This repository contains a command-line tool that can be used to post-process the output of finite element analysis performed on a cracked model in Abaqus, in order to calculate the J-integral and the stress intensity factor. The tool operates on either 2D or 3D models, meshed using either CPS6 triangular elements or C3D10 tetrahedral elements. The main novelty of the work is to calculate the J-integral for unstructured tetrahedral meshes in three dimensions, which cannot be performed in Abaqus.
Follow these instructions to set up the environment and run the analysis.
- Abaqus CAE: This tool interfaces with Abaqus, so a local installation of Abaqus CAE and accompanying license is required to run the model creation and data export scripts. The analysis script can be run stand-alone, using previously exported results.
- Python: Python 3.9 or higher.
- pip: Python package installer.
-
Clone the repository:
git clone https://github.com/allabright/abaqus-j-integral-sif.git cd https://github.com/allabright/abaqus-j-integral-sif.git -
Create and activate a virtual environment (recommended):
# For macOS/Linux python3 -m venv venv source venv/bin/activate # For Windows python -m venv venv .\venv\Scripts\activate
-
Install the required dependencies:
pip install -r requirements.txt
This work was created as part of an MSc thesis. The full report is available in report/full_report.pdf (), and the accompanying LaTeX files are also available in report/latex. The results of the parametric studies are available in the results directory, along with a Jupyter Notebook used to plot the necessary graphs.
➡️ Read the full thesis here (PDF)
--dimension - The dimension of the model being analysed, either 2d or 3d.
--model - The name of the model being analysed. The only model supported at present is Edge_Crack. If you have your own ready-made model that you want to analyse, it needs to be dragged into the data/models directory, and the name of the directory needs to be passed as the value for this argument.
--input -- By default, the number of the last directory is used. For example, if the last directory in the data/models/Edge_Crack_2D directory is Model_008, then that will be used for the export by default. If you want to use another model - for example Model_005, then 005 needs to be passed as an argument here.
--config -- By default the config.ini file in the config directory is used to specify the configuration of the analysis. If you want to use a different file, then drag it into the config directory and pass the filename as the value for this argument.
The model must first be created in Abaqus, using the model.py file in the src directory. This can be done manually as long as the correct approach is followed, but it is automated here for the edge-crack case study. Usage is as follows:
python model.py --dimension X --model Y
This creates a model with dimension X using the model Y, using the parameters specified in the configuration file, and runs the analysis. A new directory is created within the data/models/Edge_Crack_{dimension} directory for this specific model. The configuration file is copied to this directory, and the results are saved here. This script must be run on a machine which has Abaqus CAE installed, along with the appropriate license.
Before the analysis can be performed, it is necessary to dump the raw data (e.g. stresses, strains) from Abaqus. This is done using the export.py file contained in the src directory. This is a command line script, and is run using:
python export.py --dimension X --model Y
This script creates a new directory for the export under data/exports/Edge_Crack_{dimension}, and copies over the model data, and the configuration data used for that model. The data is then dumped using the Abaqus API into a JSON file called export.json. This script must be run on a machine which has Abaqus CAE installed, along with the appropriate license.
After exporting the raw data from Abaqus, the analysis can be performed. This is done using the analysis.py file contained in the src directory. This is also a command line script, and is run using the same arguments as the previous script:
python export.py --dimension X --model Y
This script creates a new for the analysis under data/analysis/Edge_Crack_{dimension}, and copies over the model data, export data, and configuration data. The analysis is then performed, and the results are stored in a JSON file and a CSV file within the run directory.
No model has been created yet, so you call:
python model.py --dimension 3d --model edge_crack
This creates the directory /data/models/Edge_Crack_3D, copies over the config/config.ini configuration file, then creates and runs a 3D analysis within Abaqus, for a 3D edge-cracked part.
Now, for example, you have run four models. However, you want to perform an export using the third. You call:
python export.py --dimension 3d --model edge_crack --input 003
A new directory is created within data/exports/Edge_Crack_3D. The latest model located within data/models/Edge_Crack_3D/Model_004 is ignored because the --input argument was passed. Instead, the data is copied from data/models/Edge_Crack_3D/Model_003 to the run directory. The results of the analysis are exported, and saved in the run directory.
Now you have a set of exports within the data/exports directory. However, you want to change the weight function used for the analysis, without having to re-run your model and export. You therefore alter the config/config.ini file to change the weight function, and save it as config/config_edited.ini. You then call:
python analyse.py --dimension 3d --model edge_crack --config config_edited.ini
Since no --input argument was specified, the most recent export in the data/exports/Edge_Crack_3D directory is used. A run directory is created for the analysis in data/analysis/Edge_Crack_3D, and the model, export, and configuration data is copied over. Since a --config argument was specified, the customized configuration file is used for the analysis. The analysis is performed, and the results are saved in the run directory.