| Project Detail |
A Swiss-French consortium aims to develop a fast and scalable numerical approach for dynamic fragmentation, which will be released as open-source software. Explosions and impacts result in the initiation of cracks that propagate at high speeds, coalesce to form fragments, and that may collide with one another. This process, called dynamic fragmentation, is of critical importance in many applications and in particular for aerospace industry, where accurate models are direly needed to characterize and estimate the evolution of space debris orbiting the earth. The research will promote robust physics-based software to predict the generation of debris during impact of structures, with a focus on obtaining accurate statistical distributions of fragment sizes, shapes, and velocities.The modeling approach is built on the complementary and unique expertise of two teams. The Swiss team has expertise in dynamic fragmentation and has developed high-performance algorithms for the insertion of cohesive cracks. Cohesive cracks give an explicit representation of crack surfaces and simplify the treatment of contacts between fragments, a crucial factor to predict debris velocities. The good scalability performance of the approach was shown to result in robust statistical data of fragment sizes for engineering problems. However, the cohesive approach is known to suffer from mesh dependency, with crack paths that depend on the underlying mesh, resulting in non-robust predictions of fragments shapes. The French group has a renowned expertise in the development of computational methods for mesh independent crack paths either with the extended finite element method (X-FEM), the Thick Level Set approach to fracture, or more recently the Lip-field approach to fracture. The latter belongs to the family of diffuse crack approaches in which the crack faces are not explicitly modeled. Unlike the cohesive crack approach, diffuse crack approaches lead to mesh-independent results but the computational costs are large and fragments are difficult to extract.The project will promote a new theoretical model for fracture and apply it to high-performance computing (HPC) fragmentation simulations. The model is called CLIP for Cohesive LIPschitz approach. Cracks are to be created in a cohesive manner but the originality is that they induce a diffuse damage around them, reducing mesh dependency. The diffuse damage is related to the cohesive one by an explicit Lipschitz type relation bringing good computational efficiency and mitigating the computational drawback of diffuse crack approaches.The research plan is composed of three work packages (WPs). WP1 will be carried by a PhD student at Ecole Centrale de Nantes (ECN) in France, and will investigate the performance of CLIP for static and dynamic crack propagation. Benchmark problems will be solved to demonstrate the convergence properties of the approach for fracture mechanics, with a particular focus on mesh independence of the solutions. WP2 will be carried by a PhD student at Ecole Polytechnique Fédérale de Lausanne (EPFL) in Switzerland, and will focus on a robust treatment of interactions between fragments, and on the physics of dynamic fragmentation in light of aerospace applications. WP3 unites WP1 and WP2 through the work of two part-time scientific computing experts at ECN and EPFL, with end goal to release a high-performance open-source software.
Grant number
212935
Funding scheme
Weave/Lead Agency
Call
Project funding in Mathematics, Natural sciences and Engineering (division II) 2022 April
Approved amount
380,184 CHF
Status
Ongoing |
