Car crashworthiness global sensitivity analysis for USD 2024

Prerequisites

This code was developped using Python 3.6.8 and the following modules:

• Numpy 1.19.4
• Matplotlib 3.36.2
• SALib 1.3.11
• openturns 1.19.post1
• dynareadout 23.10.2

LS-DYNA executable is ls-dyna_smp_s_r1010_x64_redhat5_ifort160.

LS-DYNA car crash model

The model is located in lsopt_car6_v3 folder. The other lsopt_xxxfolders contain older/developement versions of the model. There are 3 LS-DYNA files:

• the main file is main_v223.k. It contains parameters set at their initial values;
• the main file main_v223_param.k is used in case the values of the parameters are changed;
• the rest of the definition of the model is included in car6_crash_v223.k.

Running the model from Python

Running the model from Python and performing the Global Sensitivity analysis is done with LSDYNAmodel.py module.

Post-treatment: plot all the Morris analyses

To plot the various Morris analyses, use GSAutils.py module. It reads and plots data stored manually in GSA/car_v223_right-impact_v30/morris_nXX_output.md, where XX stands for the number of Morris trajectories used for each analysis.

Polynomial Chaos metamodel and Sobol indices

Module PCEsobol.py builds a PCE metamodel and computes the Sobol indices. It relies on OpenTURNS.

The input and output of the car crash model are generated with LSDYNAmodel.py, they are stored in the LHS folder.

• LHS-nnn_X.csv, where nnn stands for the number of samples, contains the input samples;
• and LHS-nnn_Y_xxxx.csv, where xxxx stands for the name of the output, contains the output values.