Ultrasonic-Based Condition Assessment of Wooden Utility Poles
[Thesis]
Lee, Yishi
Mahoor, Mohammad
University of Denver
2020
180
Ph.D.
University of Denver
2020
More than 300 million utility poles shoulder the utility grid in the United States. However, the ineffectiveness of the current inspection process causes roughly a third of utility poles removed from the service deemed suitable for reuse. Due to the utterly essential role of the power infrastructure, budget shrinkage, and the structural degradation of the modern distribution grid, this Ph.D. dissertation addresses the challenges by proposing a physics-based signal analysis method with a jointly developed ultrasonic UB1000 system to enhance the objectivity in ultrasonic-based nondestructive evaluation (NDE). The proposed methodology has been deployed commercially in the field and featured in articles by the Missouri Public Utility Alliance and the Western Cooperative Electric. This dissertation proposes an embedded waveguide as an ultrasonic radiation source. A systematic analytical model is developed based on the classical elastodynamic formulation to study the excitation and the propagation characteristics of the resulted elastic wave. Based on the steady-state assumption with a set of half-space boundary and interface loading conditions, the obtained closed-form displacement field yields the diffusive property of the shell region propagation as a function of the Poisson's ratio. This diffusive property is discovered under the quasi-steady load condition, a reasonable model to describe the behavior of a narrow-band ultrasonic transducer. The estimated diffusive propagation is demonstrated through the numerical finite element method (FEM). This study developed the first high-fidelity numerical model of a wooden pole cross-sectional region. It is capable of modeling a porous orthotropic medium under the cylindrical symmetry enabling high moisture content and/or incipient decay conditions to be simulated. Using a transient imposed boundary condition, the model uncovers different arrival wave modes resulted from propagating in various regions within the cross-section. By dissecting the waveform and isolating the corresponding arrival wave, it allows a direct examination of the wave energy content within the shell region, which is a critical area in the cross-section that dictates the overall strength of a wooden pole. By modifying the physical and the poroelastic properties of the medium to simulate the incipient decay and high moisture content, this study discovers a correlation between the selected features within the received waveform and the physical property of the medium (e.g., modulus of elasticity and the moisture saturation levels). The findings from both the numerical and analytical approaches motivate the proposed physics-based signal analysis to extract both the temporal and spectral information at the resonant frequency of the ultrasonic wave via the time and frequency (TF) transformation. A comparative study using the numerical results was performed to examine the Short-Time Fourier Transformation (STFT) and Gabor Continuous wavelet transform (GCWT). Due to its superior temporal and spectral resolutions, the GCWT is selected to analyze signals from different simulated conditions. The results produce a pronounced difference in the selected features in all the different simulated cases, suggesting a viable analysis approach for characterizing the medium. Based on the proposed physics-based signal analysis approach, this work develops and details a corresponding pole analysis algorithm. The experiments were carried out with specimens of different known Groundline (GL) conditions (healthy, decay and high-moisture) to examine the efficacy of the proposed waveguide design and the associated analysis algorithm. The collected signals are fed through the GCWT analysis algorithm to extracted the features sensitive to those conditions. The results suggest a high moisture content pole would have a typical energy attenuation of around 35% compared to a healthy pole with low moisture content, while the time when the peak energy occurs is relatively the same in both cases. In a decay wood specimen, the result suggests an 11% latency in peak energy time-of-flight (TOF), 50% energy reductions, and a 50% increase in diffusivity measuring the full width half maximum (FWHM) around the energy peak. A further refinement of the analysis places the peak energy TOF and energy reduction levels as the selected features on a feature space to assess fifteen poles with known categories. A clear decision boundary is discovered prompting a future research opportunity of using linear and/or non-linear classifiers to determine the wooden pole at the GL region.