Skip to Main content Skip to Navigation

Modélisation et prédiction du comportement macroscopique du sel gemme dans le contexte du stockage souterrain

Abstract : The optimal design of underground facilities in salt formations depends on the quality of the experimental measurements used to develop constitutive laws and to calibrate their parameters, as well as on the predictive abilities of these modelson the short and long term. A comprehensive experimental campaign was carried out, its analysis along with that ofthe existing experimental data base in MINES ParisTech, raised questions concerning the assumption of specimens representativity during laboratory tests. The effect of specimens heterogeneity on the macroscopic behavior of rock saltwas numerically investigated within a virtual laboratory. This investigation showed that specimens heterogeneity can explain the irregularities often observed in experimental measurements and that the tested specimens can be smaller than the required representative volume element (RVE). In an approach allowing to elaborate a macroscopic constitutivelaw and to overcome representativity related issues, the required RVE size was correlated to that of the heterogeneities.The pertinence of the predictive abilities of this macroscopic law was addressed and a new law, aiming to overcome the limitations of existing models in terms of the quality of their long term predictions, was proposed.
Document type :
Complete list of metadata
Contributor : ABES STAR :  Contact
Submitted on : Thursday, July 8, 2021 - 5:14:07 PM
Last modification on : Wednesday, November 17, 2021 - 12:31:41 PM
Long-term archiving on: : Saturday, October 9, 2021 - 7:07:45 PM


Version validated by the jury (STAR)


  • HAL Id : tel-03282134, version 1


Mejda Azabou. Modélisation et prédiction du comportement macroscopique du sel gemme dans le contexte du stockage souterrain. Sciences de la Terre. Université Paris sciences et lettres, 2021. Français. ⟨NNT : 2021UPSLM012⟩. ⟨tel-03282134⟩



Record views


Files downloads