Skip to Main content Skip to Navigation

Étude numérique de la nocivité des défauts dans les soudures

Abstract : Defects can be formed during assembly processes such as welding and additive manufacturing. Current non-destructive inspection techniques are able to detect these voids in a mass production context. Thus there is a critical need of methods able to assess the harmfullness of these defects. These methods must be reliable to derive acceptation criteria. Their application in a industrial context requires efficiency to give an answer in a short time. This work aims at proposing an innovative design methodology that takes these defects into account. This work focuses on a particular welded joint : two titanium alloy parts are assembled by means of a pulsed laser. Three main topics are adressed. First, the cyclic behavior of the fusion zone is studied and modeled. X-Ray tomography is used to characterize the defects formed by the process. Their morphology is studied and a library of defects is built. The spatial distribution of the defects is closely linked to the welding process. A quantitative analysis provides a model able to generate realistic defect populations. To decrease the computational cost of numerical simulations of structures with defects, a model order reduction technique is proposed. It relies on the hyper-reduction method. A multi-scale reduced order basis building enables the use of data coming from tomographic analysis. Moreover, the offline phase cost is decreased.
Document type :
Complete list of metadata

Cited literature [120 references]  Display  Hide  Download
Contributor : ABES STAR :  Contact
Submitted on : Friday, March 20, 2020 - 8:57:09 AM
Last modification on : Wednesday, November 17, 2021 - 12:28:39 PM
Long-term archiving on: : Sunday, June 21, 2020 - 1:07:45 PM


Version validated by the jury (STAR)


  • HAL Id : tel-02512870, version 1


Laurent Lacourt. Étude numérique de la nocivité des défauts dans les soudures. Matériaux. Université Paris sciences et lettres, 2019. Français. ⟨NNT : 2019PSLEM050⟩. ⟨tel-02512870⟩



Record views


Files downloads