Crack growth monitoring using shear guided waves

Abstract

Crack monitoring in critical sections of steel structures is a subject of growing interest. High frequency ultrasonic techniques have good detection sensitivities but poor inspection coverage per transducer that makes it impractical to monitor large areas. Corrosion detection and monitoring of pipelines can be performed by using low frequency guided waves, but are insufficiently sensitive for the detection of small cracks. This study evaluates the crack monitoring performance of a fundamental shear horizontal mode (SH0) system at frequencies just below the high-order mode cut-off. The scattering solutions valid at low frequencies were developed for both 2D and 3D cracks; most importantly the 3D solution showed that the SH0 reflection ratio is proportional to frequency to the power 1.5, to the effective crack size cubed, and is inversely proportional to the plate thickness and to the square root of the distance from the crack to the receiving sensor. Finite element (FE) analysis was used to validate these power coefficients and to calculate the proportionality constant. The predicted 3D solution was validated using experimental data generated by a ring of transducers on a 6-inch diameter pipe with a progressively grown notch that simulates crack growth; baseline subtraction with temperature compensation was applied to compensate for changes in the environmental and operational conditions. It was found that the residual signals after baseline subtraction can be assumed to be normally distributed so the random fluctuations could be reduced by coherent averaging; it was thereby possible to reliably detect a 2 mm wide and 1 mm deep notch simulating a crack located one pipe diameter along the pipe from the transducer ring. The damage detection performance at different locations along the pipe was assessed by analysing receiver operating characteristic (ROC) curves generated by adding simulated defects to multiple experimental measurements without damage. At a fixed standoff distance, the damage detection performance increases with the square root of the number of averaged signals, and is also improved by averaging the signals received by transducers covering the main lobe of the reflection from the defect; the common source method can be applied to reduce the effects of phase cancellation when using signals from multiple receivers. When the defect is located more than about one pipe circumference from the transducer ring, the optimal performance is obtained by averaging across all the transducers in the ring, corresponding to monitoring the T(0,1) pipe mode. The effects of temperature cycling and the presence of a large reflector, in this case a weld, near a simulated crack were investigated experimentally by data generated from electromagnetic acoustic transducers (EMATs) on a 6-inch diameter pipe with a weld. The results show that the measured reflection ratios after temperature compensation can still be assumed to be normally distributed, similar to the findings from the previous experiment, but there were uncompensated changes which result in larger variations towards the second half of the temperature cycling experiment. A notch was progressively grown near the weld of this same pipe and it was shown that a 1 mm wide and 0.5 mm deep notch located one pipe diameter along the pipe from the EMAT receiver can be detected; a 95% probability of detection with a 1% false alarm rate can be achieved after a sufficiently large number of measurements. In conclusion, an SH0 mode monitoring system with a short transducer standoff distance from the inspected area has good potential for crack monitoring applications, for example for girth welds in pipes, given that transducers designed for stable and long term use are utilised.