Experiment-based Analysis Of Interfacial Damage In Laminates And Study On Interfacial Mechanical Properties

Due to their excellent fatigue and shock resistance, Fiber Metal Laminates arebeing widely used in large-scale structures such as fuselages. For the production of awide body fuselage, splices are unavoidable in their metal sheets. Experiment findsthat interfacial delamination often initiates where splice exists and this usually leadsto the failure of spliced laminates. It is very necessary to investigate the mechanicalbehavior of interface for the security estimation of laminates. Delamination oflaminates can be simulated with interfacial mechanical model. However, difficultieswill be met when determining interfacial parameters of a practical structure, becausethese parameters usually cannot be obtained accurately by means of simpleexperiment. Based on experimental results, the interfacial mechanical parameters ofa laminate are identified through the inverse solution techniques in this paper. Theidentification integrates experiment and numerical analysis in which an interfaceelement suitable for simulation of delamination damage of laminates is used.First, combined with the bilinear constitutive relation, the cohesive element isapplied in this paper to simulate the interfacial damage of the laminates with doublesplice, single splice and asymmetrical splice. The results indicate that thedelamination initiation and propagation process for laminates near the splice can besimulated successfully by using cohesive elements without premade cracks. Thesimulation results show that the interfacial strength and fracture toughness are themain factors which affect the mechanical response of laminates. Second, based onthe nonlinear mechanical response obtained from experiments for the laminates andin conjunction with the FE numerical analysis for interfacial delamination, twotypical intelligent optimization algorithms, i.e. improved genetic algorithm andRadial Basic Function neural network, are applied to quantificationally identify the interfacial characteristic parameters of the laminates. Genetic algorithm is improvedin terms of elitist strategy and adaptive mutation probability to avoid prematureconvergence of the problem. In order to overcome large amount of calculation ofgenetic algorithm, this paper applies Radial Basic Function neural network definedby three different functions to identify interfacial parameters. The results indicatethat generalization ability of the generalized neural network of Radial BasicFunction is excellent for the problem in this paper, the interfacial parametersobtained by this method is quite close to that by genetic algorithm. Finally,delamination damage behavior and interlaminar stresses for the laminates are furtherdiscussed according to the identified interfacial parameters. It is proved thatexperiment-based hybrid/inverse method can be applied to study the interfacemechanical properties of the laminates effectively.