Hi everyone,

I hope someone can help me.

Now I am modeling particles flow by using DPM. I want to inject particles from nodes of a curved surface.

So what I did was as follow;

First, I went to gambit and got the x,y,z location of the nodes on the curved surface by using summary of face mesh. Second, I created particle injections. I wrote a file for injections. I gave x,y,z locations which are obtained above to each particle in this file.

This file are imported in fluent, but the problem is that fluent are not able to locate some injections. Actually, some injections are located succesfully. Why is this happening? Some nodes which are existing in gambit are missing in fluent??

Please help me... I am sorry for my bad English.

Thanks, AKR

Which fluent version do you use? I think I can remember in version 6.1 I had the same Problem. If it´s possible try to use the surface as injection-type. You can choose that in the injection panel instead of file.

Usually when you have this type of problem, the initial location of the particles is outside of the domain. Double (or triple check) this.

Thanks.

I will try surface injection. I want to ask to you more.

I have done to solve the gas velocity distribution in domain. So now I want to inject some particles in the velocity distribution. If I use surface injection, particles are injected from the centers of the cells?? If I want to inject many particles from one cell, how can I do that? I am thinking about grid adaption. But I am not sure. Is it OK if I use grid adaption to make one cell smaller after the calculation of gas velocity distribution, and then inject particles? Or if you have any other ideas, could you let me know please?

Regards, AKR

For injecting many particles from one cell u can use pressure swirl atomizer type injection.

Hi,

I have also seen this problem. It can occur sometimes when the x,y,z, location defined in the injection file is numerically equal to a boundary. The injection point has to be inside the domain and I find if I add a small amount (sometimes as little as 1e-6m)to move the injection point into the domain, it all works fine.

Marek