Sharing calculations for the study of composite materials

Sébastien Brisard* and Frédéric Legoll materials science researchers at the Navier Laboratory, joint research unit (UMR) ENPC, Université Gustave Eiffel and CNRS

As part of their research into predicting the properties of heterogeneous composite materials, two researchers from the Navier Laboratory have contributed to the debate on data sharing and the reproducibility of experimental calculations.

How can the properties of a heterogeneous material be predicted by looking at those of its constituent elements? Mechanics of materials researchers at the Navier Laboratory, Sébastien Brisard and Frédéric Legoll, are looking into this question using a theoretical approach. They are studying the influence of microstructure parameters on macroscopic response. To reduce the cost of the numerical calculations required, they implemented a strategy of “variance reduction” based on the “equivalent inclusion method”. Using a digitally simulated composite material, they carried out a series of calculations aimed at deducing its thermal conductivity from the properties and proportions of its basic constituent elements. In practical terms, this involves varying the input parameters and observing the impact this has on the properties of the material being studied. The aim is to determine which parameters have a significant influence on the result in order to improve prediction using simplified models.

The results of this research, together with details of the calculations carried out using open-source software and scripts written by the two researchers, were published in an article: A variance reduction strategy for numerical random homogenization based on the equivalent inclusion method. They were then posted on the data-sharing platform Recherche.data.gouv.fr. The experimental data can be used by anyone, including researchers or lecturers, who wishes to reproduce the calculations or carry out new ones. Sébastien Brisard and Frédéric Legoll were committed to this approach of sharing their data, but were soon confronted with the issues it raised, both in terms of identifying and organising open data. As these data are intended to be re-used, the researchers asked themselves how they should present the data so as to optimise their accessibility, which led them to rethink the formatting of their calculations as they worked to encourage re-appropriation.

The question of how to organise shared data raises another issue that is very prominent in the scientific community, and particularly in the open science community: that of reproducibility. In this case, although the available data allows others to reproduce the calculations made during the study, it does not guarantee that their results will be identical, as the software versions used may have change over time. In addition to the data made available, precise information needs to be provided on the software versions and hardware used initially. For Sébastien Brisard, now a professor in the Civil Engineering department of the Aix-Marseille IUT and a researcher at the Mechanics and Acoustics Laboratory (LMA), the issue of reproducibility is the next step in effective data sharing.

Composite material: A composite material is an assembly or a heterogeneous mixture of at least two materials of different natures and which presents performances superior to those of the components taken separately.

Reproducibility: Research is said to be reproducible if all the information concerning the research, including, but not limited to, the text, data and programming code, are made available so that any independent researcher can reproduce the results.

For more information:
Loic Desquilbet, Sabrina Granger, Boris Hejblum, Arnaud Legrand, Pascal Pernot, et al.. Vers une recherche reproductible : Faire évoluer ses pratiques. Unité régionale de formation à l’information scientifique et technique de Bordeaux. pp.1-136, 2019, 979-10-97595-05-0. hal-02144142v1

* Sébastien Brisard was a researcher at Université Gustave-Eiffel at the Navier Laboratory when he carried out the work presented in this article. In September 2023, he joined the University of Aix-Marseille and the Mechanics and Acoustics Laboratory (LMA),

Interview performed by the Kogito agency.