Increasing economic pressure has encouraged extension of the service lives of many civil and military aircraft fleets beyond their original design limits. One significant consequence of aging aircraft is corrosion, which has become one of the main maintenance cost drivers for the aircraft industry.
Coating systems on aircraft serve as the first line of defence against corrosion. One issue related with aircraft coatings is the degradation when exposed to the environment (chemical degradation) and mechanical stress (mechanical degradation). The interaction of chemical and mechanical stresses often produces accelerated degradation, which is known as mechano-chemical degradation. In service, it is common for aircraft paint to fail by cracking or chipping at points of high strain, such as joint areas.
This thesis reports on the development of a special purpose finite element simulation program. The program combined augmented finite element method (AFEM) and cohesive zone model (CZM) to model initiation and propagation of both cohesive and adhesive cracks in heterogeneous coating systems. A linear elastic material model was used for the degraded polyurethane matrix. Three factors were involved in the failure mechanism of a cohesive crack, i.e. crack initiation/propagation criterion, crack extension direction and the cohesive law for the matrix. The principal strain criterion was employed in our study as the fracture initiation/propagation criterion for cohesive cracks. The maximum principal stress criterion was employed to predict the cohesive crack growth direction in the polyurethane matrix. The non-local stresses were used to calculate the principal direction. The failure mechanism of an adhesive crack was fully controlled by a cohesive law for the material interface. An experimental approach that used a double cantilever beam sandwich specimen to determine the cohesive law was presented. A few aspects of numerical implementation of the simulation tool were discussed, including the general flowchart of the program, calculation and assemblage of element stiffness matrix, solution of non-linear equilibrium equations, calculation details of crack extension direction, and the augmentation scheme of AFEM.
The performance of the simulation tool was validated through both numerical and experimental approaches. Two numerical examples were modelled to simulate the cohesive crack propagation. The double cantilever beam (DCB) model verified the prediction of crack length in both ceramic and polymer materials by comparing the program predictions with analytical solutions. The effect of convergence criterion and element size on solution accuracy, stability and CPU time was also investigated using the DCB model. A beam under three-point bending was modelled to predict the path of an eccentric crack in the beam. The crack propagated towards the loading point along a curved path as expected. It was compared with results from literature and the paths closely agreed with each other. An experimental compact tension test was employed to validate the prediction for adhesive cracks. Inverse finite element method was used to determine the interface properties. The predicted failure extension with the proposed CZM curve agreed with the experimental result.
The effects of filler particles on the coating lifetime were studied, including particle size, shape, volume fraction, and particle/matrix interface adhesion with idealised particle sizes and shapes. Results indicated that low volume fraction, high interface adhesion and smaller size particles improve coating durability. Among the three factors, i.e. filler size, shape and volume fraction, the filler volume fraction had the most prominent influence on coating lifetimes while the filler shape had the least. Sensitivity study indicated that the simulation results with the idealised filler particles had an average error percentage of 5% caused by different distributions of filler particles. Coatings with the elliptical particles had a slightly larger confidence interval than that with round particles probably due to the influence of the orientation of the elliptical particles. A real coating system was also studied and the lifetime was predicted, which was a bit longer than predictions for coatings with identical filler sizes and idealised filler shapes, i.e. 5% longer than coatings with round particles and 3% longer than elliptical particles.
The significances of this Ph.D. research are twofold. From an industrial perspective, it is economically attractive to develop a tool for prediction of coating degradation. From a scientific point of view, it is imperative to acquire a fundamental understanding of coating failure mechanisms and the roles of the various parameters (e.g. filler size, shape, volume fraction and interface properties) which can influence the rate of coating degradation. The output of this research will make an important and fundamental contribution to the development of the structural health management system for aircraft in the future.