29 Abr Doctoral Thesis on Delamination in Large Composite Structures
On April 28, Pierre Daniel successfully defended his PhD thesis titled «A strategy for efficient modelling of composite delamination in large structures» at the University of Girona. His research was co-supervised by Dr. Pere Maimí (Universitat de Girona) and Dr. Martin Fagerström (Chalmers University of Technology).
About the research:
Delamination — the separation between layers in composite laminates — poses a critical challenge in industries relying on fibre-reinforced polymers, such as aerospace, automotive, and wind energy. While current numerical models can simulate this phenomenon with high accuracy, their computational cost often makes them impractical for large-scale structures.
Pierre’s work proposes an adaptive modelling strategy that begins with a single element for the laminate’s thickness and introduces local refinements only where delamination occurs. This drastically cuts down computational demand while maintaining accuracy where it matters most.
Key contributions:
- A novel method to reconstructs out-of-plane stresses across the laminate thickness.
- An adaptive crack growth approach based on large finite elements (~5 mm), using an energy-based criterion (Virtual Crack Closure Technique) coupled with a new cohesive law to accurately model the crack advancement while dissipating the fracture energy.
- A new algorithm capable of handling distorted meshes and complex delamination fronts, making it applicable to real-world structures.
These methods can be used independently — for instance, to identify high-stress regions in large structures or to perform efficient damage tolerance assessments — but their combined use offers a powerful and scalable modelling strategy for large composite structures, drastically reducing the computational cost while maintaining accuracy.
Explore some of his publications:
https://doi.org/10.1016/j.compstruct.2020.112675
https://doi.org/10.1016/j.compositesa.2022.107423
https://doi.org/10.1016/j.engfracmech.2024.110193
This research was funded by BTECH.
