Evan: Created page with "'''Title of the script:''' Parallel Computing in OpenFOAM '''Script:''' Divyesh Variya, Ashuthosh Shridhar '''Video & Narration:''' John Pinto, Payel Mukherjee '''Keywords..."

2025-01-09T05:06:44Z

Created page with "'''Title of the script:''' Parallel Computing in OpenFOAM '''Script:''' Divyesh Variya, Ashuthosh Shridhar '''Video & Narration:''' John Pinto, Payel Mukherjee '''Keywords..."

New page

'''Title of the script:''' Parallel Computing in OpenFOAM

'''Script:''' Divyesh Variya, Ashuthosh Shridhar

'''Video & Narration:''' John Pinto, Payel Mukherjee

'''Keywords:''' OpenFOAM, Parallel Computing, Hierarchical, Manual, Scotch, Simple, decomposition, mpirun, Paraview, Video-Tutorial

{|border=1
|-
|| '''Visual Cue'''
|| '''Narration'''
|-
|| '''Slide 1:'''

'''Opening Slide'''
|| Welcome to this spoken tutorial on '''Parallel Computing in OpenFOAM.'''
|-
|| '''Slide 2:'''

'''Learning Objectives'''
|| In this tutorial, we will learn to,
* '''Decompose''' a domain
* '''Run''' a '''simulation''' using '''Simple decomposition '''method
* '''Reconstruct''' a domain
|-
|| '''Slide 3:'''

'''System Specifications'''
|| To record this tutorial, I am using,
* '''Ubuntu Linux''' OS version 22.04
* '''OpenFOAM''' version 9
* '''ParaView''' version 5.10.1 and
* '''gedit Text Editor'''
|-
|| '''Slide 4:'''

'''Prerequisites'''

* If not, please go through the prerequisite '''OpenFOAM '''tutorial on '''https://spoken-tutorial.org'''
|| To practice this tutorial, learner should have,
* Basic knowledge of '''OpenFOAM '''software
* If not, please go through the prerequisite '''OpenFOAM '''tutorials on this website.
|-
|| '''Slide 5:'''

'''Parallel Computing'''
|| In''' parallel computing, '''a large domain is split into multiple '''domains''' of smaller size.

Each of the smaller domains is assigned an individual '''processor'''.

The '''processors''' run simultaneously and communicate with each other.

'''OpenFoam''' uses the '''OpenMPI''' library to do this.

Thus, we can speed up the '''simulation'''.
|-
|| '''Slide 6:'''

'''Processes Involved'''
|| There are three main processes involved in '''parallel computing.'''

* '''Decomposition''' of the '''domain.'''
* '''Simulation''' of the '''decomposed tasks''' in separate '''processors.'''
* '''Reconstruction''' of the '''domain.'''
|-
|| '''Slide 7:'''

'''Decomposition Methods'''
|| The four main '''decomposition methods''' available in OpenFOAM are:

* '''Simple'''
* '''Hierarchical'''
* '''Scotch '''and
* '''Manual'''
|-
|| Only Narration
|| Let us learn about Simple Decomposition method using a 2 Dimensional example.
|-
|| '''Slide 8:'''

'''Simple Decomposition'''
|| Simple decomposition method divides the domain''' '''in equal parts'''.'''

The splitting is done according to the user-specified '''coefficients'''.
|-
|| '''Slide 9:'''

'''Input Parameter - Simple'''
|| '''Simple decomposition''' requires two inputs.
* '''Decomposition coefficients''' and
* '''cell skewness factor, delta'''

A typical value for '''delta''' is 10 to the power of minus 3('''10^-3).'''
|-
|| '''Slide 10:'''

'''Decomposition Coefficients'''
|| '''Decomposition coefficients''' represent the number of '''subdomains''' in a particular direction.

Thus, the product of '''nx, ny, nz '''must be equal to the total number of '''subdomains.'''
|-
|| '''Slide 11:'''

'''Simple Decomposition - n (2 2 1)'''
|| This slide shows the '''decomposition domain''' for the '''decomposition coefficients''' of (2 2 1).
|-
|| '''Slide 12:'''

'''Simple Decomposition - n (4 1 1)'''
|| This slide shows the '''decomposed domain''' for the''' decomposition coefficients '''of (4 1 1).
|-
|| Only Narration
|| For more information on '''decomposition methods, '''Please refer to '''Additional Reading material.'''
|-
|| Only Narration
|| Let us set up a '''case.'''
|-
|| '''Press CTRL + ALT + T keys.'''
|| Open the '''terminal''' by pressing '''Ctrl''', '''Alt''' and '''T''' keys.
|-
|| [Terminal] Type:

'''cd $FOAM_RUN'''
|| At the '''prompt''', type this '''command''' to go to the '''run directory'''.
|-
|| [Terminal] Type:

'''cp -r $FOAM_TUTORIALS/incompressible/icoFoam/cavity/cavity .'''
|| Copy the '''cavity''' '''case''' from the '''tutorial directory''' into the '''run directory'''.
|-
|| [Terminal] Type:

'''cd cavity'''
|| Navigate to the '''cavity case'''.
|-
|| [Terminal] Type:

'''blockMesh'''
|| Run the '''blockMesh''' command to execute the '''mesh'''.
|-
|| [Terminal] Type:

'''cp -r $FOAM_TUTORIALS/multiphase/interFoam/laminar/damBreak/damBreak/system/decomposeParDict system/'''
|| Copy a''' decomposeParDict''' file from''' '''the '''dam break''' '''tutorial''' to the '''system '''folder.
|-
|| Only Narration
|| To see the maximum number of '''processors''' available in your '''machine''', do the following.
|-
|| [Terminal] Type:

'''lscpu'''
|| Type '''lscpu''' on the '''terminal.'''
|-
|| [gedit - '''terminal'''] Highlight

'''CPU(s): 24'''
|| My '''machine''' has '''24 CPUs .'''
|-
|| [gedit - '''terminal'''] Highlight

'''Thread(s) per core…. 2'''
|| '''12 physical cores''' per '''socket '''and '''2 threads''' per '''physical core.'''
|-
|| '''Slide 13:'''

'''Number of processors'''
|| Multiplication of these 3 quantities gives the number of '''logical cores.'''

This is equal to the maximum number of '''processors'''.
|-
|| Only Narration
|| We will run this case in 4''' processors'''.
|-
|| Only Narration
|| Note:

Please assign at least 4 processors to your Virtual Machine.
|-
|| [Terminal] Type:

'''gedit system/decomposeParDict'''
|| Open the '''decomposeparDict '''in a '''text editor.'''
|-
|| [gedit '''decomposeParDict'''] Highlight numberOfSubdomains '''4'''<nowiki>;</nowiki>
|| The '''number of subdomains '''is defined as '''4.'''
|-
|| [gedit '''decomposeParDict''']

'''method simple;'''
|| '''Simple''' method of decomposition is being used.
|-
|| [gedit '''decomposeParDict'''] Highlight '''coeffs'''
|| Go to '''coeffs sub-dictionary.'''
|-
|| [gedit '''decomposeParDict''': Highlight '''n'''
|| '''n '''represents the Decomposition coefficients in each direction.

I will modify''' n '''so that we have:
* 2 '''divisions''' in '''x''' and''' y''' directions.
* And 1 division in '''z '''direction.
|-
|| [gedit '''decomposeParDict''']

Save >> Close the window
|| '''Save''' and Close the file.
|-
|| [Terminal] Type:

'''decomposePar -cellDist'''
|| To '''decompose '''the '''domain''', type the following '''command.'''
|-
|| [Terminal] Highlight:

'''cellDist'''
|| The argument '''cellDist''' writes a file named '''cellDecomposition''' inside the '''constant''' folder.

This file contains the details of the '''processor '''allocation.
|-
|| Only Narration
|| The '''domain''' is decomposed.
|-
|| Only Narration
|| The '''command''' '''decomposePar''' creates a directory for each '''processor'''.

These directories contain the '''subdomain''' and the corresponding''' geometric fields.'''
|-
|| [Terminal] Type:

'''ls'''
|| Type '''ls''' to view the folders assigned to each '''processor.'''
|-
|| [cavity - '''directory'''] Highlight

'''Processor0 …..Processor3'''
|| Since we use''' '''4''' processors''', we have 4''' processor''' directories.
|-
|| [Terminal] Type:

'''ls processor0'''
|| Type '''ls''' '''processor0 '''to view the content of''' processor0.'''
|-
|| [Terminal] Highlight

'''0 constant'''
|| It contains the '''0''' and '''constant '''directories.
|-
|| Point to '''0 directory'''.

Point to '''Constant directory'''.
|| '''0''' directory contains the '''decomposed geometric field'''.

The''' Constant '''directory contains the '''decomposed mesh files.'''
|-
|| [Terminal] Type:

'''gedit constant/cellDecomposition'''
|| Open the''' cellDecomposition '''file from the Constant folder in a '''text editor.'''
|-
|| [gedit - '''cellDecomposition'''] Highlight '''400'''
|| The number on the first line represents the number of '''cells''' in the '''domain.'''
|-
|| [gedit - '''cellDecomposition'''] Highlight

'''(.......'''
|| The entries within the bracket represent the '''processor''' number.
|-
|| [gedit - '''cellDecomposition'''] Highlight

'''0'''
|| '''0''' here indicates the '''processor''' number that is assigned to the corresponding '''cell.'''
|-
|| [gedit - '''cellDecomposition''']

Close the window
|| Close the file.
|-
|| Only Narration
|| Let’s view the '''subdomains''' in '''paraview.'''
|-
|| [Terminal] Type:

'''paraFoam'''
|| Type '''paraFoam '''on the '''terminal.'''
|-
|| [ParaView] Uncheck:

'''Properties Tab >> Scroll Down - Mesh Parts >> Uncheck fixedWalls & movingWall'''
|| Uncheck all other fields other than '''internalMesh''' from '''Mesh Parts'''.
|-
|| [ParaView] Click:

'''Properties Tab >> Scroll Down - Fields'''

'''>> check cellDist'''
|| Check all '''Volume Fields''' in the '''Properties tab'''.
|-
|| [ParaView] Click:

'''Apply'''
|| Click on the '''Apply''' button in the '''Properties tab.'''
|-
|| [ParaView]

'''Click on Solid Color >> Click on cell Dist'''
|| Click on '''Solid Color '''available in the '''Active Variable'''

'''Controls.'''

Select '''cellDist''' with a '''block''' from the '''menu.'''
|-
|| Point to the dark blue block and the light blue block.

Point to the beige block and the maroon block.
|| One can see the divided''' '''domain.

The '''dark blue''' '''block''' belongs to '''processor number 0'''

The''' light blue''' '''block''' belongs to '''processor number 1'''

The '''beige block''' belongs to '''processor number 2''',

And the''' maroon '''belongs to '''processor number 3'''
|-
|| ['''paraview''']

Close the window
|| Close the '''ParaView''' window.
|-
|| '''Slide 14:'''

'''Command - Parallel run'''
|| '''OpenFOAM '''uses''' '''the '''openMPI '''library for '''parallelization'''.

* '''mpirun''' is the '''process''' '''launcher'''.
* The '''argument''' '''np''' represents the number of '''processors'''.
* In our case we use '''4 processors.'''
* Then the name of the '''application '''entered.
* We need to specify that the '''application''' has to be run in '''parallel'''.
* The '''output''' of the '''simulation''' is written in a file named '''log.'''
* This is done by using the '''right''' '''angle''' '''bracket'''.
* The '''&''' indicates that the '''process''' will be '''run''' in the '''background'''.

This allows us to use the current '''terminal''' when the '''application''' is '''running'''.
|-
|| [Terminal] Type:

'''mpirun -np 4 icoFoam -parallel > log &'''
|| Type this '''command''' in the '''terminal.'''
|-
|| [Terminal] Type:

'''tail log'''
|| Type the following '''command''' to view the end of the '''log '''file.
|-
|| [Terminal]

Highlight''' End'''
|| The '''simulation '''is complete.
|-
|| Only Narration
|| To post-process the case, we need to '''reconstruct''' the domain.
|-
|| [Terminal] Type:

'''reconstructPar'''
|| To do that, type''' reconstructPar''' in the '''terminal.'''
|-
|| [Terminal] Type:

'''ls'''
|| Type''' ls'''
|-
|| [Terminal] Highlight

'''0.5'''
|| The end time directory is reconstructed.
|-
|| '''Slide 15:'''

'''Summary'''
|| With this we come to the end of this tutorial. Let us summarize.

In this tutorial, we have learned to,
* '''Decompose''' a domain
* Run''' '''the simulation using the '''simple decomposition method'''
* '''Reconstruct''' a '''domain'''
|-
|| '''Slide 16:'''

'''Assignment'''
|| As an assignment:

Simulate the same '''case''' using,
* '''Simple decomposition''' with '''2''' '''processor'''
* '''Scotch decomposition''' with '''processor weight'''
|-
|| '''Slide 17:'''

'''About the Spoken Tutorial Project'''
|| The video at the following link summarizes the Spoken Tutorial project.

Please download and watch it.
|-
|| '''Slide 18:'''

'''Spoken Tutorial Workshops'''
|| We conduct workshops using Spoken Tutorials and give certificates.

Please contact us.
|-
|| '''Slide 19:'''

'''Spoken Tutorial Forum'''
|| Please post your timed queries in this forum.
|-
|| '''Slide 20:'''

'''FOSSEE Forum'''
||
* Do you have any general/technical questions?
* Please visit the forum given in this link.
|-
|| '''Slide 21:'''

'''FOSSEE Case Study Project'''
||
* The FOSSEE team coordinates solving feasible CFD problems of reasonable complexity using OpenFOAM.
* We give honorarium and certificates to those who do this.
* For more details, please visit these sites.
|-
|| '''Slide 22: Spoken Tutorial'''
|| Spoken Tutorial project was established by the Ministry of Education, Govt. of India

This tutorial is contributed by Divyesh Varya, Nishit Pachpande, John Pinto and Payel Mukherjee from IIT Bombay.

Thank you for joining.
|-
|}

OpenFOAM-version-7/C3/Parallel-Computing-in-OpenFOAM/English - Revision history

Evan: Created page with "'''Title of the script:''' Parallel Computing in OpenFOAM '''Script:''' Divyesh Variya, Ashuthosh Shridhar '''Video & Narration:''' John Pinto, Payel Mukherjee '''Keywords..."