Issue 77

Y. C. Arun et alii, Fracture and Structural Integrity, 77 (2026) 316-339; DOI: 10.3221/IGF-ESIS.77.19

Conversely, SiC particle size exhibits a positive coefficient (+0.000042), implying a slight increase in CoF due to enhanced asperity interaction. The intercept (0.3635) represents the baseline CoF. Overall, load and SV dominate friction reduction, while other factors show comparatively minor effects. The coefficient of friction linear regression model has an R 2 of 0.7734, which means that the model explains 77.34% of the total variability. After taking the number of predictors into consideration, the modified R 2 of 69.25% shows a modest level of predictive reliability. However, the relatively small, adjusted value and the difference between R² and adjusted R² indicate that the linear model may not adequately represent the intricacy of the CoF response. This suggests that there are important interaction effects and non-linear interactions between the process parameters, which the current linear formulation does not sufficiently capture. The residuals are roughly normally distributed, as seen in Fig. 12, where most of the dots follow the reference line. The random dispersion about zero in the residuals vs. fitted figure suggests constant variance and no discernible model bias. The absence of a systematic trend in the residuals vs. observation order confirms the independence of mistakes. The model's underlying assumptions are largely met.

Figure 12: Residual plots for coefficient of friction model.

Confirmation test for wear loss and coefficient of friction For both wear loss and coefficient of friction (CoF) of GF/PPS/CNF hybrid nanocomposites, the confirmation test findings shown in Tab. 10 show a strong agreement between the experimental values and the regression model predictions. Systematic adjustment of SiC particle size and filler content produced consistent trends under constant testing settings (8 N load, 0.45 m/s sliding velocity, and 40 m abrading distance), which the created models accurately reflected. Whereas CoF prediction errors fell between 2.86% and 6.50%, wear loss percentage errors varied between 3.73% and 5.09%. Interestingly, Trial 3 showed the best prediction accuracy with the lowest wear loss deviation (3.73%). The robustness and dependability of the regression models for forecasting tribological performance are generally confirmed by the low error margins (all below ~6.5%).

Experiment CoF

Regression model CoF

Experiment wear loss (g)

Regression model Wear loss (g)

Trails

% Error

% Error

0.3308 0.3226 0.3141

1 2 3

0.0149 0.0234 0.0361

5.09 4.93 3.73

0.3538 0.3321 0.3273

0.0157 0.0223 0.0375

6.50 2.86 4.03

Table 10: Confirmation test results for wear loss and CoF.

330