Hi all
The model I am trying to simulate is two cylinder ( outer and inner) which only the inner one rotates and a two-phase fluid flows between them. All walls are no-slip condition. Eulerian system is used to describe the flow. In COMSOL, is there the possibility to do that? I have tried "Rotating Domain" but the velocity on inner cylinder is zero.
And there is no "Rotating Shaft" as in Mixture Module of it.
Can anyone help me on this please ?
Ok so you problem is not only a rotating mixing problem but a two-phase.... out of experience, that might be tricky to solve due to the addition of time-dependent solver. I suggest you work first with a 2D geometry to get a feel on the required modelling parameters and their effect for a more complex 3D systems.
To create the 2D rotating geometry, follow these steps:
1. Create the inner and the outer circles and form an assembly with an identity pair type
2. To define the rotation of the inner domain, you need to define the rotation angle (an) as a function of time. This can be done by defining a variable and placing the following equations in it:
an = 1[rad/s]*t ;
change the value to match the angular rotation of you model but be sure you ramp it up correctly.
3. To define the rotation of the inner domain in the Cartesian coordinate, create a new variable for the model component and place the following equations
DX = -X+X*cos(an)-Y*sin(an)
DY = -Y+Y*cos(an)+X*sin(an)
DXdt = ant*(-X*sin(an)-Y*cos(an))
DYdt = ant*(-Y*sin(an)+X*cos(an))
4. In the fluid flow node, add a flow continuity option and pick the identity pair. Only the walls of the rotating domain need to have a "moving wall" boundary condition. For the velocity of the moving wall place DXdt and DYdt. This assure the momentum transfer of the fluid due to the moving wall.
5. The moving mesh node (I am not sure which version of Comsol you are using, but this method can be also applied to the newer versions):
Define a "prescribed deformation" node for the rotating domain with the prescribed mesh displacement as: DX and DY
6. After setting up the physics, make sure that the time-dependent solver runs stably. For that I usually use direct solver (for robustness). You can also increase the relative tolerance of the solver to make solve faster.
In total, I suggest you start small with a single fluid first and then ramp up the complexity to a two-phase model and complex geometries.
Good luck with your project.
I agree with Zaid Jildeh, to build your understanding of the problem, it is recommended to start with 2-D domain.
The boundary condition that simulate the rotation of the cylinder surface can be interpreted by giving the velocities (u and v) on the cylinder surface. These velocities relate the rotational speed of the cylinder and the coordinates. In the attached paper, you may found useful boundary conditions.
If you want to induce rotation of the wall and track the change, you must include the ALE method to the model (i.e. moving mesh). This can be found in mathematics node from Add Physics... In the newer version of Comsol (> v. 5.3), the moving mesh interface has been integrated in the definition node.
Concerning the rotation, i advise you to make two cylinders, then a union between them (keep the boundaries) and then a split. After that, form an assembly and the identity pair will form automatically.
From my experience, a rotating domain is easier to define than a rotating wall. Just follow the points explained in the first post.
The addition of the domain won't affect the solution time, since the inner domain is not critical to the fluid simulation.
Finally, i re-advise you to consider starting with a 2D model, just to get a feel of the comsol software and the necessary parameters to set the simulation correctly.
Good luck with your project.
- Badges
- Science topic