En:using ww3:2013 fall waves school:triangles 201
- Polymesh manual ... some discussion of it is in this video.
- Slides for "triangles 201" :These discuss the optimization of grid time steps for explicit schemes, using a fixed external contour and island nodes. this video, after time 19 minutes.
Tutorials: hands-on lab
Copy TOOLS/POLYMESH from your yellow submarine.
Go to the directory, then in your terminal (if using gnu compilers)
mkdir $HOME/bin make clean make cd triangle make all
For Intel compilers first change Rules.make
Mesh generation and editing
Exercise 1: Editing the Hawaii mesh. You can now go to this page:
Exercise 2: designing a grid with Polymesh
Dowload the test cases from [here] Test case with the "HISWA" break water.
You can run polymesh for the following cases in this archive. These correspond to the laboratory basin (8 m by 8 m roughly). All the Polymesh behavior is specified by the meshin.nml file. In the archive the meshin.nml has an output format set to OUTPUTFORMAT = 'XFN' , In order to visualize the results you should change that to OUTPUTFORMAT = 'TRIANGLE' if you want to use showne, and OUTPUTFORMAT = 'MSH' to have the proper format for WAVEWATCH III, and to visualize the grid in TWIST. Note that the grids that you will make here will have coordinates in meters (not longitude and latitude), so that when you display it in TWIST it will wrongly think that they are degrees of lon and lat: this will result in the timesteps estimate to be too large by a factor 111000.
- meshing with depth only (randonly)
- add island (randisland)
- add structure
- refine with polygon
- refine but witout structure
- refine with CFL constraint
- refine with DZ constraint: more elements on the steeper slopes.
Exercise 3: generating the island and contour file for polymesh
In your case you should have a shoreline data in the ~/DATA/SHORELINE folder of your thumb drive. It is called france_edited.line Read this one with the '*.line coastline format' option of TWIST.
If you follow the tutorial you should end up with two files rand.dat and insel.dat in your ~/TOOLS/IDL directory. These two files, and the meshin.nml file, are the minimum information needed to run Polymesh.
An example of these two files was provided in the AFTER_IDL folder... but it was making a mesh in only one of the islands (the number 10). This is a polymesh bug. This island is particular because its center of gravity is not inside of the island. This bug is corrected in the version of polymesh from November 8.
The test case will run with the old version of polymesh if this island (number 10) is deleted.
Also, several plotting bugs (contour line ignoring the last 2 points, wrong zoom region on initial loading ...) have been corrected in the rev299 update of the TWIST idl tool. The corrected version (take both the rt_visumain.tar.gz and rt_visumain.sav) is here.