Noritaka Yusa

Large-scale parallel computation for the reconstruction of natural stress corrosion cracks from eddy current testing signals

An inversion algorithm for the reconstruction of cracks from eddy current signals is developed in this study and applied to the profile evaluation of natural stress corrosion cracks that were found in steam generator tubes of a nuclear power plant. A crack is modeled as an assembly of small regions having conductivities inside so that eddy currents that flow across the cracks are considered. The conductivity of each region, which is assumed to be a discrete value, is reconstructed by means of the algorithm. Since the algorithm is based upon a tabu search that usually requires a large number of evaluating solution candidates, simulations are carried out on a supercomputer with the use of parallel computation using up to 128 CPUs so as to reconstruct the crack profiles within a reasonable computational time. It is demonstrated that the algorithm can estimate the profiles of the natural cracks with sufficient accuracy. The simulations also show that the algorithm is highly compatible with parallel computation. Additional simulations using other models of natural cracks are performed. Reconstructed profiles of the natural cracks, as a notch with zero conductivity, are very different from the true profiles, even though the reconstructed signals agree well with the measured values. This reveals that it is necessary to take the internal conductivity into consideration when dealing with natural cracks.

Shape reconstruction of multiple cracks from ECT signals by means of a stochastic method

This paper presents an approach to reconstruct multiple cracks from eddy current testing signals by means of a stochastic method such as tabu search. The size, number, and locations of cracks are considered unknown in the inversion process, whereas only the orientation of cracks is supposed to be revealed in advance. Eddy current inspections of a stainless steel plate with closely located EDM slits are performed, and the inversions of measured signals are carried out. The results agree well with the true ones, which validates the proposed algorithm

Eddy current inspection of closed fatigue and stress corrosion cracks

This study evaluates the effect of loading on closed crack openings on eddy current signals caused by the crack. Three plate specimens having a fatigue crack and another three specimens having stress corrosion cracking are prepared. Four-point bending of the specimens is carried out to introduce compressive stress to close the opening of the cracks. Then eddy current signals due to the crack are gathered using a plus point type probe. The bending was conducted with a variable load; the eddy current signals remain almost unchanged regardless of the loading.

Fast simulation of ECT signal due to a conductive crack of arbitrary width

This paper proposes a strategy for the fast simulation of eddy current testing signals due to a conductive crack of arbitrary width. To cope with a crack of width less than that selected for establishing the database, which is necessary in the fast-forward analysis scheme proposed by authors, a new finite element is introduced to treat the case when the crack boundary is contained in the element. By using such a new element, the fast-forward analysis scheme becomes suitable for the reconstruction of both the shape and width of a planar crack. It is verified that such a multiple material element is efficient for an ECT sensor inducing eddy current parallel with the crack surface. For the case with perpendicular eddy current component, however, the approach is not valid because of a scalar potential jump at the crack surface. Finally, the reason of such a singularity is investigated through numerical simulations

Profile reconstruction of simulated natural cracks from eddy current signals

Based upon the fact that internal conductivity of the crack has a serious effect on the eddy current signals, two kinds of inversion schemes that can reconstruct cracks with non-vanishing conductivity from eddy current signals are proposed in this paper. One is a physics-based approach that modifies crack parameters iteratively and the other is a neuronet-based approach. Two models that are supposed to be appropriate models of a natural crack are also proposed. After detailed explanation of the schemes, several reconstruction results using simulated data are presented. The results show these schemes can reconstruct crack profile with high accuracy where not only shape of the crack but also internal conductivity is unknown. Advantages and disadvantages of both schemes are discussed at the end of the paper.

Some advances in numerical analysis techniques for quantitative electromagnetic nondestructive evaluation

In this paper, some progresses in numerical techniques mainly made in our research group for the forward and inverse simulation of electromagnetic nondestructive evaluation (ENDE) signals are introduced. For the first part, efficient forward analysis schemes for the simulation of eddy current testing (ECT), remote field ECT (RFECT) and magnetic flux leakage testing (MFLT) signals are described respectively, in addition to some numerical examples. Fast and accurate ECT signal simulation is realised by introducing a database type strategy using precalculated unflawed potential field data. To meet the high accuracy requirement of the simulation of RFECT signals, a hybrid scheme using 2D and 3D geometry and a new formula for pickup signal are proposed. To improve the efficiency of MFLT signal simulation, a fast scheme is developed based on a FEM–BEM hybrid code of polarisation method. In addition, a phenomenological method is also described in the first part, which is developed for the qualitative estimation of eddy current distribution and pickup signals. The second part of this review paper is on the reconstruction of defect from the detected ENDE signals (mainly ECT signals). Reconstruction schemes based on conjugate gradient (CG) method of deterministic category and NN method, metaheuristic methods of stochastic category are developed and sizing of both artificial and natural cracks are performed by using measured signals. It is clarified through applications that a deterministic optimisation method is more efficient for treating simple cracks, while a stochastic way is prefer for defects of complicated geometry such as a stress corrosion crack and multiple cracks. In the crack modelling and parameterisation, an element of discontinuous material property is introduced to treat crack of arbitrary shape based on a given regular mesh. Several numerical models are proposed for natural cracks, which makes the reconstruction of some natural cracks become possible.

Generalized neural network approach to eddy current inversion for real cracks

This paper proposes a generalization for the solutions of an inversion method developed by the authors. The method is based upon an artificial neural network that simulates mapping between eddy current signals and crack profiles. One of the biggest advantages of the approach is that it can deal with conductive cracks, which is necessary to reconstruct natural cracks. However, it has one significant disadvantage: the reliability of reconstructed profiles was unknown. This paper also proposes a novel parameter that provides an index for assessment of the crack profile and overcomes this disadvantage. After the parameter is validated by reconstruction of simulated cracks, it is applied to reconstruction of natural cracks that occurred in steam generator tubes of a pressurized water reactor. It is shown that the parameter is applicable to not only simulated cracks but also real (natural) ones.

Development of computational inversion techniques to size cracks from eddy current signals

This paper reviews the recent advance in computational inversion techniques to size defects using eddy current signals. Studies reporting the sizing of artificial slits with the aid of computational inversion techniques are briefly overviewed and then several recent studies dealing with the sizing of fatigue and stress corrosion cracks are introduced. The results of the studies confirm computational techniques for sizing slits are sufficiently mature, and even though the measured signals are polluted with high noise, the boundary profile of slits can be evaluated from the signals. Furthermore, the studies demonstrate fatigue cracks can be dealt with almost identically to artificial slits from the viewpoint of eddy current testing; fatigue cracks introduced inside weld with a rough surface are well sized using eddy current testing with the aid of computational inversion techniques. In contrast, there is a large discrepancy between artificial slits and stress corrosion cracks, and thus conventional approaches are not always applicable to the sizing of stress corrosion cracks. This paper introduces several recent activities in the authors group to overcome this problem.

A nondestructive strategy for the distinction of natural fatigue and stress corrosion cracks based on signals from eddy current testing

In this paper, a novel nondestructive strategy is proposed for distinguishing differences between a stress corrosion crack (SCC) and a fatigue crack (FC) based on signals from eddy current testing (ECT). The strategy consists of measurement procedures with a special ECT probe and crack type judgment scheme based on an index parameter that is defined as the amplitude ratio of the measured signals. An ECT probe, which can induce eddy current flowing mainly in a selected direction, is proposed and applied to detect crack signals by scanning along the crack with different probe orientations. It is clear that the ratio of the amplitudes of signals detected for parallel and perpendicular probe orientations is sensitive to the microstructure of the crack, i.e., the parameter is much bigger for a fatigue crack than that of a SCC. Therefore, whether a crack is a SCC or a FC can be recognized nondestructively by comparing the index parameter with a threshold value that can be previously determined. In order to verify the validity of the proposed strategy, many artificial SCC and FC test pieces were fabricated and ECT inspections were performed to measure the corresponding crack signals. Numerical simulations were also conducted to investigate the physical principles of the new methodology. From both the numerical and experimental results, it is demonstrated that the strategy is very promising for the distinction of artificial SCC and FC; there is also good possibility that this method can be applied to natural cracks if the threshold value can be properly determined.

Reducing Joint Resistance by Heat Treatment During Fabrication of a Mechanical Joint of High-Temperature Superconducting Conductors

A mechanical joint of high-temperature superconducting (HTS) conductors has been investigated for segmented fabrication of HTS magnets proposed for a future fusion reactor. In a previous study, joint resistance of 1.8