Download the following tar zipped file. The instructions below are for installing the program in HHPC. The tar zipped file is also saved in /Users/Shared in the server. Contact Bahram Notghi for help. These instructions are not complete.
A.Notes to access the HHPC Server
- For every test create separate folders in the respective individual user's folder.
- source code folder ( CCNScar 3D V5.1 ) is located in the main (root) folder. (Access using “cd”)
- Compile the source file using './makeit -tahoe'
- Copy 'binary' into test folder using cp command. Example: 'cp binary ../../../../scratch0/hhpc_hn1/rbhardwaj/myfolder/testfolder'
- Export input files from terminal , if necessary.
- Make the binary readable, writable and executable by all by 'chmod 777 binary'
- The availability of processors is checked using the command 'showbf'.
- The time on the server is asked by submitting a jobrequest , which is done by the file 'jobHHPC2.pbs' . Edit this file to allocate the amount of wall time requested based on the availability of processors and the name of the run to distinguish from tests of others. Once done, submit the job using the command 'qsub jobHHPC@.pbs'.
- Once submitted the job is given an ID which helps to uniquely identify this job.
- To check the status of the submitted job type 'qstat <username>' or 'qstat <jobID>'.
- To delete a submitted job type 'qdel <jobID>'.
In order to restart the code change the irest_tahoe in input file to 1 and irest in CCNS to 1. The restart process doesn’t happen automatically, before restarting the process ./CopyRestart need to be executed and then you can submit the job.
Pre-Process: After creating everything in Cubit, convert the mesh to *.unv ( this file will be used by fortran file to convert it to the format that CCNS can read it ) and then*.e and then *.off Open the .stl with Meshlab and will convert it to .off , this is surface mesh and the fortran file will change it to a readable format for CCNS. At the end journal file there is few line of script that take care of the conversion mostly.
- Create the mesh convert it to Unv and off
- Use PROCESS_UNV_V4.3.F90 to create the unv version of scaled and fitted in to globe ( this file create some necessary input for CCNS and also transformed version of the mesh( UNV)
- Open the unv and convert it to *.off ( use cubit and meshlab similar to previos step)
- Use Process_OFF
Specific Note: Everything related to isotropic model *.unv can be obtained based on *.jou ( this model have sclera as one block.)
For the Anisotropic model, some of the surfaces and meshes were fixed manually after cutting the sclera. In order to get this model, first load aftercut.cub in cubit and then run .jou in the aniso folder. ( cutting the sclera created so many issues that make it impossible to fix it with automatic *.jou , there were silver in mesh that needed to be removed manually).
- Future work: in order to have everything automatic , it is recommended to start from scratch in Jou and modify it, instead of cutting it.
- The scelra with Vibrational thickness use different PROCESS file than the Uniform thickness, all the file are in corresponding folders.
Simulation:
- CCNS the fluid solver will be defined
- Input Tahoe model will be defined
- canonical_body_in.dat is for the surface Fluid-solid interaction.
- For each new mesh all of these input files will be changed.
- For new material properties only input file will be changed.
