Cohesive zone model identification on mode I bonded assembly: sensitivity analysis
Langue
en
Communication dans un congrès avec actes
Ce document a été publié dans
Proceedings of the 22nd International Conference on Composite Materials - 2019, The 22nd International Conference on Composite Materials (ICCM22), 2019-08-11, Melbourne. 2019p. 1-12
Résumé en anglais
Adhesive bonding is usually modelled using cohesive zone models (CZM) which are defined by traction-separation (TS) law. For mode I loading condition these phenomenological laws simply represent the evolution of the peel ...Lire la suite >
Adhesive bonding is usually modelled using cohesive zone models (CZM) which are defined by traction-separation (TS) law. For mode I loading condition these phenomenological laws simply represent the evolution of the peel stress as a function of the two adherends relative displacement normal to the joint. However, TS law shape is often empirically chosen rather than being measured. The uncertainty on parameter estimation is generally not indicated even though it strongly influences the reliability of the bonded joint strength prediction. Moreover there are several mechanical data that can be obtained experimentally from crack initiation and propagation experiments on a Double Cantilever Beam Test (DCB). In general, TS parameters are chosen from load-displacement curves, which is the most straightforward mechanical response to obtain. However, the development of digital image correlation has enabled to access more numerous data, such as adherends’ deflection and rotation along the overlap and at loading point. The latter can be directly used to obtain the J integral. Adherends’ deformation can also be measured through the use of resistive strain gauges. Therefore, these different identification methods need to be compared in terms of parameter estimation confidence intervals. To do so, a numerical test campaign has been carried out for each mechanical response (i.e. load-displacement, J integral, and strain measurement) a synthetic noise is added to the nominal response in order to artificially represent measurement data. The noisy response is then used for the identification of the parameters using a nonlinear least square minimization. Once the data are fitted, the parameters sensitivity and confidence intervals can then be established enabling the rigorous evaluation of these different techniques to capture the best parameters for a chosen CZM shape.< Réduire
Mots clés en anglais
DCB
Cohesive zone model
Chi-square minimization
Sensitivity
Confidence interval
Origine
Importé de halUnités de recherche