| Thermal barrier coating is a high-temperature thermal protection material applied to the surface of aero-engine components,which is one of the key technologies to further improve the temperature resistance of turbine blades and extend the aero-engine service life.However,conventional thermal barrier coating materials partially Y2O3 stabilized ZrO2(YSZ)will undergo phase transformation,sintering,and oxygen permeation in the high-temperature environment above 1200℃,leading to the premature failure of coatings.Thus,the development of a new thermal barrier coating system is an urgent requirement and strategic goal to further enhance the turbine intake temperature and promote the engine to higher efficiency and thrust-to-weight ratio.Among them,double ceramic layer thermal barrier coating systems(DCL TBCs)are considered to be an effective way to overcome the shortcoming of conventional YSZ coatings,increase the service temperature,enhance the thermal protection capability and improve the thermal cycle life of coatings.Therefore,this thesis focuses on DCL TBCs and conducts failure analysis and performance optimization study of DCL TBCs.The main work of this thesis includes the stress field analysis of DCL TBCs,the failure analysis of the cracks near the interfaces,the comprehensive performance optimization,and the intelligent prediction of coating performances.The specific work and the main findings are as follows:Stress concentration is the most important factor for the propagation of cracks in coatings,so an accurate assessment of the stress field is the basis for failure analysis of DCL TBCs.Therefore,LaZr2O7(LZ)/YSZ DCL TBCs were selected as the research object in this thesis to carry out the study on the stress field analysis.The influence of the thickening of the Thermally Growth Oxide(TGO)layer,creep,plastic deformation,geometry parameters,and interfacial roughness on the durability performance of the coating were considered in the analysis,and the effects of interfacial action and interfacial roughness parameters on the interfacial stresses under the two typical ceramic layer thickness distributions were systematically investigated.The results show that the interfacial morphology of the YSZ/TGO interface affects the normal stress at the LZ/YSZ interface when the thickness of YSZ is below 100 μm,and the normal stress at the peak of the LZ/YSZ interface gradually decreases with the increase of the horizontal distance between the peaks of the two interfaces.Besides,the maximum tensile stress at the LZ/YSZ interface and the YSZ/TGO interface increases with the increase of the interface roughness parameter,but the position of the maximum positive stress at the LZ/YSZ interface cannot change and is always located at the peak,which leads to the crack initiation near the interface.However,for the YSZ/TGO interface,there is a critical wavelength for a certain interface amplitude.When the interface wavelength of YSZ/TGO is lower than the critical wavelength,the position of the maximum positive stress at the interface is located at the peak;otherwise,it will be located near the peak,which results in a more complicated failure mechanism and cracks morphology.To investigate the failure mechanism of DCL TBCs,based on the virtual crack closure technique,this thesis investigates the evolution of the driving forces(the strain energy release rate)for horizontal cracks near the interfaces,and discusses the effects of interfacial geometric parameters,crack size,and crack location on the crack propagation ability.Besides,the propagation capacity of horizontal cracks in the presence of surface crack was also discussed.The results show that cracks closer to the interface are more likely to propagate.Meanwhile,the horizontal cracks will propagate once the conditions are met,and the crack length increases.When the cracks spread above the interface valley,the crack driving force will be reduced,and the cracks will no longer propagate.In addition,horizontal cracks near the LZ/YSZ interface are more tend to form large-scale cracks.Long surface cracks have positive significance in mitigating the propagation of cracks,while when the coalescence of two type cracks occurs,the propagation ability of interfacial cracks can be significantly strengthened,which accelerates the propagation of interfacial cracks,and leads to localized peeling and delamination failure of systems.In addition,the process of crack propagation and the failure mechanism of DCL TBCs were also described based on analysis results,which are in good agreement with the experimental phenomena.It is believed that this thesis can provide an effective theoretical basis for the design of high-durability coatings.The residual stress level and thermal insulation performance are two important performance indicators that need to be focused on when applying thermal barrier coating systems,but they often contrast with each other,so improving one aspect of the performance of TBCs has limited effect on the practical application of DCL TBCs.For this reason,according to the characteristics of TBCs,this thesis proposes a multiobjective optimization workflow combining the back-propagation neural network(BPNN)and constrained non-dominated sorting genetic algorithm-II(Constrained NSGA-Ⅱ)based on the results of theoretical and numerical models.The workflow considers the constraints of YSZ operating temperature and the sintering temperature of top coatings,which demonstrates that it is possible to find comprehensive and excellent coating structure design solutions for various DCL TBCs.In this study,the multi-objective optimizations were performed for LZ/YSZ,LaPO4/YSZ,and La2(Zr0.7Ce0.3)2O7/Y SZ DCL TBCs,and the Pareto fronts and recommended structural design references of these systems were obtained.In addition,correlation analysis was also utilized to quantify the effects of top coating properties and structural parameters on the optimization objective,which can explain the differences in the Pareto fronts of the three DCL TBCs and provide a reference for finding a new generation of top coating materials.Accurate prediction and evaluation of thermal barrier coating properties is the fundamental and critical issue for improving the performance of coatings.In recent years,data-driven models based on deep learning methods have shown the outstanding potential for discovery,characterization,and preparation of materials.However,datadriven models based on 3D convolutional neural networks(CNNs)for 3D features often need more training parameters and higher computational costs,which limits the industrial applications of such models.To this end,a novel data-driven model based on projection features was proposed in this thesis,which can greatly reduce the training scale of the model while ensuring the prediction accuracy by compression of 3D features before learning the model.In this paper,the model was also used to achieve the prediction of the effective thermal conductivity of an atmospheric plasma spray(APS)thermal barrier coating system.The results show that the model exhibits superior prediction accuracy compared to other theoretical models.In addition,when the model was applied to other two-phase composites,it is found that the model can still show good prediction performance,indicating that the model has great scalability and is expected to be applied to the prediction of other material structure-related properties. |
PDF Full Text Request