bug: solitary wave - McCowan and Grimshaw

The codes of the solitary wave based on the solution McCowan and Grimshaw have a few bugs.

The following codes marked with "New" fixed the bugs.

McCowanNewWaveModel.C GrimshawNewWaveModel.C

GrimshawNewWaveModel.H McCowanNewWaveModel.H

mentioned in issue #2234 (closed)

Thanks - can you share a test case/results that show the effect of your changes/fixes?

There is also a bug in the solitary wave-making method based on the Boussinesq solution.

All V&V to the three wavemaking methods will be provided before the weekend of this week.

@barjasg - are these known issues? thanks

changed the description

Here, the bug-fixed method of Boussinesq is also uploaded. BoussinesqNewWaveModel.H

BoussinesqNewWaveModel.C

In numerical simulations, the water depth is 1.0 m and the wave height is 0.20 m. The horizontal and vertical mesh resolutions are identically 0.05 m. The bottom is slip. Surface elevations are recorded every one meter. The length of the wave flume is 50.0 m.

Results of the default methods and bug-fixed methods are compared, as shown in this figure.

New comparisons are here.

Hi @lizhisongsjtu, Hi @andy,

It looks very good to me, but I would like to compile and test it before anything. I am quite surprise about the big difference in the default implementation between generated and expected, because I tested/validated with different cases.

I will try to replicate @lizhisongsjtu results to validate and certificate the great work.

I will update my results here.

Best, Gabi

hi Gabi, I noticed another issue that the solitary wave can not propagate from xmax to xmin (from right to left), or from ymax to ymin (from front to back).

Could you check this one? Thank you.

Best regards Zhisong

Hi @lizhisongsjtu ,

This is not an issue, it is as it has been developped. The idea of the implementation of the waveModels library is to replicate experimental facilities (two dimensional wave flumes and three dimensional wave basins) in which the wavepaddles are always placed in the same position. When we need to replicate different directions for wave impacts in objects, we do as in the laboratory, that is, turning or/and moving the object.

But yes, numerically speaking it can be done, generating and absorpting in different directions, but I would not include it, because from my point of view it is not the original idea of the code and because it messes up the code and the tutorials.

Best, Gabi

Hi @lizhisongsjtu, Hi @andy,

I include my verification of the corrected boundary conditions provided by Zhisong in OF-v2406. I am using the case provided by Zhisong, without modifying the numerical parameters. WaveHeight is 0.2m and initial water depth is 1.0m.

I am plotting results from a free-surface gauge placed 1m far from the generation patch.

Boussinesq solitary case: Mccowan solitary case: Grimshaw solitary case:

It can be seen that the new implementation can reduce the unwanted perturbations after the generation.

My only concern is about the wave height: from my results it is smaller than the 0.2m for the three cases. I have run cases with smaller cell sizes, and results are improved but the generated wave is smaller than 0.2m.

However, the plots from Zhisong show the opposite: eta/h is almost 1.0 in all cases.

Best,

Gabi

Hi, Gabi, Thank you for your reply.

In your simulations, the results are quite different from my results.

In my cases, the solitary waves are sound with expected wave heights and unnoticeable small perturbations. Here, only the results with Boussinesq solution are shown.

Results with dimensionless wave height: where blue dots are of of2406.

Results with dimension wave height: where blue dots are of of2406.

Hi @lizhisongsjtu ,

I am using a modified version of OF-v2406 in which I am developping other things, maybe there is something there that is not working properly yet.

Therefore, I have started compiling a new version of OF-2406 (without any modifications) and I will repeat the cases. Once done it I will share my results with you

Best,

Gabi

added community label

Hi @andy , hi @lizhisongsjtu,

I have tested and validated, and therefore, I confirm from my side that Zhisong work is correct. He has improved the solitary generation for the three proposed wave theories (Boussinesq, Grimshaw and McCowan).

Best,

Gabi

Thanks @barjasg @lizhisongsjtu - we will integrate the updates shortly