Open boundary "imposevelocity" problem

Hello

We are modeling the entrance chamber to a sewage pumping station (on a reduced scale 1 in 10)

The model has a single entry, defined as a boundary condition with imposed and fixed velocity (imposevelocity mode=”0”)

It has 6 outputs (one for each active pump) that are represented as free discharges to the atmosphere

My problem is that inlet velocity, and therefore the flow, does not remain constant throughout the simulation but decreases

Below is the velocity as a function of time at several points of a cross section of the inlet pipe, sufficiently far from the boundary condition

Unlike the solved examples of the library, in this case the boundary condition is completely submerged and the water level in the chamber is higher.

How can I solve this unwanted variation in the inlet velocity, which should be fixed?

I attach files so you can check

Thank you!


Comments

  • Hi (?)

    Your inlet is leaking through the bottom of the buffer. You have to make sure that the inlet buffer covers the entire inner area of the channel, the way I see it from your XML, there is a row of buffer particles missing.

    Also, you want to make sure that strict inlet behavior is enforced by using "false" for convertfluid.

    There are other things you can do to make the simulation go smoother, but try these first and see how it goes.

    Angelo Tafuni

  •  Angelo

    Thank you very much for your answer! It was very useful

    I did what you said:

    - I could not get the buffer to cover the entire entrance section because the liquid particles could not be so close to the solid edges. But I added a lid on the back of the entrance channel to close it and avoid leaking

    - I used “false” to convert fluid

    With these modifications, the velocities in the same cross section as before are the following:

    We can see that velocities have important variations (like pulses) but their average value is maintained around the expected value (0.42 m / s) which is an improvement over the previous condition.

    If points are measures much further from the edge condition, fluctuations are also observed but less pronounced

    Do you have any suggestion?

    Greetings and thanks

    Pablo Chiesa

  • Hi Pablo,

    You should be able to fill the channel with a buffer as wide and tall as the channel dimensions, I do not understand why you can't. If you are using the GUI, that may be the reason why. Try editing the XML file directly.

    Another suggestion I can give you is to avoid the empty area between the buffer and fluid. Create a buffer that is adjacent to the fluid and initialize the buffer and fluid with the same velocity, to avoid strong velocity gradients in the vicinity of an open boundary.

    Fixing these two things should improve the results.

    You also have wall effects that you are not considering by imposing a uniform velocity at the inlet, but this is something less straightforward to fix, and would require minor code implementations.

    Angelo

  • Angelo

    Again, thank you very much for your collaboration.

    I made the calculation again according to the suggestions of your previous post, but I still don't get good results.

    The pulsations in the velocity continue being very pronounced

    On the other hand, near the open boundary buffer there are some liquid particles that pass through the solid and escape. It should not happen, because in that region the solid is closed (I don't know if this influences the results or not)


    Throughout the simulation, speed in the inlet duct is quite chaotic, and upon reaching the first chamber, the flow direction is not as expected (adheres to the back wall instead of following the most direct path)


    I reply to your previous email with some comments. Maybe you have any other suggestions


    “You should be able to fill the channel with a buffer as wide and tall as the channel dimensions, I do not understand why you can't.”

    I cannot fully occupy the section because the following error aborts the execution

    I must leave a small separation between buffer and solids, approximately equal to 1 dp (Initial inter-particle distance)


    “If you are using the GUI, that may be the reason why. Try editing the XML file directly.”

    Yes. I edited the XML file


    “Another suggestion I can give you is to avoid the empty area between the buffer and fluid. Create a buffer that is adjacent to the fluid and initialize the buffer and fluid with the same velocity, to avoid strong velocity gradients in the vicinity of an open boundary.”

    Thank you. I did it.

    I avoided the empty spaces between the fluid and the buffer and imposed an initial velocity on the fluid equal to the speed of the buffer


    “You also have wall effects that you are not considering by imposing a uniform velocity at the inlet, but this is something less straightforward to fix, and would require minor code implementations.”

    I didn´t do this. I kept the velocity uniform


    Regards

    Pablo


Sign In or Register to comment.