On the characterization of the viscoelastic response of a class of materials using acousto-ultrasonics: A pattern recognition approach.

Abstract

This thesis describes the experimental work carried out at the University of Ottawa for characterizing linear viscoelastic materials using acousto-ultrasonic wave propagation technique. The objective of the research program is to asses the feasibility of utilizing acousto-ultrasonics in the characterization of mechanical response and strength prediction of solid polymers as a class of engineering materials. The basic principles of acousto-ultrasonic technique, the instrumentation involved, and the factors affecting waveform measurements are described. A review on the analysis of the resultant acousto-ultrasonic waveform using statistical pattern recognition methodology is presented. The time-dependent properties of creep, and stress-relaxation of linear viscoelastic materials are reviewed. Also, fatigue, macromechanical defects and yielding of polymers are discussed. Experimental results obtained in a series of acousto-ultrasonic characterization tests performed on Polyvinylchloride (PVC) and Polycarbonate (PC) specimens are presented. Good correlation was obtained between the acousto-ultrasonic parameter (AUP) and stress in material during a time-dependent stress relaxation testing for a particular strain level. The acousto-ultrasonic parameter (AUP) was seen to decrease with increasing stress levels during uniaxial tensile loading. Also, induced fatigue states in the material and macromechanical defect states were seen to have good correlation with the AUP. Statistical pattern recognition classifiers were designed for all characterization experiments performed and tested with unknown samples and found to be very effective. Acousto-Ultrasonics (AU), combined with statistical pattern recognition methodology applied in real time environment, proves to be an efficient non-destructive quantitative evaluation process for the class of solid polymers investigated.