Estimating the probability of detection of cracks in metal plates using lamb waves

Abstract

This paper focusses on the development of a data-driven damage detection method to quantify fatigue crack in metal plates using Lamb waves and its reliability using a probability of detection (POD) technique. The guided Lamb waves are generated and sensed using an array of direct-write (DW) polyvinylidene fluoride (PVDF) annular comb shaped transducers designed to explicitly generate a desired guided wave mode in the test specimen. The annular comb design helps generate a single desired wave mode in the specimen thereby suppressing the energy of other wave modes that can be generated simultaneously. The guided wave responses are obtained through a simulation study and are recorded at different progressions of crack. A damage index (DI) is constructed as a function of crack size that can effectively track the change in ultrasonic response variations and for diagnosing fatigue crack in the metallic specimens. This DI is then further used in the POD model to estimate the crack detection probability. The POD curves can be helpful to check the reliability of the proposed inspection system as well as identify the critical experimental parameters that can significantly influence the crack detection results.