Session: Research Posters

Paper Number: 119762

119762 - Experimental Ultrasound Computed Tomography for Material Characterization Using a Linear Array Pair 

For quality assurance and in-service safety, sufficient characterization of the materials is essential during the development, manufacture, and processing of a material in any industrial setting. Evaluation of elastic coefficients, material microstructures, morphological features, and related mechanical properties are all part of characterizing a material. In the context of material characterization, ultrasound offers various techniques through nondestructive evaluation (NDE). Ultrasonic signals are sensitive to material properties such as wave speeds, attenuation, diffusion backscattering, microstructural variation, and elastic characteristics (e.g., elastic modulus, hardness, etc.). Ultrasound computed tomography (USCT) is an emerging imaging method that can be implemented to obtain a quantitative estimation of material properties. USCT is a tomography-based ultrasound imaging technique that transmits and reflects ultrasonic energy through the object of interest.

In recent years, applying a partial differential equation-constrained, nonlinear optimization technique known as full waveform inversion (FWI) to ultrasound tomography-based imaging (i.e., USCT) in NDE has shown promising results. FWI is based on full wave-field modeling and inversion to extract material parameter distribution using wave equations. The FWI method generates a velocity model by iteratively determining and minimizing the waveform residual, which is the difference between the measured data and the modeled output. FWI evaluates its objective function of data misfit measurements in the time or frequency domain. In FWI, the high dimensional gradient of the objective function is efficiently computed using the adjoint state method. The gradient is evaluated by interacting the back-propagated data misfit with the forward-propagated source wave field. A new updated model is generated from the previous reference model and the computed scaled gradient at each iteration. The iterations continue until the data misfit falls below predefined thresholds or when other convergence measures are met. Such a process requires accurate temporal source function information. Significantly incorrect source information may force the inversion towards a local minimum, severely compromising the inversion and leading to significant artifacts in the resultant velocity model. As a result, providing accurate source information to have a correct forward model is one of the critical components to accurate FWI results.

This study initially proposed a source estimation technique to obtain the source time function for accurate forward modeling by constructing a linear inverse problem for the unknown transducer modeling. Later, a material characterization approach was proposed with accurate source estimation to extract wave speed distribution from an elastic material by employing FWI. Systematic performance analysis of the proposed FWI model with accurate source estimation was assessed using experimental and synthetic data obtained from a 6061-aluminum sample. Overall, the proposed FWI technique has successfully reconstructed the wave speed distribution, exhibiting the potential of the proposed method of material characterization in various engineering applications.

Presenting Author: Md Aktharuzzaman The University of Alabama

Presenting Author Biography: Akthar is a graduate student at the University of Alabama. He is 4th year Ph.D. student in the department of Aerospace Engineering and Mechanics at the University of Alabama, Tuscaloosa. Akthar completed his Bachelor's in Aerospace Engineering from Military Institution of Science and Technology (MIST), Mirpur, Bangladesh in 2017. He joined the Computational Imaging and Smart Structure (CISS) lab on 2019 as a graduate student. Akthar is experienced in ultrasound data acquisition and processing, transducer prototyping, finite element and spectral element based modeling, numerical optimization, and computational imaging based material characterization. He enjoys outdoor activities, fishing, and traveling.

Authors:

Md Aktharuzzaman The University of Alabama
Shoaib Anwar The University of Alabama
Dmitry Borisov The University of Kansas
Jiaze He The University of Alabama