Finite element model driven rapid localization method for multi-damages on pressure vessels

Torispherical structures are one of the main head or bottom cover parts of special equipment such as pressure vessels. Conventional nondestructive testing methods are not quite applicable for online damages inspection at the head or bottom part of large storage tanks in-service. Piezoelectric transducers (PZT) are capable for both exciting and receiving Lamb waves to scan large surface areas and detect damages. This paper constructs a compact uniform circular array using PZTs and presents a virtual array sparse feature based multi-damage localization method for hemispherical head. Firstly, array Lamb wave signal propagating on hemispherical structures is modeling for constructing array sparse features using the array steering vector. Secondly, a finite element analysis model of pressure vessel head was created in ABAQUS to construct virtual array sparse feature library. Finally, their locations can be quickly determined by damages imaging using similarity compared the array steering vector of damage signals with the virtual array sparse feature library. Both the numerical and experimental results verify the virtual array sparse feature modeling based multiple damages localization method can effectively monitor axisymmetric structures with high accuracy.