Data-Driven Approach to Predicting Durability and Property Retention in Polymer Matrix Composites
- 1KTH Royal Institute of Technology
Abstract
Predicting the long-term performance of reinforced polymer composites from short-duration accelerated aging data remains a central challenge in composite durability engineering. This work presents a data-driven kinetic and mechanical-performance characterization of a glass-flake reinforced unsaturated polyester composite exposed to deionized water at 20, 40, 60 and 80 °C for up to 14.5 months, integrating long-term gravimetric sorption, flexural testing, and dynamic mechanical analysis (DMA), including substantial portions of the dataset that have not previously been analyzed or reported. The initial moisture sorption rate constant obeys an Arrhenius relationship across all four temperatures (activation energy Ea = 45.1 kJ/mol, R2 = 0.97), providing a kinetic basis for estimating uptake behavior within the tested temperature range. Extended gravimetric monitoring reveals a two stage hygrothermal response: an initial Fickian-like uptake stage reaching an apparent plateau (4.4-7.5 wt% depending on temperature and replicate), followed by a degradation-associated mass-loss stage, with weight gain at 60 °C falling from a 4.5% plateau to 1.8% after 14.5 months, and markedly greater coupon-to-coupon variability at 80 °C. Flexural modulus and strength retention under wet conditions (22-50% and 28-60% of dry values, respectively) is reported across all four temperatures at 2 and 6 months, alongside DMA glass transition temperature data showing an initial rise with exposure temperature followed by a decrease at 80 °C. No meaningful relationship between initial moisture sorption kinetics and 2-month mechanical retention could be established from the present four temperature-matched averages (Pearson r = 0.43, n = 4; retention is non-monotonic in ln(k)); the 6-month data instead indicate that degradation kinetics and exposure duration, rather than the initial sorption rate, are what track mechanical performance loss. The study does not establish a validated model for predicting mechanical-property retention; rather, it provides a quantitative experimental basis and structured dataset for future data-driven and machine-learning-based durability models of this and structurally similar glass-flake-reinforced polyester and related polymer matrix composite systems.
Related
- Progress and perspectives on plant-based natural fiber-reinforced polymer composites: Structure–property relationships and emerging design strategies
- Functional Polymer Nanocomposite Coatings for Naval and Maritime Applications: Advances in Protection, Durability, and Surface Engineering
- Decision Framework for Polymer Composite Linings in Trenchless Renewal: Integrating Materials Durability, Operational Performance, Application Quality, and Environmental Sustainability
- Enhanced Interfacial Adhesion and Dynamic Seawater Durability of Basalt Fiber/Epoxy Composites via PDA-Assisted In Situ Growth of UiO-66-NH2
- Redox-Mediated Incorporation of Sulfated Cellulose Nanocrystals into PVAc Latex: A Comprehensive Structure–Property-Durability Investigations
BibTeX
@misc{kharazmi2026datadriven,
title = {{Data-Driven Approach to Predicting Durability and Property Retention in Polymer Matrix Composites}},
author = {Parastou Kharazmi},
howpublished = {ChemRxiv},
year = {2026},
month = oct,
doi = {10.26434/chemrxiv.15010219/v1},
url = {https://chemrxiv.org/doi/full/10.26434/chemrxiv.15010219/v1},
note = {CC BY},
}
RIS
TY - UNPB
TI - Data-Driven Approach to Predicting Durability and Property Retention in Polymer Matrix Composites
AU - Parastou Kharazmi
PY - 2026
DA - 2026/10/11
PB - ChemRxiv
DO - 10.26434/chemrxiv.15010219/v1
UR - https://chemrxiv.org/doi/full/10.26434/chemrxiv.15010219/v1
AB - Predicting the long-term performance of reinforced polymer composites from short-duration accelerated aging data remains a central challenge in composite durability engineering. This work presents a data-driven kinetic and mechanical-performance characterization of a glass-flake reinforced unsaturated polyester composite exposed to deionized water at 20, 40, 60 and 80 °C for up to 14.5 months, integrating long-term gravimetric sorption, flexural testing, and dynamic mechanical analysis (DMA), including substantial portions of the dataset that have not previously been analyzed or reported. The initial moisture sorption rate constant obeys an Arrhenius relationship across all four temperatures (activation energy Ea = 45.1 kJ/mol, R2 = 0.97), providing a kinetic basis for estimating uptake behavior within the tested temperature range. Extended gravimetric monitoring reveals a two stage hygrothermal response: an initial Fickian-like uptake stage reaching an apparent plateau (4.4-7.5 wt% depending on temperature and replicate), followed by a degradation-associated mass-loss stage, with weight gain at 60 °C falling from a 4.5% plateau to 1.8% after 14.5 months, and markedly greater coupon-to-coupon variability at 80 °C. Flexural modulus and strength retention under wet conditions (22-50% and 28-60% of dry values, respectively) is reported across all four temperatures at 2 and 6 months, alongside DMA glass transition temperature data showing an initial rise with exposure temperature followed by a decrease at 80 °C. No meaningful relationship between initial moisture sorption kinetics and 2-month mechanical retention could be established from the present four temperature-matched averages (Pearson r = 0.43, n = 4; retention is non-monotonic in ln(k)); the 6-month data instead indicate that degradation kinetics and exposure duration, rather than the initial sorption rate, are what track mechanical performance loss. The study does not establish a validated model for predicting mechanical-property retention; rather, it provides a quantitative experimental basis and structured dataset for future data-driven and machine-learning-based durability models of this and structurally similar glass-flake-reinforced polyester and related polymer matrix composite systems.
ER -