Experimental validation of a novel structural damage detection method based on empiricalmode decomposition

Abstract

Structural health monitoring, as a non-destructive damage detection technique, has beenadopted by many researchers as well as various industries to keep up with the increasingdemand for preventive maintenance routines. Oil and gas is one of the major industries inwhich effective damage detection can ensure reliable transportation of chemicals andprovide tremendous savings in maintenance costs. The present work investigates theintegrity of a novel damage detection method through a set of experiments conducted ona standard steel pipe commonly used in the oil and gas industry. The methodinvolved collection of the vibration signature of the pipe via certain sensors, anddecomposition of the signature (signals) by empirical mode decomposition. This wasfollowed by the use of a damage index defined on the first intrinsic mode functionresulting from the decomposition. The experiment consisted of a cantilever steelpipe equipped with piezoceramic sensors to monitor the free vibration of a pipeimpacted by an impulse hammer. Three damage cases were studied including singlehalf-circumferential, single full-circumferential, and multiple circumferential andhalf-circumferential damages. A MATLAB code was developed to process and todecompose the signals according to the proposed methodology. The results showed that theproposed damage detection methodology could successfully detect the presence andlocation of damage, and could also identify the severity of the damage. A set ofexperiments was also done to examine the effect of support flexibility on thedamage index, and to examine the sensitivity of the method to support conditions.