Identification de fractures dans un milieu poreux

Abstract : This PhD is dedicated to the mathematical study of an inverse problem in hydrogeology: the goal is to identify fractures in porous medium, knowing measurements of the underground flow. The number, the location and the physical parameters of the fracture are looked for. This problem is formulated as the least squares minimization of a function evaluating the misfit between measurements and the result of the direct model. We used a model describing the flow of a monophasic incompressible fluid (Darcy's law), in a porous medium containing some fractures represented by interfaces. The direct problem is the fracture model discretized by the mixed hybrid finite element method. To solve this inverse problem, we developed an iterative algorithm, which is based on the use of fracture indicators that have been developed durig the thesis. These indicators give a first order information concerning the effect of the addition of a new fracture. As these indicators are inexpensive, a large number of configurations of new fractures is tested at each iteration. The algorithm was programmed, validated and tested numerically in various situations, using synthetic measurements. It gives very satisfactory results, although this problem is considered difficult. Finally, an early study of identifiability of the inverse problem of fractures in porous medium has been achieved. It allowed to prove the identifiability for a simplified model (very permeable faults, which is common in the underground). The question of identifiability for the full fracture model remains open.
Document type :
Liste complète des métadonnées
Contributor : Abes Star <>
Submitted on : Tuesday, March 21, 2017 - 10:49:54 PM
Last modification on : Thursday, February 7, 2019 - 1:32:35 AM
Document(s) archivé(s) le : Thursday, June 22, 2017 - 2:52:06 PM


Version validated by the jury (STAR)


  • HAL Id : tel-01415825, version 2


Fatma Cheikh. Identification de fractures dans un milieu poreux. Analyse numérique [math.NA]. Université Pierre et Marie Curie - Paris VI, 2016. Français. ⟨NNT : 2016PA066417⟩. ⟨tel-01415825v2⟩



Record views


Files downloads