the time it arrives at the pipe bend. In our simulations, we have only a straight section with a total length of 5D and constant inlet properties. We will now restart the simulation with an inlet profile that we extract from a slice of the pipe. It is a planar slice with the normal in the Z-direction, located at Z=2.5D from the inlet. Because in a multiblock mesh, nodes are duplicated on the boundary, so we first perform an *Extract Block* and select only the interior nodes. We then apply the slice on this extracted block. To keep the original quadrilateral cells, we de-select the option *triangulate slice*. Also make sure that the option *Merge duplicated points in the slice* is selected. In the information tab, you should see that the number points is 618, and we will check later that this corresponds with the number of nodes that SU2 expects in the inlet boundary file.
119
119
Now save the slice data, making sure that only the slice is selected in the Pipeline Browser. Choose the .csv file format, and choose as _Arrays To Write_ only the variables *Omega*, *Pressure*, *Velocity* and *Turb_Kin_Energy*. The point coordinates are automatically saved. Also choose scientific notation with precision "6" to make sure that you have enough digits for the accurate interpolation of the inlet profile. A simple pvpython script is provided in the tutorials folder. This is a python script that you can run with paraview's pvpython command and it will generate an inlet.csv file for you :
120
120
121
-
```$ pvpython paraview_extract_slice_data.py```
121
+
```
122
+
$ pvpython paraview_extract_slice_data.py
123
+
```
124
+
125
+
122
126
123
127
With python, matlab or a simple *awk* command it is easy to put the data in the correct format for the SU2 inlet profile. The awk command below reads the file *inlet.csv*, it assumes that the comma acts as a separator, it skips the first line (the header), and then prints the necessary columns and outputs it in the file inlet_test. We want symmetric data in the y-axis, so we check if the y-coordinate in column 2 is positive, and then we write this data for positive and negative y-coordinates. We write the data into the file *inlet_test*.
The solution was saved as *restart.dat* and the filename used for the restart is named *solution.dat* (locate the option for the filenames in the config file!). You can simply copy the *restart.dat* file to *solution.dat* and restart the simulation with the above options. If you restart and the file *inlet.dat* is not present, SU2 will stop with the message
136
146
137
-
```
138
-
Looked for: inlet.dat.
139
-
Created a template profile file with default values named example_inlet.dat
140
-
You can use this file as a guide for making your own profile specification.
141
-
```
147
+
``` Looked for: inlet.dat.```
148
+
```Created a template profile file with default values named example_inlet.dat```
149
+
```You can use this file as a guide for making your own profile specification.```
142
150
143
151
You will now have an example inlet profile saved in the file *example_inlet.dat*. You can now check that this file has 618 rows of data, for each of the nodes on the inlet plane. We only need the header (the first 5 lines of the file), so you can open the file, copy the first 5 lines, open the file *inlet_data* and add the 5 lines to the top of the file. You can now save this file as *inlet.dat* and restart the simulation.
144
152
145
153
### Results
146
154
147
155
We will now compare the simulation results of the pipe bend with results from the paper of *Sudo et al.* Some velocity measurements were performed at several locations in the pipe bend, and the location that we will use for comparison are shown in the figure below.
148
156
149
-
157
+
150
158
151
-
152
-
153
-
Figure 4: Pipe bend showing the location of the slices that will be used for comparison with measurements.
159
+
Figure (4): Pipe bend showing the location of the slices that will be used for comparison with measurements.
154
160
155
161
Below are some of the contour plots taken at several locations in the pipe, showing the velocity normal to the plane with the isocontours in black. The top half are the simulation results and the bottom half are the experimental results from *Sudo et al (1998)*. We see that the main flow features are captured in the simulation.
156
162
163
+
157
164
165
+
158
166
159
-
160
-
161
-
167
+
Figure (5): mean flow velocities in several pipe cross sections.
162
168
163
-
169
+
Below is a comparison of the axial velocity on the horizontal symmetry line through the vertical plane at $$\phi=90^0$$. The flow separation in the bend is difficult to capture correctly, and it leads to an underestimation of the local velocity near the inner radius of the bend.
164
170
165
-
Below is a comparison of the axial velocity on the horizontal symmetry line through the vertical plane at $\phi=90$. The flow separation in the bend is difficult to capture correctly, and it leads to an underestimation of the local velocity near the inner radius of the bend.
171
+
166
172
167
-
173
+
Figure (6): velocity on the symmetry line in the vertical planar cross-section at $$\phi=90^0$$.
0 commit comments