T-junktion - problem at corners
 

Hi all,

I am simulating a T-junktion, droplet formation simulation. My main channel is filled with water and the two side channels with oil, so that droplet formation occurs. I am using Gambit to create a structural mesh (around 100000 elements). Everything seemed ok at the begining when I noticed that at the corners of the geometry, where the two fluids mix, an unuasual gradient of the velocity occurs. The velocity at the corners is 5 times bigger than the average. I thing this is a numerical error of some sort and it affects the physical results ... I tried to work with unstructured mesh, the problem dissapeared but the physical properties of the droplets also change in an unwanted way. I tried to refine the mesh but this did not change anything. I am thinking the problem comes from the surface tension at this corner faces where the two fluids mix, the solver is probably just confused at this point. So my idea (and also question) is that somehow I turn off the surface tension for the mesh at the corners. Is this possible and does someone have experince with this situation. Thanks!

Martinchhttp://file:///C:/Users/49157/Tjunktion100000PART.png