Jet impingement on cylindrical obstacles
 

Hi All,

I am new here, and have just started a CFD project for my masters. First step is to model a jet (water into water), at about 250 m/s coming out of a nozzle of 141 micron diameter and then hitting a cylinder, splitting into two jets, which then each hit another cylinder.

I have made the geometry and set the properties in the workbench to 2D analysis type. I am using ansys 14.

In setup, i am using k-epsilon model, with the nozzle as a velocity inlet and the exit as a pressure outlet at 0 Pa. Obstacles are zero-velocity inlets. pressure exit has a monitor on it for x velocity.

I know what the jet distribution pattern should look like as experiments with this configuration have been done. The plan is to make a model that gets the same results as the experiment, and then use the model to predict the behavior of the jets in different configurations.

As i am new to ANSYS, i have a few questions, if anyone could help i would be very grateful :)

1) While solving i keep getting an error like "reversed flow in 112 faces on pressure-outlet 8". I have tried giving the exit some back pressure and moving the exit further away from the obstacles, but i still get this. Do i need to worry? The solution still converges, and pattern is quite similar to real world results.

2) I get good results with my current domain, however making the domain larger with the same mesh, i get bad results, the jet that hits the first obstacle, instead of the two subsequently formed jets hitting obstacles further behind, the jets curve backward and never hit the second obstacles... I also get this if i use my current domain but make the mesh smaller.

3) How do i make sure i have made my 2D model symmetric?

Any tips, ideas on model setting etc... would be great :) Below are two screenshots showing what happens when i make the domain larger... It should look like the top pic but it doesn't and i don't know why.

Thanks guys for any of your help. Looking forward to learning about CFD and becoming part of the community :)

Rob

With current domain:
http://i883.photobucket.com/albums/a...ps45e9e68e.jpg


Making domain larger with same mesh size:
http://i883.photobucket.com/albums/a...ps0ebc185f.jpg

1) Look at the velocity vectors to find where the flow is reversed, and check if it matches your experimental results. If the jet is the only inlet, there will be large scale recirculation of entrained fluid, and it may not be worth making the domain as huge as it would need to be to accurately capture the entire recirculation.

2) You might try opening up the walls upstream of the jet as pressure inlets to avoid the recirculation, then tweak them down to help understand what's going wrong with different meshes and domain sizes.

3) Symmetry is just a boundary condition. Whatever edge is your symmetry plane should be set as 'symmetry' in boundary conditions. In your results, you should see that in the direction perpendicular to the edge both the velocity component and pressure gradient are zero. Speaking of boundary conditions, you may want to use the 'wall' condtion for your obstacles rather than a 0 velocity inlet. At a glance, they may seem the same, but Fluent has specific equations to model near wall effects of turbulence that it won't use unless the wall boundary condition is selected.