Abstract:
This study investigated the characterization and predictive modeling of thermally aged
Glass Fiber Reinforced Plastic (GFRP) Composites. The experimental part of the study
explored the effect of fiber orientation, laser cutting and thermal aging on GFRP
mechanical properties. The development of a predictive model for estimating the
mechanical properties of thermally aged GFRP was explored in the computational part.
GFRP composites were fabricated with woven and random glass fiber and epoxy resin
hardener and subjected to mechanical and laser machining. Mechanical property testing
reveals that Tensile and flexural properties are found to be superior in mechanically cut
samples. Compromised surface integrity due to thermal damage in the case of laser cut
samples is also noted. All results indicated that woven GFRP has superior mechanical
properties than random GFRP. Woven GFRP tensile test samples were thermally aged at
50°C, 100°C, 150°C and 200°C for 30 mins, 60 mins, 90 mins and 120 mins. The samples
showed a gradually increasing brown color at temperatures above 150°C. The tensile test
showed that the Ultimate Tensile Strength (UTS) value had a general decreasing trend as
the thermal aging temperature increased. The predictive model read the photographic image
of a thermally aged sample and used the color change due to thermal aging as an identifier
for the image processing algorithm. Artificial Neural Networks (ANN) estimated the
thermal aging temperature and time from the image processing algorithm’s Red Green Blue
(RGB) color matrix output. A regression equation was also developed which creates a
mathematical relationship between the UTS values and the thermal aging variables from
the experimental data. Finally, the ANN’s output was forwarded to the developed
regression equation to get the estimated UTS. The predictive model’s estimated UTS
showed an average accuracy of 97% compared to the experimental results. The results of
the characterization of mechanical properties of thermally aged GFRP can contribute
meaningful insights into the existing literature. The developed predictive model can have
potential applications in aerospace line maintenance operations with the promise of cost
and time savings.
