[Christian Steiner]

Hi folks,

i am currently simulating the flow through an L-shaped pipe. The fluid enters at the horizontal inlet at a radius r_0 and leaves the pipe at the vertical outlet at a radius r_5 which is bigger than r_0.

The x-axis (horizontal axis) of the pipe is tilted by 11° around the radial axis, so it doesn't match the global/absolute x-axis.

I start my calculations with a no rotation around the global x-axis.

I tried:

Pressure Inlet: Total pressure and gauge pressure given, for the first iterations no direction vector is specified

Pressure Outlet: A pressure smaller than the total pressure at the inlet is assumed, no backflow direction is specified

I use a pressure based solver and run time-steady calculations.
For turbulence i use the k-e modell with enhanced wall functions.

For the solution methods, i picked the SIMPLE scheme, and use second-order upwind for all discretization schemes.

I found out that a hybrid solution initialization works best and i tried very low URF (<0.1) and moderate URF (~0.3) for my calculations.

Problem

The problem is, that i always got backflow at the pressure outlet, so we attached a cylindrical volume with diameter and height in the dimension of 20 times the pipe radius.
This seems to fix the backflow at the outlet for me.
BUT: i still get backflow at my inlet, which according to my supervisor cannot be. So we made a compromise and set the direction at the inlet to normal to boundary at all times. For a rotational speed of 0rpm, this works fine, but the moment i raise the rotational speed above 300rpm (and also the total pressure), we experience backflow which doesn't go away (after more than 5000 iterations).
This happens for incompressible and compressible flow.


Do you guys have any ideas?

Thank you in advance!


Best Regards

Christian

[Gkchpa]

Originally Posted by Christian Steiner

Hi folks,

i am currently simulating the flow through an L-shaped pipe. The fluid enters at the horizontal inlet at a radius r_0 and leaves the pipe at the vertical outlet at a radius r_5 which is bigger than r_0.

The x-axis (horizontal axis) of the pipe is tilted by 11° around the radial axis, so it doesn't match the global/absolute x-axis.

I start my calculations with a no rotation around the global x-axis.

I tried:

Pressure Inlet: Total pressure and gauge pressure given, for the first iterations no direction vector is specified

Pressure Outlet: A pressure smaller than the total pressure at the inlet is assumed, no backflow direction is specified

I use a pressure based solver and run time-steady calculations.
For turbulence i use the k-e modell with enhanced wall functions.

For the solution methods, i picked the SIMPLE scheme, and use second-order upwind for all discretization schemes.

I found out that a hybrid solution initialization works best and i tried very low URF (<0.1) and moderate URF (~0.3) for my calculations.

Problem

The problem is, that i always got backflow at the pressure outlet, so we attached a cylindrical volume with diameter and height in the dimension of 20 times the pipe radius.
This seems to fix the backflow at the outlet for me.
BUT: i still get backflow at my inlet, which according to my supervisor cannot be. So we made a compromise and set the direction at the inlet to normal to boundary at all times. For a rotational speed of 0rpm, this works fine, but the moment i raise the rotational speed above 300rpm (and also the total pressure), we experience backflow which doesn't go away (after more than 5000 iterations).
This happens for incompressible and compressible flow.


Do you guys have any ideas?

Thank you in advance!


Best Regards

Christian

Hi Christian, I assume you solved the Issue, If so, I have the similar problem now , Can you tell me how did you fixed that issue ?