[Lloyd Sullivan]

Hi All,

I am new to CFX and have a question regarding the setup of a simulation (I have read the forum/manual for related subjects).

The Problem
I am using V18 to determine added mass effect of a cylinder travelling through water in a fixed single direction using the inbuilt rigid body solver.
I will be setting different masses for the cylinder and will be applying a CEL expression for a prescribed acceleration of the cylinder (or external force F=ma).

The Questions
I am trying to determine the best method to translate the mesh around the cylinder, or the entire domain in one direction (X-direction), whilst keeping the mesh rigid .

I have found two possibilities (I don't want to use the Remeshing tool)

1. A sliding mesh that travels at the same rate as the rigid body (like in the attached Multi_domain.gif) where the motion of the rigid body is prescribed using an external force and somehow making the mesh domain slide with it.
https://imgur.com/Q2O48fX

2. Use a large domain and a smaller subdomain with a rigid mesh, but moves with the rigid body (much like the decoupled buoy tutorial) and using the 1.0[m^5 s^-1]/volcvol to ensure the mesh deforms very far away from the cylinder. (an example mesh is attached - please ignore the coarseness of the mesh)
https://imgur.com/JmKxOAx

I would ideally like to use the first approach. However, i'm not entirely certain how to implement the idea.
Any new suggestions or help with the problem (tutorials/guides that i haven't seen, etc) will be enormously appreciated

Best regards,

Lloyd Sullivan

[ghorrocks]

Before you dive into the complexities of FSI and rigid body motion - is FSI/rigid body motion the only way of modelling your application?

Can't you model the cylinder as a stationary object and have the fluid flow past it? This does not need FSI or rigid bodies at all. If this is inappropriate please explain why.

[Lloyd Sullivan]

Originally Posted by ghorrocks

Can't you model the cylinder as a stationary object and have the fluid flow past it? This does not need FSI or rigid bodies at all. If this is inappropriate please explain why.

Hi Glenn,

Thank you very much for replying so promptly.

In response to your question:
I need to apply different masses to the cylinder, with different amounts of acceleration so as to determine how much error there is between the analytical curve (example in link with some arbitrary inputs https://imgur.com/SZdxwGL ).

Therefore, this leads me to believe a rigid body solution will be required to add the mass with an external force applied in one direction (unless there is a simpler method involving CEL expressions).

Once again, any help is greatly appreciated.

Best regards,

Lloyd

[ghorrocks]

The cylinder mass is irrelevant, the fluid effects and mass effects are easily separated. but if you are modelling the cylinder accelerating then you probably need to model it as you suggest as the approach I described is accurate only for inertial frames (ie zero acceleration).

To answer your question - this appears to be a moving mesh simulation, not rigid body or FSI. You appear to be applying a known motion to the body, so you can simply just move the mesh for that motion and extract the forces. Rigid body and FSI are used when the forces then affect the motion.

[Lloyd Sullivan]

Hi Glenn,

Sorry for the delayed response, I have been away.

Thank you very much for the helpful information. I'm assuming when you say easy to separate, you mean to find the force with the applied acceleration function then implement F=(mass_solid + mass_added) *acceleration to find the added mass?

However, I would like to note that it is important that I model the mass of the cylinder, as the next step of the simulation is to analyse the number of iterations required for convergence. As according to the theory, as the mass of the cylinder relative to the fluid decreases, the number of iterations should increase for each time step to converge.

Once again, any feedback is greatly appreciated.

Best regards,

Lloyd

[ghorrocks]

In your previous post you said you apply different accelerations and wish to determine the added mass effects. Do you know the entire motion in advance? In other words, you apply known accelerations to the body and get the added mass from that?

Your second paragraph is very confusing. What do you mean by number of iterations to convergence? Convergence of what? And what theory says you should have more iterations as the cylinder mass increases?