About Abaqus

Abaqus is a comprehensive finite element analysis (FEA) suite developed by Dassault Systèmes under its SIMULIA brand. It is designed to simulate complex real-world engineering problems involving stress, deformation, heat transfer, mass diffusion, acoustics, piezoelectricity, and electrochemistry across diverse industries.

The suite consists of several core modules:

• Abaqus/Standard: An implicit finite element solver, best suited for static, quasi-static, and moderate dynamic problems.

• Abaqus/Explicit: A dynamic explicit analysis package optimized for highly nonlinear, transient events such as impacts and crashes.

• Abaqus/CAE: A finite element pre- and post-processing package used for building models and visualizing results.

Abaqus is recognized as an advanced analysis code specializing in production analysis, with a comprehensive, modular set of capabilities and consistent, intuitive problem-definition rules. It enables analysts to predict how structures respond to realistic loading conditions such as impact, vibration, thermal effects, and material failure, and its implicit and explicit solvers support everything from detailed component analyses to large system-level models.

Current Licensed Software Version

Abaqus 2026 Golden

Installation

The installation files and the installation document for the latest version of the software, along with network licensing information are available on our FTP server. Abaqus is also installed and available on our Linux clusters, BwUniCluster3.0 and Horeka to run computationally intensive simulations. The Abaqus software license agreement is also available on the FTP server.

Note: Installation files for versions that are not listed above can be obtained upon request by contacting scs-contact∂scc.kit.edu.

Licensing

Network License: To use the network license, please specify the Dassault Systèmes License Server (DSLS) license server name and port during installation as indicated in the network license configuration file available on the FTP server. An active internet connection is required to use the software. If you are accessing it from outside the KIT network domain, please connect via VPN in order to reach the license server. For instructions on setting up and using the VPN, please refer to the link: Remote Access (VPN).

Note: If you encounter problems connecting to the DSLS license server during the installation even after using VPN, we recommend selecting the "Skip licensing configuration" option. This allows the installation of Abaqus to complete successfully. After the installation is complete, navigate to the following directory:

C:\ProgramData\DassaultSystemes\Licenses

If the file DSLicSrv.txt does not exist, please create it. Then enter the DSLS license server information (hostname:port) provided exactly as in the network license configuration file available on the FTP server and save the file. You should then be able to launch Abaqus.

Abaqus simulations with large memory or CPU requirements should be executed on the Linux HPC cluster in batch mode. Batch execution allows users to utilize multiple CPU cores, large memory resources, and long wall-clock times managed by the Slurm workload manager. We recommend you use the latest version of the software available on the cluster.

Prerequisites

Before submitting a job to the cluster:

• Create your Abaqus input file (.inp) locally, or export it from Abaqus/CAE. Please note that the simulation input files must be created using the exact version of Abaqus that is available on the cluster. Using a different version may result in compatibility issues or prevent the simulation from running successfully.
• Copy the input file and any additional required files (e.g., user subroutines, include files) to your home or project directory on the cluster.

Abaqus Module on HPC Cluster

The cluster uses the Environment Modules system. To view all available Abaqus versions:

module avail cae/abaqus

Load the latest version of the software:

module load cae/abaqus/2026

Create and Submit Slurm Batch Script

For Single-Node Job, create a job script (for example, abaqus_job.sh) with the following contents. This example uses a single node with threads-based parallelization.

#!/bin/bash
#SBATCH --partition=cpu
#SBATCH --nodes=1
#SBATCH --ntasks-per-node=48
#SBATCH --cpus-per-task=1
#SBATCH --time=03:00:00
#SBATCH --mem=10G
#SBATCH --error=00_input-%j.err
#SBATCH --output=00_input-%j.out

module load cae/abaqus/2026

abaqus job=00_input \
       input=00_input.inp \
       cpus=$SLURM_NTASKS \
       mp_mode=threads \
       double=both \
       output_precision=full interactive

Always try to use appropriate values for the slurm parameter. In the above shell script content,

• cpus specifies the total number of CPUs used by Abaqus (typically $SLURM_NTASKS).
• mp_mode=threads runs the job using shared-memory threading, suitable for a single node.
• double=both and output_precision=full enable full double precision for both analysis and results.
• interactive keeps the job running in the foreground so Slurm can track its completion status.

For Multi-Node Job that span multiple nodes, Abaqus requires MPI-based parallelization (mp_mode=mpi) and a host list generated from the Slurm node allocation.

#!/bin/bash
#SBATCH --partition=cpu
#SBATCH --nodes=2
#SBATCH --ntasks-per-node=48
#SBATCH --cpus-per-task=1
#SBATCH --time=03:00:00
#SBATCH --mem=90G
#SBATCH --error=00_input-%j.err
#SBATCH --output=00_input-%j.out

module load cae/abaqus/2026

# Generate host list for multi-node MPI execution
HOSTLIST=$(scontrol show hostname $SLURM_NODELIST | \
           awk -v n=$SLURM_NTASKS_PER_NODE '{print "[\x27" $1 "\x27, " n "]"}' | \
           paste -sd, -)

echo "mp_host_list = [$HOSTLIST]" > abaqus_v6.env

abaqus job=00_input \
       input=00_input.inp \
       cpus=$SLURM_NTASKS \
       mp_mode=mpi \
       double=both \
       output_precision=full interactive

Where

• The HOSTLIST line reads the Slurm-allocated node names and writes them into abaqus_v6.env, which Abaqus reads at startup to distribute the MPI ranks across nodes.
• mp_mode=mpi switches Abaqus to MPI-based parallel execution required for multi-node jobs.
• --mem should be scaled up appropriately, since memory is now split across --nodes nodes.

Submit the job to Slurm using:

sbatch abaqus_job.sh

Check the status of your job:

squeue

Best Practices

• Create and validate your input deck locally (or via Abaqus/CAE) before submitting it to the cluster.
• Always copy your .inp file and any additional required files to the cluster before running.
• Use mp_mode=threads for single-node jobs and mp_mode=mpi only when running across multiple nodes.
• Choose --ntasks-per-node and cpus according to the allocated Slurm resources, and match --nodes to your license's MPI/token allowance.
• Check the .log and .msg files after the simulation completes for solver progress, convergence information, and memory usage.