2D and 3D results completely different

ameyadurve · October 18, 2012, 02:30

I am simulating a Jet Loop Reactor.
The tank dimensions are: 1m height and 1 m diameter. The jet diameter is 0.0156 m and is located 0.5 m from the tank bottom. The jet velocity is 12.5 m/s. Fluid is water.

My problem is, when I float the simulations in 2D (k-epsilon Model) I can see the jet impinging on the tank bottom and recirculation zones created in the tank. However, when I float the simulation in 3D using same solution strategy, the jet loses all its momentum the moment it hits the bottom and there are no recirculation zones in the tank.

I do not understand why such a discrepancy exists in the 2D and 3D simulations. I can understand if there is difference in the accuracy of predictions but getting totally different flow patterns is not expected.

Can anyone help me understand why this is happening ?

Please note:
1) The mesh is not an issue. I have a best possible mesh for the domain. Skewness less than 0.4

2) Experimental observations show that 2D results represent the system more accurately

Marion · October 18, 2012, 04:39

Hi,
Are you talking about 2D-Axisymmetric? Is your jet axis the same axis as the tank?
Are you sure that the geometry in 3D is the same as in 2D?
I've done the comparison previously, and between a 2D-axi and a 3D (just revolving the 2D surface around the axis) I had no difference at all. That is why I think there may be an issue with the set up of one of the models.
Marion.

ameyadurve · October 18, 2012, 06:00

Hello Marion,

It is 2D and not 2D axisymmetric. The geometries are exactly the same, as mentioned above. I have checked and double checked that part.

.

Have you done the comparison for turbulent jets ???

Regards
Ameya

Marion · October 18, 2012, 07:29

It's not jets, but I work with high pressure fluids, cavitation, turbulence.

Why are you modelling a cylindrical tank with 2D and not 2D-axi?

Standard 2D is used for when you have 1 of the dimensions very large compared to the others. 2D-axi is used when your geometry and boundary conditions hava an axis of revolution.

Therefore if you are modelling in standard 2D it is normal that it won't correlate with a 3D model of a cylinder.

ameyadurve · October 18, 2012, 08:38

Well, the tank has an outlet situated at 0.5 m from the bottom and hence 2D axisymmetry is not an option.

Thanks for the input regarding use of 2D flows.

I will now simulate a similar system, in which I have some good quantitative experimental data, in both 2D and 3D. I will keep you posted about the results of the study

Thanks again

Marion · October 18, 2012, 08:58

I understand the geometry better now (I think!).
In my opinion, you can neither use standard 2D nor axisymmetric 2D, both are wrong since you want to model a diameter of 0.0156 m for the opening in a cylindrical tank.

October 18, 2012, 02:30	2D and 3D results completely different	#1
ameyadurve New Member Ameya Durve Join Date: Jun 2009 Location: Mumbai Posts: 10 Rep Power: 16	I am simulating a Jet Loop Reactor. The tank dimensions are: 1m height and 1 m diameter. The jet diameter is 0.0156 m and is located 0.5 m from the tank bottom. The jet velocity is 12.5 m/s. Fluid is water. My problem is, when I float the simulations in 2D (k-epsilon Model) I can see the jet impinging on the tank bottom and recirculation zones created in the tank. However, when I float the simulation in 3D using same solution strategy, the jet loses all its momentum the moment it hits the bottom and there are no recirculation zones in the tank. I do not understand why such a discrepancy exists in the 2D and 3D simulations. I can understand if there is difference in the accuracy of predictions but getting totally different flow patterns is not expected. Can anyone help me understand why this is happening ? Please note: 1) The mesh is not an issue. I have a best possible mesh for the domain. Skewness less than 0.4 2) Experimental observations show that 2D results represent the system more accurately

October 18, 2012, 04:39		#2
Marion Senior Member Marion Join Date: Jul 2012 Location: France Posts: 122 Rep Power: 14	Hi, Are you talking about 2D-Axisymmetric? Is your jet axis the same axis as the tank? Are you sure that the geometry in 3D is the same as in 2D? I've done the comparison previously, and between a 2D-axi and a 3D (just revolving the 2D surface around the axis) I had no difference at all. That is why I think there may be an issue with the set up of one of the models. Marion.

October 18, 2012, 06:00		#3
ameyadurve New Member Ameya Durve Join Date: Jun 2009 Location: Mumbai Posts: 10 Rep Power: 16	Hello Marion, It is 2D and not 2D axisymmetric. The geometries are exactly the same, as mentioned above. I have checked and double checked that part. . Have you done the comparison for turbulent jets ??? Regards Ameya

October 18, 2012, 07:29		#4
Marion Senior Member Marion Join Date: Jul 2012 Location: France Posts: 122 Rep Power: 14	It's not jets, but I work with high pressure fluids, cavitation, turbulence. Why are you modelling a cylindrical tank with 2D and not 2D-axi? Standard 2D is used for when you have 1 of the dimensions very large compared to the others. 2D-axi is used when your geometry and boundary conditions hava an axis of revolution. Therefore if you are modelling in standard 2D it is normal that it won't correlate with a 3D model of a cylinder.

October 18, 2012, 08:38		#5
ameyadurve New Member Ameya Durve Join Date: Jun 2009 Location: Mumbai Posts: 10 Rep Power: 16	Well, the tank has an outlet situated at 0.5 m from the bottom and hence 2D axisymmetry is not an option. Thanks for the input regarding use of 2D flows. I will now simulate a similar system, in which I have some good quantitative experimental data, in both 2D and 3D. I will keep you posted about the results of the study Thanks again

October 18, 2012, 08:58		#6
Marion Senior Member Marion Join Date: Jul 2012 Location: France Posts: 122 Rep Power: 14	I understand the geometry better now (I think!). In my opinion, you can neither use standard 2D nor axisymmetric 2D, both are wrong since you want to model a diameter of 0.0156 m for the opening in a cylindrical tank.