Launch abaqus python: 
====================
abaqus cae noGUI
>>> import sys
>>> print sys.argv
['/sw/abaqus/2017/SIMULIA/CAE/2017/linux_a64/code/bin/ABQcaeK', '-cae', '-noGUI', '-lmlog', 'ON', '-tmpdir', '/tmp', '-academic', 'RESEARCH']
>>> 

By default, Abaqus will run itself in the background, but that doesn’t work well with PBS, which will assume the job has completed when Abaqus goes into the background 
and end the job before Abaqus completes its job. So, you should always use the interactive option when running Abaqus from a PBS script.

Modify the following example script to reflect the correct parameters for the resources you are using. We recommend that you use /scratch/snumber directory when running abaqus. #!/bin/bash
####  PBS preamble
#PBS -N abaqus_job
#PBS -m abe
#PBS -M YourName@griffith.edu.au
#  Request 2 processors on 1 physical machine
#PBS -l select=1:ncpus=2:mem=12gb,walltime=5:00:00
#PBS -q workq
####  End PBS preamble
module load abaqus/2017
#  Include the next three lines always
if [ -e "${PBS_NODEFILE}" ] ; then
   # Print nodes the job is running on
   uniq -c $PBS_NODEFILE
fi
#  Put your job commands after this line

#  Create a temporary directory for sample run
cd  $PBS_O_WORKDIR
mkdir $PBS_O_WORKDIR/sample
cd $PBS_O_WORKDIR/sample

#  Get a sample file from the Abaqus test suite source  /usr/local/bin/s3proxy.sh
abaqus fetch job=t5-std

# requested 6gb per CPU, which is a little more than
#   what we ask for here, which is 5gb per CPU;
#   you should always add a little for program overhead.  
#  

abaqus job=jobname inp=t5-std cpus=2 memory=5gb 

Interactive run:
===========
qsub -I -l select=1:ncpus=2:mem=12gb,walltime=5:00:00

Once you are inside the job:
cd  $PBS_O_WORKDIR
mkdir $PBS_O_WORKDIR/sample
cd $PBS_O_WORKDIR/sample

module load abaqus/2017
source  /usr/local/bin/s3proxy.sh
abaqus fetch job=t5-std
abaqus job=jobname inp=t5-std cpus=2 memory=5gb interactive