SQUID NDE and
Probability of Detection

SQUID
(superconducting QUantum interference device) magnetometers are being used as tools for Nondestructive Evaluation (NDE) to detect and characterize defects in aging aircraft. To evaluate SQUID NDE reliability, we have performed a probability of detection (POD). A boundary element method (BEM) measurement model using a Green's function developed specifically for crack problems has been constructed for use in the POD analysis. The model simulates the 2-D images of the magnetic field obtained by scanning a SQUID magnetometer over a plate containing a crack and carrying an injected dc-current. POD curves were generated through Monte Carlo simulation using distributions derived from sensitivity analyses and experimental noise measurements.

The preliminary POD analysis resulted in minimum detectable crack lengths corresponding to 90% probability of detection at 95% confidence of 1.4 mm for the dc-measurement and 0.0134 mm for the ac-measurement. The very small minimum detectable crack length determined for the ac-measurement is due to the large noise reduction through use of the lock-in amplifier in this type of measurement. However, there is a large discrepancy between experimental measurements on real fatigue cracks and the results of this analysis. If the model is to be used to calculate realistic probability of detection values for SQUID systems, more work has to be done to quantify additional sources of noise to be incorporated into the analysis, for example the effects of edges and sample inhomogeneities on detectability.

SQUID Magnetomety Home | back | next