From waste to value: Oil palm frond biomass for data-driven modeling and optimization of Remazol brilliant blue R dye adsorption

Dye contaminants in industrial effluents pose serious environmental and human-health risks, creating a need for treatment methods that are both effective and sustainable. In this study, oil palm frond (OPF) biomass was evaluated as a low-cost and energy-efficient adsorbent for the removal of Remazol Brilliant Blue R dye (RBBR) from aqueous solution. A Box–Behnken design with 27 runs was used to examine the effects of initial pH, adsorbent dosage, contact time, and initial dye concentration on dye removal efficiency. Predictive models were developed using response surface methodology (RSM), artificial neural networks (ANN), and support vector machine regression (SVM), and their performances were compared using leave-one-out cross-validation (LOOCV). To account for the bounded nature of dye removal efficiency (0–100%), model training, validation, and hyperparameter tuning were conducted on the logit-transformed response, and predictions were subsequently back-transformed to the original scale. Under LOOCV, the SVM model achieved the best predictive performance, with the lowest root mean square error (RMSE = 6.76), lowest mean absolute error (MAE = 5.64), and highest coefficient of determination (R2 = 0.907), followed by ANN (RMSE = 7.74, MAE = 5.76, R2 = 0.880). The reduced RSM model showed slightly lower predictive accuracy (RMSE = 9.33, MAE = 6.88, R2 = 0.835). The highest experimentally observed dye removal was 94.20%. Overall, OPF is a promising sustainable adsorbent for RBBR removal, while SVM and ANN showed slightly better predictive performance and RSM remained valuable for interpretability and process insight.

Text 125990.pdf - Published Version Restricted to Repository staff only Download (6MB) Official URL or Download Paper: https://linkinghub.elsevier.com/retrieve/pii/S1383...

Actions (login required)

View Item