Inverse parameter identification of n-segmented multilinear cohesive laws using parametric finite element modeling

Research output: Contribution to journalJournal articleResearchpeer-review

21 Citations (Scopus)
27 Downloads (Pure)


Delamination in laminated composites are efficiently modelled with the cohesive zone model (CZM). The shape of the cohesive law becomes important when simulating delamination in material systems experiencing large scale fiber bridging, as several competing damage mechanisms occurs in the fracture process zone at multiple length scales. For this purpose a multilinear cohesive law has recently been developed in [1], which readily can be adapted to a variety of shapes. However, a key challenge in applying such cohesive laws is their model calibration, i.e. identification of parameters that define the shape of the cohesive law. In this work, a new methodology for experimental characterization of multilinear cohesive laws is proposed. The methodology is an inverse approach, which identifies cohesive laws by iteratively varying cohesive zone parameters using a gradient-based optimization scheme to minimize the error in structural response between a fracture mechanical experiment and a parametric finite element model. The method is demonstrated on a moment loaded double cantilever beam (DCB) specimen made of unidirectional glass-epoxy showing large-scale fiber bridging. Multilinear cohesive laws are characterized which result in an excellent agreement between the finite element simulation and the experiment. The results and sensitivity studies demonstrate the accuracy and robustness of the proposed methodology, even for a large number of design variables in the optimization problem.

Original languageEnglish
Article number111074
JournalComposite Structures
Publication statusPublished - Oct 2019


  • Delamination
  • Experimental characterization of cohesive laws
  • Gradient-based optimization
  • Inverse finite element modelling
  • Large scale bridging


Dive into the research topics of 'Inverse parameter identification of n-segmented multilinear cohesive laws using parametric finite element modeling'. Together they form a unique fingerprint.

Cite this