Thin material layers have found various applications with various roles of functions, such as in fibre reinforced laminated composite materials, in integrated electronic circuits, in thermal barrier coating material system, and etc. Interface delamination is a major failure mode due to either residual stress or applied load, or both. Over the past several decades, extensive research works have been done on this subject; however, there are still uncertainties and unsolved problems. This thesis presents the new developed analytical studies on local delamination failure of thin material layers. Firstly, the analytical theories are developed for post-local buckling-driven delamination in bilayer composite beams. The total energy release rate (ERR) is obtained more accurately by including the axial strain energy contribution from the intact part of the beam and by developing a more accurate expression for the post-buckling mode shape. The total ERR is partitioned by using partition theories based on the Euler beam, Timoshenko beam and 2D-elasticity theories. By comparing with independent test results, it has been found that for macroscopic thin material layers the analytical partitions based on the Euler beam theory predicts the propagation behaviour very well and much better than the others. Secondly, a hypothesis is made that delamination can be driven by pockets of energy concentration (PECs) in the form of pockets of tensile stress and shear stress on and around the interface between a microscopic thin film and a thick substrate. Both straight-edged and circular-edged spallation are considered. The three mechanical models are established using mixed-mode partition theories based on classical plate theory, first-order shear-deformable plate theory and full 2D elasticity theory. Experimental results show that all three of the models predict the initiation of unstable growth and the size of spallation very well; however, only the 2D elasticity-based model predicts final kinking off well. Based on PECs theory, the room temperature spallation of α-alumina oxidation film is explained very well. This solved the problem which can not be explained by conventional buckling theory. Finally, the analytical models are also developed to predict the adhesion energy between multilayer graphene membranes and thick substrates. Experimental results show that the model based on 2D elasticity partition theory gives excellent predictions. It has been found that the sliding effect in multilayered graphene membranes leads to a decrease in adhesion toughness measurements when using the circular blister test.

[pt] O aumento dos custos relacionados aos processos de manutenção em estruturas como aeronaves, aliadas à crescente demanda das mesmas, alimentam a necessidade de investimentos em técnicas inovadoras de monitoramento estrutural. Dessa forma, o trabalho realizado nesta tese, busca o desenvolvimento de uma técnica de monitoramento ativo, visando o acompanhamento de parâmetros da estrutura analisada, a fim de identificar e caracterizar processos de dano não visíveis, tais como corrosão e delaminação. A metodologia empregada, teve como base a análise dos padrões de deformação superficial, obtidos com o uso de grades de sensores à fibra óptica baseadas em redes de Bragg (FBG). Inicialmente, tais padrões foram provocados por carregamentos estáticos (tração), e posteriormente por atuadores PTZ fixados à estrutura. Estes últimos são submetidos a uma voltagem alternada e frequência fixa. Esta técnica apresenta todas as vantagens dos sensores FBG (massa e dimensões reduzidas, imunidade eletromagnética, elevado poder de multiplexação e alta sensibilidade entre outras), alem de permitir a visualização de alterações nos padrões de deformação, provocados por danos, através da variação da frequência de excitação. Com relação à interpretação dos resultados, a estratégia empregada consistiu em separar o problema de detecção e caracterização dos danos. Dessa forma, a detecção é realizada comparando a energia das deformações superficiais dos corpos de prova nos casos com e sem defeito, enquanto a caracterização é obtida através a utilização de redes neurais artificiais (RNA), por meio de rotinas de reconhecimento de padrões. / [en] The higher costs related to maintenance processes in structures such as aircraft, coupled with the growing demand of them, fueling the need for investment in innovative techniques for structural monitoring. Thus, the work done in this thesis seeks to develop a technique of active monitoring, aiming at monitoring of structure parameters analyzed in order to identify and characterize processes of hidden damage such as corrosion and delamination. The maid methodology was based on the analysis of patterns of surface deformation, obtained with the use of nets of optical fiber sensors based on fiber Bragg gratings ( FBG ). Initially, these patterns were caused by static loads (tension ), and later by PTZ actuators fixed to the frame, who are subjected to an AC voltage and fixed frequency. This technique has all the advantages of the FBG s sensors (mass and small dimensions, electromagnetic immunity, high multiplexing s power and high sensitivity among others), in addition to allowing visualization of changes in the patterns of deformation caused by damage, by varying the frequency excitation. With respect to the interpretation of the results, the strategy employed was to separate the problem of detection and characterization of damage. Thus, the detection is performed by comparing the deformation energy of the surface of the specimens in the cases with and without defect, whereas the characterization is obtained through the use of artificial neural networks (ANN) by means of pattern recognition routines.