Passive structural monitoring based on matched processing

Abstract

The fourth generation of nuclear reactor designs use liquid sodium as the core coolant. Due to both significant primary loads and thermal variations, important supporting structures are susceptible to thermo-mechanical fatigue. Periodic inspections are planned in order to check the health of the immersed structures, particularly to detect or to characterize potentially hazardous damage. The presented approach relies on both the a priori knowledge of the inspected structures (geometry and properties) and the low-frequency ambient noise in the reactor vessel. Indeed, the latter is used as a source of opportunity to inspect the immersed elastic structures. Two kinds of potential defects could occur: localized defects (e.g., fatigue cracks) and global defects (e.g., deformation of the structure). The feasibility of using data derived from ambient noise for detecting and localizing these defects is carried out on an academic example: a cylindrical shell immersed in water. A cross-correlation based procedure is used to construct the replica and datasets. Then, techniques, such as Matched Field Processing and Matched Mode Processing, are investigated to detect and locate the damage. Both conventional and adaptive processors are implemented and their performance compared. Presented results are obtained from numerical data.