FAQ: Abaqus

1. How do I create my model in ABAQUS

ABAQUS/CAE can be used to create your model. To start, issue the following commad at the prompt:

For graphica sessions with 3D OpenGL graphics

prompt:> abaqus cae

For graphical sessions using X11 (without 3D OpenGL graphics:

prompt:> abaqus cae -mesa

The graphical interface is user-friendly and it gives the new user an option to run the ABAQUS/CAE Tutorials at the start of the program.

2. What is the abaqus_v6.env file and how do I use it?

The abaqus_v6.env is the environment file which you can create to set certain environment variables that Abaqus reads when the software is started up. Abaqus reads the environment variables from the following files in the sequence shown:

the global abaqus_v6.env file set by the administrators
the abaqus_v6.env file in the user’s home directory
the abaqus_v6.env file in the current working directory

Only items ii and iii are accessible by normal users.

3. What are the typical variables found in the abaqus_v6.env file?

The detailed list of variables that users can set in the abaqus_v6.env file is found in the online manual. A sample abaqus_v6.env file in SVU may contain the following:

#pre_memory = "1024 mb"
#standard_memory = "1024 mb"
ask_delete=OFF
scratch="/hpctmp2/scratch"
#
# for mpi jobs
#
mp_rsh_command = 'ssh -l %U %H %C'

The first 2 lines set the pre-execution memory and the standard memory to 1024 MB. These are normally not needed, and hence they are commented with the “#” character. “ask_delete=OFF” tells Abaqus not to prompt before deleting old result files. “scratch=/hpctmp2/scratch” tells Abaqus to use /hpctmp2/scratch to store temporary files during analysis runs. mp_rsh_command tells Abaqus to use the ssh command when running parallel MPI jobs.

4. How do I run my ABAQUS/Standard and ABAQUS/Explicit analysis jobs?

The ABAQUS/Standard and ABAQUS/Explicit solvers are only available on NUS IT’s compute clusters. You will not be able to run analysis from within a CAE session running on MobileVisLab. All jobs will have to be submitted to our job scheduler which will launch the analysis job on our computational clusters as a batch process. The job scheduler will manage the job and select the least loaded host for your job to run.

First, create your input files using ABAQUS/CAE Job Manager window (do not submit the job from ABAQUS/CAE). After creating the input file, place the input file and any user subroutines you have in a directory that is commonly accessible such as /hpctmp2 and login to the compute server via ssh. At the compute server, issue the following command in the directory where the input files are placed:
To submit an Abaqus analysis job to all the queues in PBS Pro, please follow the procedure described below. In the examples given, we assume that you already have an Abaqus input file, sample.inp created using Abaqus, and it is in the /hpctmp2/testdir directory that you created and you have permission to write in this directory.

Submitting a job to PBS Pro Scheduler
Create a job script using a text editor such as emacs or vi:

prompt:> cd /hpctmp2/testdir
prompt:> emacs MyJobScript.txt

and type the following lines into your script file:

#!/bin/bash
#PBS -P My_Abaqus_Analysis_Job
#PBS -j oe
#PBS -N My_Model_001

### The following will run on 24 cpus
## Queue selection line
#PBS -q parallel24
## Compute resources selection line
#PBS -l select=1:ncpus=24:mpiprocs=24:mem=160GB

## the following lines are required, do not edit / modify
cd $PBS_O_WORKDIR;
np=$( cat ${PBS_NODEFILE} |wc -l ); ### get number of CPUs

## Make a temporary scratch space (automatically created in /hpctmp2/)
scratch=/hpctmp2/${USER}/${PBS_JOBID}
export TMPDIR=$scratch
mkdir -p $scratch

## Run ABAQUS job as below
abaqus job=sample cpus=$np interactive
wait

Submit the job using the command below:

prompt:> qsub MyJobScript.txt

Note: You may select the following queues instead of parallel24:

To run on 8 cpus:

## Queue selection line
#PBS -q parallel8 
## Compute resources selection line
#PBS -l select=1:ncpus=8:mpiprocs=8:mem=10GB

To run on 12 cpus:

## Queue selection line
#PBS -q parallel12 
## Compute resources selection line
#PBS -l select=1:ncpus=12:mpiprocs=12:mem=10GB