Random Fractures

Repository containing the tools to perform the calculations for the thesis Quantification of fracture patterns with random parameter fields for phase field models. The aim of this thesis is to quantify the influence of uncertain material parameters in the porous medium on the crack propagation.

Fracture simulation

The pde models describing the crack propagation in porous media is solved with an inhouse code PANDAS. The data are output as tecplot data, containing the data for the variables at the triangle nodes, but also at the gauss nodes of the used Finite-Element shape function.

Installation

git clone https:/kaiserls/RandomFracturePatterns.git
cd RandomFracturePatterns
pip install .

To install PANDAS, get a copy of the PANDAS software from MIB, place it in the folder RandomFracturePatterns and compile it following the user manual. You propably need to install the library libreadline-dev with apt to compile pandas successfully.

Running the simulations

To run the simulations from this thesis you need to get the custom pandas app folder used in this thesis. Write an email to me to get in contact.

python src/main.py

If this does fail, due to failing imports, try:

python3 -m src.main

Analysis of the simulation results

The simulation results will already be post-processed, analyzed and visualized by the main script. More control and the settings used in the thesis can be obtained by using the jupyter notebook evaluate.ipynb.

Computing on a server

1. Connect to local network over vpn.
2. Copy your ssh key to the server
3. Start the computation, e.g. with screen

Screen

1. Start computation with screen -dm bash -c 'python3 -m src.main' -S theOptionalScreenName -L
2. Reattach with screen -r theOptionalScreenName
3. Detach from screen with strg+a strg+d
4. List screens with screen -ls to check if finished
5. Kill screen if needed with screen -S theOptionalScreenName -X quit

Kill rogue tasks

ps -o pid= -u your_username | xargs kill -1