Digital Design strategies to certify and mAnufacture Robust cOmposite sTructures

Part of EU Open Research Repository

There is a newer version of the record available.

Published February 12, 2025 | Version v1
Dataset Open

Data of "Stochastic Deep Material Networks as Efficient Surrogates for Stochastic Homogenisation of Non-linear Heterogeneous Materials"

Authors/Creators

  • 1. ROR icon University of Liège

Contributors

Data curator:

Producer:

  • 1. ROR icon University of Liège

Description

Stochastic Deep Material Networks as Efficient Surrogates for Stochastic Homogenisation of Non-linear Heterogeneous Materials

This directory contains the data and algorithms generated in publication1

Table of Contents

  1. Dependencies and Prerequisites
  2. Structure of Repository
  3. Images/Geometries and IB-DMN training data of the 6 SVEs
  4. Stochastic analysis - Direct numerical simulations of SVEs
  5. Training of the reference IB-DMN
  6. Stochastic analysis - Stochastic IB-DMN
  7. Reproduce paper[^1] figures

Dependencies and Prerequisites

  • Python, pandas, matplotlib, texttabble and latextable are pre requisites for visualizing and navigating the data.

  • For generating mesh and for vizualization, gmsh (www.gmsh.info) is required.

  • For running simulations, cm3Libraries (http://www.ltas-cm3.ulg.ac.be/openSource.htm) is required.

Instructions using apt & pip3 package manager

Instructions for Debian/Ubuntu based workstations are as follows.

python, pandas and dependencies

 sudo apt install python3 python3-scipy libpython3-dev python3-numpy python3-pandas

matplotlib, texttabble and latextable

 pip3 install matplotlib texttable latextable

Pytorch

  • Without GPU
 pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cpu
  • With GPU
 pip3 install torch torchvision torchaudio

Libtorch (only when using cm3Libraries)

  • Without GPU: In a local directory (e.g. ~/local with export TORCHDIR=$HOME/local/libtorch)
 wget wget https://download.pytorch.org/libtorch/cpu/libtorch-shared-with-deps-2.3.0%2Bcpu.zip 
unzip libtorch-shared-with-deps-2.1.1+cpu.zip
  • With GPU: In a local directory (e.g. ~/local with export TORCHDIR=$HOME/local/libtorch)
 wget https://download.pytorch.org/libtorch/cu121/libtorch-shared-with-deps-2.1.1%2Bcu121.zip 
  unzip libtorch-shared-with-deps-2.1.1+cu121.zip

Structure of Repository

Images/Geometries and IB-DMN training data of the 6 SVEs: 6SVE_Example

  1. 6SVE_Example/6SVE_Data: Images/Geometries and IB-DMN training data of the 6 SVEs
  2. 6SVE_Example/6SVE_DNS:
  1. 6SVE_Example/6SVE_DMN:

Stochastic analysis - Direct numerical simulations of SVEs: Stochastic_DNS_LinearHardening

  1. Functions and data used to generate SVE of UD fiber reinforced composites following 2:
  1. Test setup
  1. Generate SVEs and running simulations
  • Stochastic_DNS_LinearHardening/MicroSample_UDComp.py Generates the SVEs and runs the DNS (requires cm3Libraries)
  • Calls previously listed files to generate SVE
  • Nsim = 100 to set the number of SVE realisations to be generated
  • TestKey = 'Tensile' or TestKey = 'Shear' or TestKey = 'UniStrain' to run under uniaxial stress, shear or uniaxial strain
  • minVf = 0.3 and maxVf = 0.35 to set the bounds on the fiber volume fraction
  • prefix = './Res'+TestKey+'/Vf_03_035/' sets the directory where to save the results (to be consistent with the volume fraction bounds and directory should exist)
  • Results of simulations stored in Stochastic_DNS_LinearHardening/ResTensile and Stochastic_DNS_LinearHardening/ResShear for uniaxial stress and shear

Training of the reference IB-DMN: Vf_Mat

  1. Training of general IB-DMN:

Stochastic analysis - Stochastic IB-DMN: Stochastic_DMN_LinearHardening

  1. Files used to initialize the IB-DMN material law of levels 4, 5 and 6:
  1. Adding a random perturbation on the reference IB-DMN parameters and perform nonlinear simulations:
  • Stochastic_DMN_LinearHardening/Tool.py
    • Called by the following file (not stand alone)
  • Stochastic_DMN_LinearHardening/GenParaDMN.py
    • GenPara = False is adding perturbation on the reference IB-DMN
    • GenPara = True is generating a random IB-DMN
    • Load = 'Tensile' or Load = 'Shear' to run under uniaxial stress or shearing
    • FigureOnly = True to plot figures using archieved results, FigureOnly = False to generate new stochastic IB-DMN, new data and run tests (requires cm3Libraries) before plotting figures
    • Id_Para = '_partial' or Id_Para = '_a25_b05', Id_Para = '_a5_b075', Id_Para = '_a10_b05' defines the saving directory and if only partial distribution (in ln=[depth] and lw=[depth]) or not are considered
    • a= 2.5 and b = 0.5 define the distribution parameters
    • ``level = 4, level = 5``` or ```level = 6``` specifies the level of the IB-DMN to be used
    • Results are stored in ./Res* for the different loading cases and combinations of parameters a and b of the random perturbation distributions (partial or not).

Reproduce paper1 figures

 python3 Plot_DNS_DMN.py
  • Fig. 5:
    • The command to be run from the root directory . to plot the beta distribution is
 python3 Beta.py
  • Fig. 6, Fig. 7, and Fig. 8:
    • Set FigureOnly = True
    • Parameters: Id_Para = '_a25_b05', a=2.5, b=0.5
    • Use successively level = 4, level = 5, and level = 6
    • Use successively Load = 'Shear' and Load = 'Tensile'
    • The command to be run from the directory Stochastic_DMN_LinearHardening is
 python3  GenParaDMN.py
  • Fig. 9 and Fig. 10:
    • Set FigureOnly = True
    • Parameter: level = 5
    • Use successively Id_Para = '_a25_b05', a=2.5, b=0.5, then Id_Para = '_a5_b075', a=5, b=0.75, and then Id_Para = '_a10_b05', a=10, b=0.5
    • Use successively Load = 'Shear' and Load = 'Tensile'
    • The command to be run from the directory Stochastic_DMN_LinearHardening is
 python3  GenParaDMN.py
  • Fig. 11:
    • Set FigureOnly = True
    • Parameters: Id_Para = '_partial', ln=[0,1,2,3,4], lw=[], a=2.5, b=0.5 and level = 5
    • Use successively Load = 'Shear' and Load = 'Tensile'
    • The command to be run from the directory Stochastic_DMN_LinearHardening is
 python3  GenParaDMN.py
  • Fig. 12:
    • Set FigureOnly = True
    • Parameters: Id_Para = '_partial', ln=[], lw=[0,1,2,3], a=2.5, b=0.5 and level = 5
    • Use successively Load = 'Shear' and Load = 'Tensile'
    • The command to be run from the directory Stochastic_DMN_LinearHardening is
 python3  GenParaDMN.py
  • Fig. 13:
    • Set FigureOnly = True
    • Parameters: Id_Para = '_partial', ln=[4], lw=[3], a=2.5, b=0.5 and level = 5
    • Use successively Load = 'Shear' and Load = 'Tensile'
    • The command to be run from the directory Stochastic_DMN_LinearHardening is
 python3  GenParaDMN.py
  • Fig. 14:
    • Set FigureOnly = True
    • Parameters: Id_Para = '_partial', ln=[3], lw=[2], a=2.5, b=0.5 and level = 5
    • Use successively Load = 'Shear' and Load = 'Tensile'
    • The command to be run from the directory Stochastic_DMN_LinearHardening is
 python3  GenParaDMN.py
  • Fig. 15:
    • Set FigureOnly = True
    • Parameters: Id_Para = '_partial', ln=[0,1,2,3,4], lw=[], a=2.5, b=1 and level = 5
    • Use successively Load = 'Shear' and Load = 'Tensile'
    • The command to be run from the directory Stochastic_DMN_LinearHardening is
 python3  GenParaDMN.py
 python3 Plot_rve6_DMN.py
 python3 Plot_DNS_DMN.py

Disclaimer

This project has received funding from the European Union’s Horizon Europe Framework Programme under grant agreement No. 101056682 for the project “DIgital DEsign strategies to certify and mAnufacture Robust cOmposite sTructures (DIDEAROT)”. The contents of this publication are the sole responsibility of ULiege and do not necessarily reflect the opinion of the European Union. Neither the European Union nor the granting authority can be held responsible for them.

  1. The present work is described in: "Wu, L. and Noels, L. (Submitted). Stochastic Deep Material Networks as Efficient Surrogates for Stochastic Homogenisation of Non-linear Heterogeneous Materials , doi: " which can be downloaded. We would be grateful if you could cite this publication in case you use the files.0 ↩ ↩2

  2. The micro-structure generator is described in: "Wu, L. and Chung, C. N. and Major, Z. and Adam, L. and Noels, L. (2018) From sem images to elastic responses: A stochastic multiscale analysis of ud fiber reinforced composites, Composite Structures 189, doi: 10.1016/j.compstruct.2018.01.051 ↩

Files

2025_StochasticIBDMN-main.zip

Files (478.0 MB)

Name Size Download all
md5:e6d501934aaaf357fbb5b3befd8121b3
478.0 MB Preview Download

Additional details

Related works

Is described by
Publication: 10.1016/j.compstruct.2018.01.051 (DOI)

Funding

European Commission
DIDEAROT - Digital Design strategies to certify and mAnufacture Robust cOmposite sTructures  101056682

Software

Repository URL
https://gitlab.onelab.info/cm3/cm3Libraries/