A novel closure based approach for fatigue crack length estimation using the acoustic emission technique in structural health monitoring applications
A novel closure based approach for fatigue crack length estimation using the acoustic emission technique in structural health monitoring applications
复制标题
DOI：
10.1088/0964-1726/23/10/105033
复制
发表时间：
2014-10-01
影响因子：
4.1
通讯作者：
Irving, Philip
中科院分区：
文献类型：
--
作者：
Gagar, Daniel;Foote, Peter;Irving, Philip
使用声发射 (AE) 检测和定位金属结构中的疲劳裂纹已被广泛报道，但调查其疲劳裂纹长度估计潜力的研究却很少。裂纹扩展信息可以利用完善的断裂力学原理来预测部件的剩余使用寿命。因此，如果AE不仅可以作为检测裂纹扩展的手段，而且还可以用于估计裂纹长度，那么它在结构健康监测应用中的前景将得到显着改善。基于疲劳裂纹扩展期间产生的声发射信号与相应循环载荷之间的相关性，开发了一种推导裂纹长度的新方法。裂纹长度计算模型是使用 2 mm 厚的 SEN 铝 2014 T6 样本疲劳裂纹扩展测试中生成的 AE 数据凭经验得出的，拉伸应力范围为 52 MPa，R 比为 0.1。该模型使用在 27 MPa 应力范围内进行的单独测试中独立生成的 AE 数据进行了验证。结果表明，可以得到 10 mm 至 80 mm 范围内的裂纹长度预测，归一化绝对误差平均值在 0.28 至 0.4 之间。还使用现有的基于 AE 特征的方法进行预测，并将结果与使用开发的新方法获得的结果进行比较。
Use of Acoustic Emission (AE) for detecting and locating fatigue cracks in metallic structures is widely reported but studies investigating its potential for fatigue crack length estimation are scarce. Crack growth information enables prediction of the remaining useful life of a component using well established fracture mechanics principles. Hence, the prospects of AE for use in structural health monitoring applications would be significantly improved if it could be demonstrated not only as a means of detecting crack growth but also for estimation of crack lengths. A new method for deducing crack length has been developed based on correlations between AE signals generated during fatigue crack growth and corresponding cyclic loads. A model for crack length calculation was derived empirically using AE data generated during fatigue crack growth tests in 2 mm thick SEN aluminium 2014 T6 specimens subject to a tensile stress range of 52 MPa and an R ratio of 0.1. The model was validated using AE data generated independently in separate tests performed with a stress range of 27 MPa. The results showed that predictions of crack lengths over a range of 10 mm to 80 mm can be obtained with the mean of the normalised absolute errors ranging between 0.28 and 0.4. Predictions were also made using existing AE feature-based methods and the results compared to those obtained with the novel approach developed.