CFD Online Discussion Forums

CFD Online Discussion Forums (
-   CFX (
-   -   Filling a tank or pipe (

stu May 22, 2007 19:06

Filling a tank or pipe
A simple scenario, I wish to fill a tank or pipe with water, say I have a pipe that is 10 cm diam, 100 cm long that is standing on its end. I have the following BC's 1 inlet, 1 wall, 1 outlet or opening.

Initially the pipe is filled with air and then I start to fil it from the inlet at the bottom at say 1 cm/sec. How do I setup the physics of this simply problem?

I know that it is a multiphase problem and I use volume fractions, initially the domain is 100% air and 0% water. The inlet BC becomes water at 1cm/sec at 100% vf and air at 0 cm/s 0% vf. The outlet or opening only lets air out, air 100% vf, water 0% vf. And I only have the 1 domain. I can use the KE turb model. I have setup the two fluids as continuous fluids, not dispersed fluids.

What else and I missing?

I want to be able to view the density change as the water moves up the pipe, simple plane with density contour plotted.

For some reason I am having problems trying to get this example to work.

Transient analysis total time 10 sec, time steps 0.1 secs. High resolution (double?) backward euler.

Could someone give me guidance or post a the text based input file that gives the physics of this kind of problem.

Thanks Stu

Gert-Jan May 23, 2007 08:23

Re: Filling a tank or pipe

1) What is your initial guess?

2) I would start with laminar.

3) How does your grid look like? I hope you are using a hex mesh. If not you will get a lot of smearing due to tet elements that are not parallel with your free surface.

4) do you use the free surface sharpening?

5) I would not use two continuous phases. With 1 cm/s mixing is not likely, the way you have set this up.


stu May 23, 2007 21:20

Re: Filling a tank or pipe
Thanks for your response

In a further problem I am doing, very similar but different shape ( it is a tank) I am using laminar.

Point 5, " I would not use two continuous phases. With 1 cm/s mixing is not likely, the way you have set this up." If I basically want to view the pipe filling up with water and 'pumping / pushing' the air out, I thought this would be 2 phase. If the tank is initially filled with air vf=1, and then water (vf=1) is pumped in, how can I do this without using 2 continuous fluids? I might be missing something, please feel to point me in the right direction.

Thanks Stu

Gert-Jan May 24, 2007 18:23

Re: Filling a tank or pipe
I mean that you should use the homogeneous model (with free surface model). Then you will have one velocity field and a massfraction for water. This implies less equation and less run time. Indeed both phases are continuous.

I meant that you shouldn't use a heterogeneous model with two continuous phases for this problem as the phases will be completely demixed.


All times are GMT -4. The time now is 05:00.