Search

One of the most common methods for performing non-destructive testing (NDT) of the steel tank floors in aboveground storage tanks is dc magnetic flux leakage (MFL). This test method gives an estimate of the defect depth and width based on the MFL signal strength and its peak location, respectively. A key limitation is that the signal strength depends on not only the defect depth and width but also various other factors including a defect’s wall profile and sensor lift-off. Moreover, in a practical MFL test, the sensor lift-off changes due to surface roughness and uneven steel plate surface in tank floors. We present an ac MFL system to increase the accuracy of defect characterization by: 1) distinguishing between two common defect shapes located on the far side of the steel plate used in a typical above storage tank (AST) floor: lake-shaped and rectangular defects and 2) developing a sensor lift-off compensation scheme based on ac signal phase. Simulation results show that skewness of the ac MFL can be used to distinguish between far-side lake-shaped and rectangular defects. AC MFL signal phase is shown to be a suitable compensator for the dependence of signal strength on unintended variations in sensor lift-off. The simulation results have been validated through experiments. [ABSTRACT FROM AUTHOR]

One of the most common methods for performing non-destructive testing in steel tank floors is DC magnetic flux leakage (MFL). The magnetic dipole method is the most widely used mathematical technique to predict the MFL from defects in such structures. However, due to the complexity of an exact analytical description of an MFL system, researchers often make coarse approximations for the profile of the magnetic surface charge density $\sigma _{m}$ , orientation of the magnetic field H, and variation of relative permeability $\mu _{r}$. In this paper, the validity of these approximations is evaluated for 2-D rectangular defects in a steel plate, by comparing model predications with finite element results. The primary sources of deviation between the approximate solutions and true MFL profiles were found to be caused by assumptions that 1) $\sigma _{m}$ on the specimen surface adjacent to a flaw is zero. This assumption is equivalent to treating the orientation of H to be parallel to the specimen surface, even at locations in close proximity to a flaw and 2) local variation in permeability around the defect can be ignored. This approximation was found to cause an underestimation of $\sigma _{m}$ and, consequently, the predicted MFL. In contrast, approximating $\sigma _{m}$ to be zero at the bottom of a flaw, and approximating uniform distribution for $\sigma _{\mathbf {m}}$ on the vertical defect sides of a slot defect was found to generate only minor errors in an estimate of flux leakage. [ABSTRACT FROM AUTHOR]