[dreamchaser]

Hello,

I am trying to set up a moving mesh problem. I am new to this kind of simulation so I am sorry for the questions.

*Problem Description
The problem I am trying to simulate is shown in the first picture.
I would like to spin a sphere at some RPM. There is a force of some magnitude applied at the bottom wall. I would like to do a moving mesh where I calculate the deflection of this bottom wall due to the pressure generation that will be created by the rotating sphere.

*Boundary Conditions
I first did the Ball Check Valve moving mesh tutorial in CFX which was very helpful. Regarding my setup, I have set the bottom wall as the rigid body.
I have then defined a force acting on this bottom wall. I sometimes changed the force to a spring b/c convergence was better.

The sphere is set to a rotating wall with some RPM. The other walls are set to stationary no-slip walls.
The mesh motion for the rotating wall and the side walls are set to unspecified. I did this because as the lower wall deflects downward, I assume we want the nodes on the side of the walls to move downward.
The top wall's mesh motion is set to stationary.

*Results
What I noticed when I am running the moving mesh simulation is that the whole geometry is expanding. Instead of just the bottom wall moving downward, the size of the sphere is increasing. I am not sure what is causing this. (I attached a picture of the initial time step and one at the end of the simulation). Also, I noticed the boundary of the sphere is deforming. Any advice would be greatly appreciated!!!

Thanks!!

[Martin_Sz]

If You dont want to determine a heat transfer between fluid domain and rotating disc try to define this disc like a immersed solid. - if You do this this disc doesnt deform on Your simulation
Best regards

[ghorrocks]

I agree - if you can do this model with immersed solids it will be much easier and avoids the distortion issues you are having. Read the documentation on the restrictions for immersed solids. But as long as you are OK with the restrictions then immersed solid is far simpler.

[dreamchaser]

Originally Posted by ghorrocks

I agree - if you can do this model with immersed solids it will be much easier and avoids the distortion issues you are having. Read the documentation on the restrictions for immersed solids. But as long as you are OK with the restrictions then immersed solid is far simpler.

Thank you both for the response. I figured out a way to make it work without using immersed solids. After reading the documentation, it looks like doing an immersed solid simulation assumes the whole disk is submerged in the oil. However this is not the case for my simulation.

What I did was I defined the disk as the rigid body. I am then applying a force at the center of the disk. The problem I am facing is that the disk is either colliding with the bottom wall or the top wall (rising).

I am trying to apply a set force to the center of the disk and then would like the disk to reach a steady state condition where it would be spinning at some steady state distance from the bottom wall. Based on equations for bearings, the disk is supposed to lift when it is spun at a certain speed.

*Problem description to match
I am trying to match a case where a sphere weighing 4.084g is spinning with an average velocity of 1m/s. The weight of the sphere is .04N. Based on theory there is supposed to be a minimum film thickess of 27microns. The viscosity of the oil is .05 Pa.s.

To replicate the paper, I set the mass of the rigid body to 4.084g and rotation speed to 1m/s. Since I do not have gravity on, I put a force of .04N at the center of the rigid body to replicate its weight. When I run the simulation the sphere crashes into the bottom wall and I am not seeing any lift. As a note, if I apply no force the sphere is lifted and moves upward. (I have constrained the motion of the sphere to the y-axis)

*Questions
1)I notice that in the beginning the pressure is high because the sphere is close to the wall. However, as it moves away, the pressure of the domain increases even more (pic2). Is this because I have not specified an opening? I have everything else as a wall. (Picture 1 and 2)

2)In the paper, the lubricant fluid level is only submerging half of the sphere. This is different where I have put oil in the whole domain submerging the entire sphere. Does this extra oil cause extra weight on the sphere not making it generate enough lift? I don't have gravity on so I assume this extra fluid on top of the sphere has no effect.

3)In the rigid body settings, I can specify the center of mass and even rotation. Do I need to initialize the center of mass of the sphere?

4) A possible problem I believe is that I have not specified the viscosity as a function of pressure? In some of these papers they define the viscosity as

vis=vis_initial*exp(a*p)

where a is called piezoviscous coeff (based on fluid) and p is pressure. However my sphere is lifting for some cases so I'm not sure if this is the problem.

any advice would be appreciated.

Thanks!!

[diamondx]

This is tough, I do not have experience in mesh deformation, but reading your setup, I hope I can may be help. It is good that you have a lot of question, That is the key to troubleshooting.
1- You mention that the lubricant level is only half submerging in the fluid ? Why you did not do that ? the tutorial of CFX about the tank slushing start by patching just a region of the domain with water, may be you can do same for your domain ?

2-Why you did not set up the viscosity as function of the pressure ?

3-You don't know if you have to put wall or opening ? What about the paper ? does it mention it ??

[dreamchaser]

Originally Posted by diamondx

This is tough, I do not have experience in mesh deformation, but reading your setup, I hope I can may be help. It is good that you have a lot of question, That is the key to troubleshooting.
1- You mention that the lubricant level is only half submerging in the fluid ? Why you did not do that ? the tutorial of CFX about the tank slushing start by patching just a region of the domain with water, may be you can do same for your domain ?

2-Why you did not set up the viscosity as function of the pressure ?

3-You don't know if you have to put wall or opening ? What about the paper ? does it mention it ??

Diamondx thanks for the reply and for everyone else.

I am actually using a spring instead in my problem because I realized that convergence with an applied force was not very good. I will try to set up viscosity as a function of pressure. I would like to explain my logic and would appreciate it if anyone can verify if it makes sense.

Instead of applying a force, I am putting a spring at the center of the sphere. (Picture Attached). The objective is to design the spring constant based on the RPM of the sphere to keep the sphere from rising. As a note, I have specified the front and back planes as symmetry to simulate a 2D case.

My logic is that F=kx. This spring force must balance the generated force in the gap which is equal to Fgen=PA where P=pressure and A=area. So we have:

PA=kx leading to k=(PA)/x

*x is going to be the distance the sphere will lift up from the bottom wall. I ran an earlier steady state case (non moving mesh) where the bottom wall was 500 microns from the sphere. As a result, I have what the maximum pressure will be in the gap from this steady state simulation. So I know P and x for the moving mesh simulation.

I am confused on what to use for the area?? Regarding the steady state case I ran (no moving mesh) I know what the max pressure is going to be. I also have an analytical solution on where this max pressure will be. However,
what width in the z-direction do I use for the area if I am essentially doing a 2D simulation (front and back are symmetry conditions)??

If I know the right area to use, then I can find the right spring constant k.
Putting this spring on the sphere, for the moving mesh simulation should keep the sphere stationary. The oscillations should die out and go to zero.

I appreciate any insight you can provide.

Thanks!