A general method for numerical identifiability and sensitivity analysis of failure criteria for continuous fibre-reinforced plastics

• Published in: European journal of mechanics, A, Solids A, Solids, 2023-07, Vol.100, p.104976, Article 104976
• Main Authors: Takács, Donát M., Berezvai, Szabolcs, Kovács, László
• Format: Article
• Language: English
• Subjects: Composite; Failure; Failure criteria; Identifiability; Sensitivity
• ISSN: 0997-7538
• DOI: 10.1016/j.euromechsol.2023.104976

Modelling the static failure of fibre-reinforced composite laminates is a challenging problem, as the parameters of the relevant mechanical models might not have a physical interpretation. Failure parameters are usually determined experimentally, where the parameter fitting and experimental design can be non-standard and intricate. In order to improve efficiency in designing mechanical experiments for the fitting of composite failure model parameters, a general method for quantifying the sensitivity of failure criteria parameters to failure stress states has been developed. This method helps selecting the failure stress states with regards to a failure model that result in the most robust fit of the model parameters. The method is demonstrated on the commonly used Tsai–Wu and Puck failure models, but is valid for a wide range of other models as well. The practical application of this method is also demonstrated by evaluating an experimental data set. •A general method for sensitivity analysis of composite failure models is presented.•Stress-space representation is used with Voronoi sectoring for systematic sampling.•Measurement errors and variation of material parameters are modelled stochastically.•The method is demonstrated and validated on Tsai–Wu and Puck failure criteria.•Parameter sensitivity data and experimental design recommendations are provided.