A methodology for evaluating detection performance of ultrasonic array imaging algorithms for coarse-grained materials

Abstract

Abstract-Improving the ultrasound inspection capability for coarse grain metals remains of longstanding interest and is expected to become increasingly important for next generation electricity power plants. Conventional ultrasonic A, B, and C-scans have been found to suffer from strong background noise due to grain scattering which can severely limit the detection of defects. However, in recent years, array probes and Full Matrix Capture (FMC) imaging algorithms have unlocked exciting possibilities for improvements. In order to progress and compare these algorithms we must rely on robust methodologies to quantify their performance. This article proposes such a methodology to evaluate the detection performance of imaging algorithms. For illustration, the methodology is applied to some example data using three FMC imaging algorithms; Total Focusing Method (TFM), Phase Coherent Imaging (PCI), and Decomposition of the Time Reversal Operator with Multiple Scattering Filter (DORT MSF). However it is important to note that this is solely to illustrate the methodology; this article does not attempt the broader investigation of different cases that would be needed to compare the performance of these algorithms in general. The methodology considers the statistics of detection, presenting the detection performance as Probability of Detection (POD) and probability of False Alarm (PFA). A test sample of coarse grained INCONEL 625, manufactured to represent materials used for future power plant components and containing some simple artificial defects, is used to illustrate the method on the candidate algorithms. The data is captured in pulse-echo mode using 64 element array probes at centrefrequencies of 1MHz and 5MHz. In this particular case, it turns out that all three algorithms are shown to perform very similarly when comparing their flaw detection capabilities.

Quantification of the Effect of Multiple Scattering on Array Imaging Performance

“…In polycrystalline materials, at frequencies typically used in NDT (below 20 MHz), the absorption is negligibly small [30], so the underlying physical reason of poor detection is the increased multiple scattering. However, quantitative estimates of the detection limit are usually obtained for specific defects by numerical modeling [14] or experimental measurements [4], [31] and without direct relation to the multiple scattering rate.…”

Numerical simulations of ultrasonic array imaging of highly scattering materials

“…This is useful for separately analysing thesignal and noise data which enables monitoring the true SNR. This is a valuable tool in general, 1 as investigations are often limited to measure signals which contain noise, thereby constrained 2 to solely measuring positive SNRs [17] which offer a limited utility as a performance metric. 3…”

Optimization of element length for imaging small volumetric reflectors with linear ultrasonic arrays

“…This cannot be temporally averaged out (unlike random noise), and so it is important than an optimization study is able to model and account for its effect on imaging performance. This research area has been particularly active in recent years with detailed FE Models and first principles studies modelling and describing scattering from individual and distributions of grains in materials and welds [4,6]. Generally these approaches are computationally expensive and impractical particularly when applied to a 3D array inspection optimization problem.…”

“…Considering these assumptions, an equation for the expected Signal-to-Noise Ratio, r), from a given array can be defined, represented as a superposition of the PSF's from each grain in the material volume [7] r) (4) where the numerator describes the "signal" component at an image point r which in this case is that of a point target (a useful approximation for small volumetric flaws) and is a material property that describes the amplitude of the signal in the absence of noise. The denominator represents the noise component with the term in the square root essentially an integration of the PSF's from all grains and a material dependent variable describing the backscatter amplitude as a function of frequency.…”

“…Well proven "rules of thumb" exist for defining most of these parameters based on the wavelength of the ultrasound and the type of flaw required to be detected, yet the element length, L, is often chosen arbitrarily. In the literature, many authors have engaged in parametric studies to optimize array imaging performance but generally the out-of-plane element length is ignored [3][4][5]. This can be attributed to the common assumptions made about scatterer behavior and the element width to length ratio (which are typically defined so that there is little variation in the ultrasonic field perpendicular to the imaging plane).…”