[FelixD]

High Reef 3D Team,

I am currently looking at rendering some 2D breaking wave simulations in paraview and have noticed that the free surface becomes quite 'stepped' once it reaches the bed. I believe this is to do with the cells being significantly taller than they are long. However, this is not the case anywhere else along the free surface where the wave appears quite smooth.

I was wondering if you had any advice on generating good renders and smooth profiles once the water reaches the beach. I am also simulating sediment transport so it would be ideal to has this part of the simulation and grid as smooth as possible.

I've attached images of the free surface and my simulation txt files, including a close up that shows the grid super imposed on the free surface. I am using B 99 1 (relaxation method 1) for the numerical beach and B 96 9.0 0.0 (wave gen and beach distance) if that is any help too.

Thanks,
Felix

[kamath]

Hi Felix,

From the first attachment, I think you might be taking a contour of the wrong variable or so? The stepped profile towards the end of the domain is already a part of the solid where there is no water.

I see that you do include the free surface and the bottom slope into the render correctly in the second picture.

The zoomed in 3rd picture again seems to be from under the free surface.

So using the procedure that you used for the second picture should already give you the correct render without the "stepped profile".

Just to establish the procedure:
1. Load the main pvtu
2. Take a slice (as this is a 2D sim)
3. Take a contour on the slice, use variable phi, set value to 0.
4. Load REEF3D-Solid.vtp from the DIVEMesh_Paraview folder.
(seems like you have followed this for attachment 2, but I list out anyway to confirm)


Further, I have two suggestions that might help


dx=0.025 is a little too coarse if you are interested in the fine detail in the region where the free surface approaches a thin film. Like the sudden vertical free surface at the end seen in attachment 2.

I believe this is the Ting and Kirby case and iirc, dx=0.005 was originally used for the cfd simulation for this.

With respect to the free surface behaviour close to an object, or in the case of small amounts of fluid in run up on a solid, F 40 23, a different reinitialisation procedure for the level set method might be more useful.

[FelixD]

Originally Posted by kamath

Hi Felix,

From the first attachment, I think you might be taking a contour of the wrong variable or so? The stepped profile towards the end of the domain is already a part of the solid where there is no water.

I see that you do include the free surface and the bottom slope into the render correctly in the second picture.

The zoomed in 3rd picture again seems to be from under the free surface.

So using the procedure that you used for the second picture should already give you the correct render without the "stepped profile".

Just to establish the procedure:
1. Load the main pvtu
2. Take a slice (as this is a 2D sim)
3. Take a contour on the slice, use variable phi, set value to 0.
4. Load REEF3D-Solid.vtp from the DIVEMesh_Paraview folder.
(seems like you have followed this for attachment 2, but I list out anyway to confirm)


Further, I have two suggestions that might help


dx=0.025 is a little too coarse if you are interested in the fine detail in the region where the free surface approaches a thin film. Like the sudden vertical free surface at the end seen in attachment 2.

I believe this is the Ting and Kirby case and iirc, dx=0.005 was originally used for the cfd simulation for this.

With respect to the free surface behaviour close to an object, or in the case of small amounts of fluid in run up on a solid, F 40 23, a different reinitialisation procedure for the level set method might be more useful.

Thanks Arun,

The free surface is the same in all the attachments (phi with the value set to 0). The second attachment is the same as the first except the object is hidden in the first. I think you are correct in that it is simply an issue of grid size.

Cheers,
Felix