-7- oscillation of SERR curve appears again (e.g. 0 c = ) and the SERR reduces to a relatively low value. This behavior seems to be a result of the asymmetric loading and restraint causing by the offset of interface defect. An asymmetric loading and restraint may induce the extra deviating force for the surface crack. Therefore, the driving force would continue to increase instead of dropping as the crack approaches to interface, which may facilitate the deviation of the surface crack when it approaches to interface. Obviously, it can explain why the SERR curve for / a h= 0.8 and 0.9 exhibit irregular convex at 0.05 c =± .Interestingly, this deviation has been observed by present researchers in their experiments[6]. If the defect is remote from the surface crack, its influence on surface crack is much small so that the deviating force is inconspicuous. Therefore, for the remote defect (e.g. 0.15 c = ), the curve of SERR do not appear monotonous increasing. Figure 4. The reduced revised function ( , , ) L a c κ as a function of interface defect length for different surface crack length. The reduced revised function ( , , ) L a c κ as a function of normalized surface crack length for different interfacial defect length is plotted in Fig. 4. In present figure, the interface defect offset c has been fixed as 0.05 and the surface crack length has been enlarged in order to simulate the spread of interface defect during the service of TBCs. Since the interface defect location has been fixed, the altering of ( , , ) L a c κ , which is obtained, as mentioned earlier, by the normalized SERR, can separately demonstrate the effect of defect length. In Fig. 4, it is seen that interfacial defect length is also a significant factor that affects the surface crack driving force. For a small interface defect, the interface defect seems to have little influence to the surface crack. Noted that when the L is about 0.05, the value of SERR is close to the solution for perfect interface (the dash line in Fig. 4), which indicates that there may be an asymptotic curve for very low values of interface defect length. In other words, in present model, when the normalized length of interface defect with offset 0.05 c = is smaller than 0.05, the interface can be regarded as a perfect one. For a relatively large interface defect, the value of ( , , ) L a c κ rises to a high degree. For example, when the surface crack propagates about seventy percent of the ceramic coating, the ( , , ) L a c κ for the 0.3 L = is about 3.3, which indicate an interface defect with a length 0.3 L = would make the driving force of surface crack more than three times larger since the substrate restrain is greatly reduced due to the interface defect. Obviously, the facilitation of the large scale interface defect may induce the premature coalescent of the surface crack and the interface crack, which contributes to the spalling of the ceramic coating. Therefore, a well bonded interface should be ensured in order to enhance the durability of the TBCs. Compared to the Fig. 2, it seems that the effect of lengthening the interface defect length on the surface crack driving force is very similar to the effect of diminishing the offset of the interface
RkJQdWJsaXNoZXIy MjM0NDE=