[salahgaffar]

Hello Foamers,

To start with, the problem I am working on a 4-stroke internal combustion engine problem using Openfoam v2012 (the dot com solver branch). I have managed to solve the physics part of the problem well, however, the current issue is mesh motion. I am currently using velocityLaplacian motion solver.


To model my real engine problems, I need to use the following grid structure:

(refer to image 1 - complex problem topology)


P.S. The engine geometry libraries available in v2012 (e.g. engine valve and engine piston) were useful initially but will not give me enough controllability over the mesh quality for the grid topology I am currently using.

As an initial simple problem to get things going, I started working on the following topology:

(refer to image 2 - very simplified topology)



To connect the fluid across the different regions in my topology, it was required to use cyclicAMI boundary conditions.

At the cyclicAMI boundary, a uniform fixed value of 0 was explicitly specified for the velocity.
However, the AMI seems to have allowed the pointMotionU solution to propagate to the second region (port region) and caused the solution to not allow my uniform value BC, as shown in the following image:

(refer to image 4 - image 4 - AMI vs Wall case to separate different motion regions)


This was remedied when a wall boundary condition was used at this interface. However this is not feasible when solving the physics (not only the motion).

My questions are as follows:

1) do you guys have any solution to provide a boundary condition that acts like a wall when solving the motion and acts as a cyclicAMI when solving the flow variables? Or do you think the solution might be to separate the different cell zones and solve each separately?

P.S. I tried using multibody motion solver but I hit a brick wall when I was unable to define the motion of the liner (side wall which has a slip BC).

2) Do you have any other alternatives for slip boundary condition at the liner, since I noticed that the slip boundary condition sometimes need to be confined to a certain component (velocityComponentLaplacian in z works fine but velocityLaplacian with BCs specified in z doesnt work for certain geometries (boundary starts bulging from the sides irregularly), as shown in image 5 below:


I have attached the run folders for the 2 cases being discussed in these links:


Case 1 - highlighting the problem with the wall vs AMI issue
---------------------------------------------------------------------
https://drive.google.com/file/d/17gb...ew?usp=sharing

Case 2 - highlighting the slip boundary condition issue
---------------------------------------------------------------------
https://drive.google.com/file/d/19Ip...ew?usp=sharing

[salahgaffar]

The images are attached below

[tecmul]

Hello,

1. From a brief look at the images, I don't understand why you need a cyclicAMI condition at all. As far as I know, the cyclicAMI condition is used for sliding meshes.

2. I had a problem similar to yours. The "fixedNormalSlip" boundary condition allows you to specify a normal vector for the boundary. The displacement of the mesh nodes will then always be perpendicular to that normal vector.

[salahgaffar]

Hello tecmul,

First of all, thanks for taking the time to answer my questions

1) the cyclicAMI was needed in order to allow non-conformal zones to be used in the mesh, similar to the first two attached images (e.g. refer to image 2, where it would be much better to generate a grid for the port alone and the cylinder alone). Also, in the more complex problem (shown in image 1), I would need to use a sliding interface (cyclicAMI) to allow for the valve motion within the deformable yellow region while maintaining the head tetras region (red) stagnant.

2) As for the fixedNormalSlip, from what I understand, it requires a unit normal vector (n). In the case of a round cylinder, the direction of this normal will differ from each point to the next. My current guess is that i'd have to figure out a way to use the face normal for each boundary in the code for that purpose.

Please feel free to correct me if Im wrong and again, I really appreciate your feedback!!

[tecmul]

1) Ok I see now. Just off the top of my head, maybe specify high diffusivity around the cyclic patch? Hopefully someone with more experience will comment.

2) Yes you're right. My case was 2D so a single normal vector worked. Maybe look at how the fixedNormalSlip boundary condition works and use that to write a new condition that limits node displacement to a certain direction? The Z axis in your case.