[hwangpo]

Hey, guys
I am simulating a simple model(see attached figs).
But the results are far from reality.

Inlet: opening B.C. with static pressure(the value=rho*g*h where h=60).
Outlet: relative pressure=0
Others: wall B.C..

I can not figure out why there is so large negative pressure at the vicinity of branch pipe.

I really appreciate your help.

[Blanco]

Hi,what is the numerical setup? and what about turbulence model, mesh in the branch pipe? you are obtaining 40 m/s maximum velocity in the connection, I think you have boundary layer separation there, that s why you get negative pressure...

[ghorrocks]

The low pressure region is clearly a separation.

I can also see your contour lines are jagged which is a sure sign of your mesh being too coarse. Have you read this FAQ? http://www.cfd-online.com/Wiki/Ansys..._inaccurate.3F

[sarab]

Dear you dint not supply full information. Its simulation shows that your mesh is unstructured and not adequate. what is your turbulence model? working fluid and reference pressure. Plot Absolute pressure instead of pressure to get a clear picture.

[hwangpo]

Originally Posted by Blanco

Hi,what is the numerical setup? and what about turbulence model, mesh in the branch pipe? you are obtaining 40 m/s maximum velocity in the connection, I think you have boundary layer separation there, that s why you get negative pressure...

Thank you.
k-e model and unstructured mesh.
If there is boundary layer separation, what can I do next to fix this?
refine mesh elements or something else?

[hwangpo]

Originally Posted by ghorrocks

The low pressure region is clearly a separation.

I can also see your contour lines are jagged which is a sure sign of your mesh being too coarse. Have you read this FAQ? http://www.cfd-online.com/Wiki/Ansys..._inaccurate.3F

Except mesh refinement, anything else? I did some tests and the problem remains unsolved.
It is not right because there will be cavitation. How can I improve it or redo the simulation the right way?

I really look into that FAQ before and I cannot find out a way.

thank you for your help.

[Blanco]

Ok, yes you have to refine the mesh in the separation zone and just ahead/after it, to improve the definition of the physical domain and so improve the solution. Btw, what is the minimum absolute pressure reached in the separating zone? I suppose negative pressure you talked about was relative pressure, take a look at abs values. You talked also about cavitation, if you expect cavitation than you need to have a model for it.
Regards

[ghorrocks]

If you are using a reference pressure of 1 bar and your working fluid is water then you are miles away from cavitation.

And if the simulation is accurate and it predicts a separation - then the system you are modelling has a separation, so your simulation is correct. Or do you know that a separation does not exist?

[hwangpo]

Originally Posted by ghorrocks

If you are using a reference pressure of 1 bar and your working fluid is water then you are miles away from cavitation.

And if the simulation is accurate and it predicts a separation - then the system you are modelling has a separation, so your simulation is correct. Or do you know that a separation does not exist?

Thank you so much for your suggestion.
In this model, the reference pressure is 1 bar, but I used '9.8 kPa' to plot the pressure contour(see the figure above). The negative pressure is about 55*9.8 kPa(water head=55m if this is right).

We have the physical model in fact but cannot observe the flow because that part is inside a wall of the physical model, so we simulate it to see what happens. So I cannot be sure there is separation or not, but I think there is.

any suggestions?

[ghorrocks]

Well if you are trying to say that the pressure in that region = 101-55*9.8 = -437kPa absolute then it will (almost) definitely cavitate. It appears like you have not run a cavitation model so then all this model tells you is that it cavitates - the actual flow and pressure field it predicts is wrong.

[hwangpo]

Originally Posted by Blanco

Ok, yes you have to refine the mesh in the separation zone and just ahead/after it, to improve the definition of the physical domain and so improve the solution. Btw, what is the minimum absolute pressure reached in the separating zone? I suppose negative pressure you talked about was relative pressure, take a look at abs values. You talked also about cavitation, if you expect cavitation than you need to have a model for it.
Regards

OK. Many thanks

[hwangpo]

Originally Posted by ghorrocks

Well if you are trying to say that the pressure in that region = 101-55*9.8 = -437kPa absolute then it will (almost) definitely cavitate. It appears like you have not run a cavitation model so then all this model tells you is that it cavitates - the actual flow and pressure field it predicts is wrong.

What if there is no cavitation? The boundary-layer separation is responsible for the negative pressure or something else? how to deal with it?

Thank you so much.

[Blanco]

Well, if pressure is -437 kPa then it almost definitely cavitate as it has been said. The boundary layer separation will be present also in the real Fluid behavior, if everything is correct in the setup, and it will cause the pressure decrease that lead to cavitation, but if you don t model cavitation you can't refer to the pressure and velocity field obtained...you just know that cavitation will happen there, nothing more.

[Blanco]

Wait...just to be sure I understood right...in your plot I see ref. Pressure 9.8 kPa and then -55 Pa in the figure...so you get 9.75 kPa minum absolute pressure, not -437 kPa, is this correct? if this is correct then you simply have to check if cavitation would happen at 9750 Pa

[hwangpo]

Originally Posted by Blanco

Wait...just to be sure I understood right...in your plot I see ref. Pressure 9.8 kPa and then -55 Pa in the figure...so you get 9.75 kPa minum absolute pressure, not -437 kPa, is this correct? if this is correct then you simply have to check if cavitation would happen at 9750 Pa

Thank you for your concern. The Ref. Pressure is 1 bar for the modeling, but I use a different unit(9.8 kPa, which is not default) to plot the pressure contour.
I think maybe this cause the confusion.

I am going to refine the mesh first and see what will happen.
Thank you so much.

[hwangpo]

hey there
I refined the mesh and the problem is still there.
any suggestions?

Thank you in advance.

[hwangpo]

Originally Posted by Blanco

Well, if pressure is -437 kPa then it almost definitely cavitate as it has been said. The boundary layer separation will be present also in the real Fluid behavior, if everything is correct in the setup, and it will cause the pressure decrease that lead to cavitation, but if you don t model cavitation you can't refer to the pressure and velocity field obtained...you just know that cavitation will happen there, nothing more.

see the updates plz.
any suggestions about the boundary-layer separation?
Or is this reasonable?

thank you for your time.

[Blanco]

Ok thanks for the clarification. The separation is ok, it is physical because of the low pressure, but seeing the results you really need a cavitation model because cavitation will happen. Actual results therefore are just "indicative " because they don't include cavitation. Btw are you getting convergence in this run? what is the order of your residuals?

[hwangpo]

Originally Posted by Blanco

Ok thanks for the clarification. The separation is ok, it is physical because of the low pressure, but seeing the results you really need a cavitation model because cavitation will happen. Actual results therefore are just "indicative " because they don't include cavitation. Btw are you getting convergence in this run? what is the order of your residuals?

1e-04
it is converged after about 40 steps for the scenario with all three inlets opening(the exact scenario we just talk about before).
see the fig plz.

[Blanco]

Ok for the convergence, btw it s really fast. Is the velocity you get at the outlet consistent with your experimental results?