[BaranOz]

Dear friends,
I have a question regarding the calculation of wave forces on a pile from Reef3D.
For the wave force calculation, a wave with the waveheight of H=0.004m and wavelength of L=4.0m is propagated on to a pile with a diameter of D=0.1m.
Water depth is taken as h = 0.5m.
When I completed the simulation and plotted the time history of wave force, I saw a problem. Initial wave force was supposed to be zero (Since waves still yet to reach to the pile), but the values started with weird numbers.
I attached the waveforce vs time plot and also the control and ctrl files.
Any help would be really great.
Thanks in advance,
B.

[kamath]

Hi Baran,
The primary reason I would guess for this is the grid.
You can check how the object is represented using contour by solid = 0 in Paraview over your vtu files / its y-normal slice. (This needs P 25 1 in the input, though you can run it for just one time step with P 20 1 to get one output file for visualisation)

You will probably also observe that this value will change if you remove all your stretching input commands and run it with the base dx=0.05.

How do we know that the chosen grid is appropriate?
By a grid convergence study


Hope that helps

[BaranOz]

Thanks so much, Arun! As you said, when I removed grid stretching commands force values started nearly at zero.
I just want to ask why should I have to use the same grid size at every part of the domain?
I thought that the wave force values would be much more accurate with the stretched grids near the pile.
Thank you
B.

[kamath]

You are right in that a finer grid near the cylinder should provide a better approximation of the forces.
You are also right in that you don't need to use a uniform grid in the whole domain (that is why we have the option to stretch the grid).

The crux of the matter is that the stretching of the grid introduces a new parameter into the decisions you have to make in meshing, and therefore is both a blessing and a curse.

You will need to look at the Non-Uniform-Grid log file in DiveMesh_Log to look at the max and min size of the cells and the stretching factors. There are a few guidelines out there to restrict the deformation that is introduced by grid stretching within a certain limit, otherwise you introduce errors to to sudden jumps in the grids.
In an extreme case, a sudden jump in the grid size can also result in the simulation crashing due to extremely high velocities at the border of the twp grid zones.

In short, you can get the right results with a stretched grid, but the stretching needs to be within some bounds (for eg. a cell ratio of 1.08-1.2).

A solution for the x/y-direction would be B 101 11 where the region around your cylinder could be constant dx of your choice and the zones outside stretch / coarsen gradually to grids sufficient for wave propagation.
For the z-direction, I am generally happy with B 103 5.

Hope that helps.

[BaranOz]

Okay. Thank you.
Baran