Enhancing damage localisation in critical areas of storage tanks in floating production storage and offloadings with a guided wave structural health monitoring system

Abstract

A guided wave monitoring system for storage tanks of floating production storage and offloading units was tested on a panel which is representative of a critical region of such structures. Initially, baseline signals of the undamaged structure were acquired within a representative temperature range. These signals were then processed with the optimal baseline subtraction, optimal stretch and phase-shift compensation algorithms. Two distinct damage indexes were generated from the residual signals in order to verify the detection capability of the system. A delay-and-sum image algorithm was then applied to the residual signals in order to evaluate the defect localisation capabilities of the system. These images were than post-processed with three different methods, one based on a probabilistic approach, another on singular value decomposition and also with a combination of these two methods. The signal-to-noise ratio (SNR) gain obtained by applying the post-processing strategies was calculated and compared to the original images. Results demonstrate that the system is capable of detecting localised wall loss and indicating its correct position for damage larger than 20% wall thickness loss. Results also show that the post-processing methods helped to improve the SNR, leading to gains of more than 15dB in damage to artefact amplitude ratios.

Guided Wave SHM system for detection and quantification of damages in FPSOs storage tanks using data-driven algorithm

“…In this work, we proposed a method for detecting and quantifying damage related to thickness reduction in complex and difficult-to-access structures, using the outlier detector algorithm in a guided wave SHM system, permanently installed in a structure. Furthermore, this study complements the work of Menin et al, 40 who had developed a technique for locating damage in complex structures using the same structure and the same experimental data.…”

“…Finally, because the developed method does not perform damage localization, it is not possible to define the best pair of sensors. However, we can use the algorithm developed by Menin et al, 40 for example, which performs the localization of damage through a classical image. From the damage location, we will possibly be able to choose the best pair of sensors to perform the damage quantification through the SNR.…”