Crack detection characterization of strain sensing grids

This paper presents a new probabilistic framework for evaluating measurements to detect anomalies such as cracks. The new framework accounts for the role of errors and the role of policy parameters such as the maximum allowable rate of false alarms and the required minimum probability of detecting critical anomalies. The framework is applied to the detection of cracks using strain sensing grids. This problem is studied and illustrated by examples from the viewpoint of both analysis and design. In analysis, the crack detection capability of a strain sensing grid is probabilistically characterized. In design, the specification of grid spacing and pattern is determined at various operating conditions that include the role of errors, minimum crack size that must be detected and other parameters. The computational techniques used to study strain sensing grids is discussed in detail. The main results indicate that under plausible operating conditions, a relatively large number of strain sensing stations (more than one: station per three times the square of the minimum crack size to be detected) is needed when the coefficient of variation of the possible errors exceeds 1%. The implications on the required technology for strain sensing grids to detect cracks is briefly discussed. © 1998 Elsevier Science Ltd. All rights reserved