Measuring stress variation with depth using Barkhausen signals

Magnetic Barkhausen noise analysis (BNA) is an established technique for the characterization of stress in ferromagnetic materials. An important application is the evaluation of residual stress in aerospace components, where shot-peening is used to strengthen the part by inducing compressive residual stresses on its surface. However, the evaluation of the resulting stress-depth gradients cannot be achieved by conventional BNA methods, where signals are interpreted in the time domain. The immediate alternative of using x-ray diffraction stress analysis is less than ideal, as the use of electropolishing to remove surface layers renders the part useless after inspection. Thus, a need for advancing the current BNA techniques prevails. In this work, it is shown how a parametric model for the frequency spectrum of Barkhausen emissions can be used to detect variations of stress along depth in ferromagnetic materials. Proof of concept is demonstrated by inducing linear stress-depth gradients using four-point bending, and fitting the model to the frequency spectra of measured Barkhausen signals, using a simulated annealing algorithm to extract the model parameters. Validation of our model suggests that in bulk samples the Barkhausen frequency spectrum can be expressed by a multi-exponential function with a dependence on stress and depth. One practical application of this spectroscopy method is the non-destructive evaluation of residual stress-depth profiles in aerospace components, thus helping to prevent catastrophic failures.