Jet Ejector- Convergence

Magma · February 19, 2013, 05:38

Hi all,
I am working with jet ejector, my primary target is to simulate the suction of atmospheric air through water jet.
I have modeled and simulated various models but the problem i am facing is with the convergence. The solution does not converge. I have reduced time-steps too and increased the number of iterations too, even 3000 iterations does not work. Has anyone carried work in similar or same topic, i need help. if anyone has please help me with the detail procedure, i am already frustrated

or can any one give me suggestions why the output is wrong. I have attached the geometry file .parasolid too and the output plots.
if anyone can help please send me a mail at: bhaskar@ku.edu.np

ghorrocks · February 19, 2013, 17:00

FAQ: http://www.cfd-online.com/Wiki/Ansys...gence_criteria

In this case it appears you are running this steady state. It almost certainly is not steady state, you will need a transient simulation to model this correctly.

Magma · February 19, 2013, 22:12

Thanks ghorrocks and as you said, I have been running it in steady state and i was thinking it must be a transient case. Thanks again!!!

imnull · February 20, 2013, 10:14

#1
do not model 3D... go to pseudo 3D (2D and extrude along the axe for 1 cell).

#2
where is an atmosphere domain? you have to modify the geometry - it is MUST.

#3
mesh must be HEX not tet.

#4
be careful with boundary condition.

ghorrocks · February 20, 2013, 16:24

Sorry Dmitry, I disagree with most of your points:
1 - The geometry is not axisymetric so cannot be modelled axisymetric. The inlet is on one side and that will almost certainly make it 3D.
2 - Sounds unlikely to me. The inlet marked as atmosphere does not look as if the current location is a problem. However the outlet boundary is almost certainly too close, it will need to be moved further away.
3 - This can run fine on tet meshes. The most important parameter is good mesh quality. A good tet mesh is better than a poor quality hex mesh.
4 - Of course

imnull · February 20, 2013, 17:03

Quote:

Originally Posted by ghorrocks

Sorry Dmitry, I disagree with most of your points:
1 - The geometry is not axisymetric so cannot be modelled axisymetric. The inlet is on one side and that will almost certainly make it 3D.
2 - Sounds unlikely to me. The inlet marked as atmosphere does not look as if the current location is a problem. However the outlet boundary is almost certainly too close, it will need to be moved further away.
3 - This can run fine on tet meshes. The most important parameter is good mesh quality. A good tet mesh is better than a poor quality hex mesh.
4 - Of course

I disagree with you as well.

1. depends on what plane are you are interested in.. if you interested to see the plane ZY from pic below everything will work just fine.
2. yes I am about outlet boundary.. it must be an atmosphere domain (better use opening instead of outlet). I assume that the pressure drop in this case is BIG!
3. there is no problem to get HEX by blocking approach if #1 will work for you
4. OK

pls share the boundary condition with..

Untitled111.jpg

Magma · February 20, 2013, 21:29

Thanks both!
Actually I have somewhat poor mesh with around 30000 nodes and 150000 elements, I will try to smooth it.
Outlet is in atmospheric condition, so I probably need to keep it as opening. will do so....
The another thing what I find problematic is the streamline of air and water is ok, but when I plot the contour of velocity of either of them, it is so wrong!!! and contour of both is same!!!
Well probably I am having problems in setting Boundary conditions to... and multiphase is a pain!!
I simulated single phase, and the result was ok!!(3D and 2D) but incase of multiphase it seems like 2 fluids are not connecting with each other.

What I meant to ask is: I have a domain( I have modeled it as single domain) do I need to create two sub-domains for water to air interaction and air to water interaction? This is not the case of mass transfer.
Any ideas?

imnull · February 20, 2013, 21:58

Quote:

Originally Posted by Magma

Thanks both!
Actually I have somewhat poor mesh with around 30000 nodes and 150000 elements, I will try to smooth it.
Outlet is in atmospheric condition, so I probably need to keep it as opening. will do so....
The another thing what I find problematic is the streamline of air and water is ok, but when I plot the contour of velocity of either of them, it is so wrong!!! and contour of both is same!!!
Well probably I am having problems in setting Boundary conditions to... and multiphase is a pain!!
I simulated single phase, and the result was ok!!(3D and 2D) but incase of multiphase it seems like 2 fluids are not connecting with each other.

What I meant to ask is: I have a domain( I have modeled it as single domain) do I need to create two sub-domains for water to air interaction and air to water interaction? This is not the case of mass transfer.
Any ideas?

tell us your boundary condition...
+
what is your mesh statistic?

Magma · February 20, 2013, 22:07

Of course!
I have attached a pic with Boundary conditions. The rest is wall condition. I probably think there is no BC that sets up interaction between fluids.

imnull · February 21, 2013, 10:12

// Tutorials :: 0 // 1. Introduction to the ANSYS CFX Tutorials

there are number of tutorials avail

ghorrocks · February 21, 2013, 16:27

Quote:

I probably think there is no BC that sets up interaction between fluids

I do not understand this comment. In a multiphase simulation you define the voluem fraction at each inlet. So for your water inlet it is pure water at 2m/s and the air inlet is pure air at atmospheric pressure, the exit is an opening or outlet at atmospheric pressure. Easy.

Magma · February 21, 2013, 22:48

Quote:

Originally Posted by ghorrocks

I do not understand this comment. In a multiphase simulation you define the voluem fraction at each inlet. So for your water inlet it is pure water at 2m/s and the air inlet is pure air at atmospheric pressure, the exit is an opening or outlet at atmospheric pressure. Easy.

Yes,
I have included volume fraction for inlet conditions, velocity, pressure as they need. The problem is: do I need to set up anything more so that the fluids interact, and this is not interphase mass transfer problem, it's just a suction study.
Well i have streamlines as expected, but the contour of velocity for water and air is same???

what is the problem?
Next thing i wish to ask, for global initiation what would be better: automatic or with value... as in both boundary, each boundary is of single fluid so volume fraction tends to be 1 for each inlet. If i define 0.9 for air( air is all over venturi, and 0.1 for water, it's obvious that the result will be wrong... any suggestions?

imnull · February 21, 2013, 23:08

sorry i am out of this topic...

or upload your project.

ghorrocks · February 22, 2013, 05:05

Boundary conditions are exactly what they say - conditions on the boundary of the domain. They set up the external conditions so the physics you are modelling can do whatever it does inside the domain. So you need boundary conditions which accurately represent reality and also physics which is realistic. You need both for your model to be accurate.

February 20, 2013, 21:29	Jet Ejector-convergence	#7
Magma New Member BhaskarD Join Date: May 2012 Location: Nepal Posts: 22 Rep Power: 13	Thanks both! Actually I have somewhat poor mesh with around 30000 nodes and 150000 elements, I will try to smooth it. Outlet is in atmospheric condition, so I probably need to keep it as opening. will do so.... The another thing what I find problematic is the streamline of air and water is ok, but when I plot the contour of velocity of either of them, it is so wrong!!! and contour of both is same!!! Well probably I am having problems in setting Boundary conditions to... and multiphase is a pain!! I simulated single phase, and the result was ok!!(3D and 2D) but incase of multiphase it seems like 2 fluids are not connecting with each other. What I meant to ask is: I have a domain( I have modeled it as single domain) do I need to create two sub-domains for water to air interaction and air to water interaction? This is not the case of mass transfer. Any ideas?

Thread Tools	Search this Thread
Show Printable Version Email this Page	Search this Thread: Advanced Search
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Jet fan and Tunnel simulation	ahlo7	CFX	9	November 13, 2019 04:54
2D Plane Jet Fluent Solution	doug	Main CFD Forum	11	February 4, 2019 09:34
Jet in Supersonic Crossflow, controlling mass flow rate	ChrisA	OpenFOAM Running, Solving & CFD	3	November 13, 2012 18:20
determination of jet front	Fred	CFX	5	February 11, 2007 12:01
method to determine jet front	Fred	Main CFD Forum	0	February 8, 2007 13:15

February 19, 2013, 17:00		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,696 Rep Power: 143	FAQ: http://www.cfd-online.com/Wiki/Ansys...gence_criteria In this case it appears you are running this steady state. It almost certainly is not steady state, you will need a transient simulation to model this correctly. Magma likes this.

February 19, 2013, 22:12	Jet-ejector convergence	#3
Magma New Member BhaskarD Join Date: May 2012 Location: Nepal Posts: 22 Rep Power: 13	Thanks ghorrocks and as you said, I have been running it in steady state and i was thinking it must be a transient case. Thanks again!!!

February 20, 2013, 10:14		#4
imnull Member Dmitry Join Date: Nov 2011 Posts: 45 Rep Power: 14	#1 do not model 3D... go to pseudo 3D (2D and extrude along the axe for 1 cell). #2 where is an atmosphere domain? you have to modify the geometry - it is MUST. #3 mesh must be HEX not tet. #4 be careful with boundary condition. Last edited by imnull; February 20, 2013 at 14:41.

February 20, 2013, 16:24		#5
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,696 Rep Power: 143	Sorry Dmitry, I disagree with most of your points: 1 - The geometry is not axisymetric so cannot be modelled axisymetric. The inlet is on one side and that will almost certainly make it 3D. 2 - Sounds unlikely to me. The inlet marked as atmosphere does not look as if the current location is a problem. However the outlet boundary is almost certainly too close, it will need to be moved further away. 3 - This can run fine on tet meshes. The most important parameter is good mesh quality. A good tet mesh is better than a poor quality hex mesh. 4 - Of course

February 21, 2013, 10:12		#10
imnull Member Dmitry Join Date: Nov 2011 Posts: 45 Rep Power: 14	// Tutorials :: 0 // 1. Introduction to the ANSYS CFX Tutorials there are number of tutorials avail

February 21, 2013, 23:08		#13
imnull Member Dmitry Join Date: Nov 2011 Posts: 45 Rep Power: 14	sorry i am out of this topic... or upload your project.

February 22, 2013, 05:05		#14
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,696 Rep Power: 143	Boundary conditions are exactly what they say - conditions on the boundary of the domain. They set up the external conditions so the physics you are modelling can do whatever it does inside the domain. So you need boundary conditions which accurately represent reality and also physics which is realistic. You need both for your model to be accurate.