Search

Wendelstein 7-X (W7-X), the world’s largest nuclear fusion experiment of modular stellarator type, started operation in 2015 and will be upgraded with a water cooled first wall for steady state operation in 2020. The first wall consists of a CFC armored island divertors, adjacent baffles, heat shields, and stainless steel wall panels. Baffle and heat shield segments consist of graphite tiles, bolted with low pre-stress onto heat sinks of CuCrZr that are in turn brazed onto water cooled steel pipes. Cracks were detected before installation in the baffles in the root of the brazed seam in over 100 locations. Such cracks are attributed to the imposed plastic deformation of the pipes to bring them into the final shape following the complex 3D geometry of the plasma vessel. This paper gives an overview of the experimental and numerical work using finite element method (FEM) and dual boundary element method (DBEM), including sub-modeling to assess the risk of a water leak during operation. Details of the numerical work is published in Giannella et al. (2017), Lepore et al. (2017) and Citarella et al. (2018). First fatigue crack growth experiments were carried out on pipe material and thermal-mechanical crack growth predictions were made with FEM and DBEM. It appeared that the Stress Intensity Factor (SIF) threshold of the ductile steel is only reached when large plastic strains occur, thus violating the field of application of linear elastic fracture mechanics to forecast crack growth. Afterwards, representative brazed pipe samples were manufactured and subjected to initial plastic deformation causing cracks in 11 out of 12 samples. Some samples were tested up to 60,000 bending load cycles. Two out of four samples failed after ∼35,000 cycles. Before and after the test, the shape of the cracks was measured using 3D computer tomography scans. Equivalence between thermal load in W7-X and the mechanical load in the cyclic test was determined with the numerical models to allow for a prediction of the fatigue life in W7X. Additional modeling showed that also plastic zones away from the cracks can limit the fatigue life.

Three-dimensional numerical analyses, using the finite element method (FEM), have been adopted to simulate fatigue crack propagation in a hollow cylindrical specimen, under pure axial or combined axial-torsion loading conditions. Specimens, made of Al alloys B95AT and D16T, have been experimentally tested under pure axial load and combined in-phase constant amplitude axial and torsional loadings. The stress intensity factors (SIFs) have been calculated, according to the J-integral approach, along the front of a part through crack, initiated in correspondence of the outer surface of a hollow cylindrical specimen. The crack path is evaluated by using the maximum energy release rate (MERR) criterion, whereas the Paris law is used to calculate crack growth rates. A numerical and experimental comparison of the results is presented, showing a good agreement in terms of crack growth rates and paths.

An inverse procedure is developed for computing the material parameters of the class of intrinsic cohesive-zone models for the analysis of structural adhesive joints presented in [Valoroso N, Champaney L. A damage-mechanics-based approach for modeling decohesion in adhesively bonded assemblies, Engng Fract Mech 2006;73:2774–801.]. In particular, using the same experimental data recorded as of ISO 25217 test protocol and Double Cantilever Beam bonded specimens, a deterministic identification that overcomes the usual limitations and difficulties of ISO 25217 is proposed to estimate the mode-I cohesive parameters. The driving rationale for our approach to identification is to set up an inverse procedure based on one single experimental test that uses the measured load–deflection curve and crack length as data set. A finite element model updating scheme is used that renders extensions to more general situations such as different geometries, large displacements and nonlinearities in the bulk material, possible with minor modifications. Experiments are conducted for symmetric DCB specimens under pure mode-I bending and an optimization problem is solved to find a suitable agreement between experimental data and finite element computations. Identification results are presented and the key role played by the crack progression in the identification is discussed to assess the effectiveness of the proposed approach.