Modelling irregular tube flow
Greetings DualSPHysics users,
I'm trying to simulate 3D irregular tube flow. The results are less than satisfactory. Initially I simulated 2D low Reynolds Poiseuille flow (Re <<1), which had excellent results. In the end I simulated 3D Poiseuille flow (Re=500) which still had excellent agreement with the analytical solution. All of that was done using a straight tube, and the next step was to use a tube which had a bulge (quick expansion and then quickly narrowing again). I have attached 2 images, one shows the full structure and the other one shows the "cut" structure (shortened).
I used my previous .xml file, added in the new structure file and modified the domains a tiny bit. Due to the length of the structure, the previous dp (=0.00005) could not be used (~70 million particles). Dp was changed to 0.0005 to allow an acceptable computation time.
The result was disappointing, too high velocity and a Poiseuille flow at the exit of the pipe was not achieved. I have attached an image showing the profile across the pipe (Vmax is supposed to be 1 m/s, it is ~ 2 m/s)
What I expected to see is a fully developed profile at the exit of the pipe while at the ~centre (At the bulge) it would get more messy (Somewhat turbulent).
Things to note:
- I'm using kinematic viscosity since artificial viscosity provided not good results in the previous steps and kinematic viscosity value will be changed in the future (Extremely desired)
- Dt's were adjusted during the simulation
- 17850 particles out of 850k were removed (99% due to Rhopout)
My question comes down to, what is the best approach to proceed with this problem? My ideas were:
- Increase number of particles even further (At the price of computation time, which is quite limited)
- Try to change some of the parameters in .xml and perhaps use Delta-SPH / Shifting (Unlikely)
- Try using artificial viscosity (Quite undesired)
- Other
If anyone has any comments on my methodology or on how to proceed further, it would be greatly appreciated. If any extra information is required, I can provide it. Additionally, I have uploaded a .txt version of the .xml file.
Kind regards,
Bladomas
Edit: I somehow cannot make the pictures appear in the text, if I add them the text disappears. Hopefully they are still accessible.
I'm trying to simulate 3D irregular tube flow. The results are less than satisfactory. Initially I simulated 2D low Reynolds Poiseuille flow (Re <<1), which had excellent results. In the end I simulated 3D Poiseuille flow (Re=500) which still had excellent agreement with the analytical solution. All of that was done using a straight tube, and the next step was to use a tube which had a bulge (quick expansion and then quickly narrowing again). I have attached 2 images, one shows the full structure and the other one shows the "cut" structure (shortened).
I used my previous .xml file, added in the new structure file and modified the domains a tiny bit. Due to the length of the structure, the previous dp (=0.00005) could not be used (~70 million particles). Dp was changed to 0.0005 to allow an acceptable computation time.
The result was disappointing, too high velocity and a Poiseuille flow at the exit of the pipe was not achieved. I have attached an image showing the profile across the pipe (Vmax is supposed to be 1 m/s, it is ~ 2 m/s)
What I expected to see is a fully developed profile at the exit of the pipe while at the ~centre (At the bulge) it would get more messy (Somewhat turbulent).
Things to note:
- I'm using kinematic viscosity since artificial viscosity provided not good results in the previous steps and kinematic viscosity value will be changed in the future (Extremely desired)
- Dt's were adjusted during the simulation
- 17850 particles out of 850k were removed (99% due to Rhopout)
My question comes down to, what is the best approach to proceed with this problem? My ideas were:
- Increase number of particles even further (At the price of computation time, which is quite limited)
- Try to change some of the parameters in .xml and perhaps use Delta-SPH / Shifting (Unlikely)
- Try using artificial viscosity (Quite undesired)
- Other
If anyone has any comments on my methodology or on how to proceed further, it would be greatly appreciated. If any extra information is required, I can provide it. Additionally, I have uploaded a .txt version of the .xml file.
Kind regards,
Bladomas
Edit: I somehow cannot make the pictures appear in the text, if I add them the text disappears. Hopefully they are still accessible.