MSc Defence - Zach Evans
Date and Time
Location
MACN 415
Details
Multivariate Machine Learning Approaches to Spectroscopic Analysis of Cross-linked Polyethylene Water Pipes
MSc Candidate: Zach Evans
Abstract
Crosslinked polyethylene (PEX) potable water pipes are one of the most popular products on the planet for domestic and industrial potable water transportation. PEX being cost-effective, easily installed, and having long lifetimes in-service have contributed to the distribution and growth of this industry. Polyolefins like PEX are susceptible to oxidative degradation, and in rare cases PEX pipes fail prematurely causing undue costs for both consumer and industry. Understanding how these cracks initiate and propagate into a failure could provide to vital information leading to improved outcomes for the lifetime of PEX pipes. We use Fourier Transform Infrared Microscopy in conjunction with multivariate machine learning techniques such as Principal Component Analysis to understand initial manufacturing conditions effect on the accumulation of degradative products, and β-Variational Auto Encoders to disentangle spectral changes relevant to crack formation and propagation.
Examination Committee
- Dr. Elisabeth Nicol, Chair
- Dr. John Dutcher, Advisor
- Dr. Leonid Brown, Advisory Committee
- Dr. Aicheng Chen, Advisory Committee