Hein Htet Aung | MS Student | Case Western Reserve University
When it comes to predicting service lifetime of materials, it is important to account for different factors that impact the performance of a material, such as various material grades, stressors, and stressor levels. In addition, it is also crucial to explore the synergetic effects of different stressors that could impact the degradation rates and mechanism of a material. Data-driven models are useful when it comes to lifetime prediction since they are not restricted to have a complete understanding of the degradation mechanisms and they can also account for different factors that impact the performance of a material.
Network structural equation modeling (netSEM) is a data-driven modeling technique that has been developed at the SDLE Research Center and with a public version available as a R software package to the public on CRAN. Adapted from structural equation modeling (SEM) which is used to explore linear relationship between observed and latent variables, netSEM also captures the non-linear relationships between variables and explores the degradation pathway in a <Stressor|Mechanism|Response> (<S|M|R>) framework, providing a predictive <Stressor|Response> model (data-driven model) as well as an inferential and mechanistic <Stressor|Mechanism|Response> model that allows for rank ordering of degradation mechanisms. netSEM selects the best relationship between variables based on statistical significance such as adjusted R2 and also rank order models using Akaike Information Criterion (AIC), and Bayesian Information Criterion (BIC).
“netSEM provides both an inferential model as well as a predictive model which not only will allow us to explore how variables in the study relate to each other but also give us a way to predict the service lifetime or performance of a sample in relation to the stressor. ”
-Hein Htet Aun
The goal of Dr. Laura S. Bruckman and Hein Htet Aung (MS student) is to utilize netSEM to predict service lifetime of polymers. Using poly(methyl methacrylate) (PMMA) as a sample polymer, the team implements a lifetime and degradation science approach to predict the service lifetime. Data is collected and analyzed from real world outdoor exposure and indoor accelerated stepwise exposures, which captures the synergetic impact of degradation factors in a shorter amount of time. netSEM modeling would then be utilized to give insights on degradation pathways and predict service lifetime.
The team is utilizing netSEM to generate degradation pathway to give insights into degradation mechanisms and predict service lifetime.
In addition, the team is also working towards developing netSEM further in order to gain more insights for degradation modeling, such as comparing degradation pathways of different variants in study or effect of different stressor levels.