Machine learning approach for tribological improvement of pond ash-glass fiber/epoxy composites used in sliding-wear parts
Priyadarshi Tapas Ranjan Swain, R.S. Krishna, Mohammed Nasir, Srimant Kumar Mishra, Prabina Kumar Patnaik, Abhilash PurohitPurpose
The purpose of this research is to enhance the dry sliding wear performance of glass fibre-reinforced epoxy composites by incorporating pond ash as a sustainable filler. The study aims to evaluate how different pond ash contents influence wear rate and friction, optimise the tribological parameters through the Taguchi method, and develop an artificial neural network (ANN) model to accurately predict wear behaviour. The work ultimately seeks to demonstrate that industrial waste can be effectively converted into value-added composite materials suitable for light-duty bearing and structural applications.
Design/methodology/approach
Composites containing 0–15 Wt.% pond ash were fabricated by hand lay-up and compression curing using glass fibre-reinforced epoxy as the matrix system. Dry sliding wear tests were carried out on a pin-on-disc tribometer by varying pond-ash content and operating parameters. A Taguchi L16 orthogonal array was used to identify significant factors and determine the optimal combination for wear reduction. An ANN model based on back-propagation was developed to predict wear behaviour using the experimental data set.
Findings
Sliding wear analysis shows that the rate of specific wear reduces with the addition of pond ash content. Taguchi’s design of experiment approach confirms that pond ash content is the maximum dominating factor, followed by velocity of sliding, applied load, and distance of slide. An ANN architecture model was successfully implemented for predicting wear behaviour outside of experimental conditions. The predicted values showed less than 10% error and confirmed the accuracy and reliability of the trained ANN model for composite performance prediction.
Originality/value
This study offers a clear blend of sustainable materials engineering and data-driven optimisation. It introduces pond ash as a hybrid filler in glass fibre-reinforced epoxy composites and demonstrates how this waste material can meaningfully improve wear resistance. The work also combines Taguchi optimisation with an ANN model to validate and predict tribological behaviour with high accuracy. The integration of experimental analysis, and machine learning provides a fresh perspective on developing greener, high-performance composites for sliding applications.
Peer review
The peer review history for this article is available at: https://publons.com/publon/10.1108/ILT-11-2025-0534/