Rocznik Ochrona Środowiska 2026, vol. 28, pp. 329-348
Abdullah Faiz Al Asmari 
 |
King Khalid University, Saudi Arabia |
 |
|
 |
https://doi.org/10.54740/ros.2026.023 |
   |
|
 |
The increasing need to use sustainable pavement materials has prompted the pursuit of alternatives to Petroleum bitumen made of bio-based materials. This work explores the environmental and mechanical behavior of bio-oil-modified bitumen within the framework of a hybrid experimental and machine-learning approach. Bio-binder was waste cooking oil used to partially replace conventional bitumen (ranging from 0 to 20 percent). The experimental analysis aimed to assess the binder's high-temperature deformation resistance using the rutting parameter (G*/sinδ). Three machine learning models were created with the use of key input variables such as bio-oil content, mixing temperature, aging condition, viscosity, and density to improve the predictive power and decrease the amount of effort expended on the experiment, including Artificial Neural Network (ANN), Random Forest (RF), and Extreme Gradient Boosting (XGBoost). Statistical measures such as R2, RMSE, MAE, and MAPE were used to assess model performance. XGBoost proved to be the most accurate, with higher accuracy than ANN and Random Forest. The environmental benefits have been estimated by quantifying the binder-related CO2 reduction when using bio-oil instead of oil. A Pareto optimization model was used to determine the best trade-off between environmental sustainability and mechanical performance. The findings suggest that a replacement level of bio-oil of 10–15% offers the most promising compromise, as it retains acceptable rutting resistance whilst achieving a considerable carbon reduction. In general, the suggested hybrid experimental-machine learning method can be considered an effective instrument for optimizing sustainable asphalt binder formulations and offers the possibility of the continuous incorporation of environmentally friendly materials into pavement engineering. The novelty of this study lies in integrating experimental characterization, machine learning prediction, and Pareto-based environmental optimization within a single framework for bio-oil-modified bitumen. Unlike previous studies that focused separately on mechanical performance and predictive modeling, the present work simultaneously evaluates rutting resistance, CO₂ reduction potential, and predictive accuracy using ANN, Random Forest, and XGBoost models to develop a sustainable and optimized asphalt binder formulation.
bitumen, XGBoost, machine learning, CO₂ emission
 |
 |
AMA Style
Asmari A.. Hybrid Experimental–Machine Learning Framework for Environmental Optimization of Bio-Modified Bitumen. Rocznik Ochrona Środowiska. 2026; 28. https://doi.org/10.54740/ros.2026.023
ACM Style
Asmari A.. 2026. Hybrid Experimental–Machine Learning Framework for Environmental Optimization of Bio-Modified Bitumen. Rocznik Ochrona Środowiska. 28. DOI:https://doi.org/10.54740/ros.2026.023
ACS Style
Asmari A., Hybrid Experimental–Machine Learning Framework for Environmental Optimization of Bio-Modified Bitumen Rocznik Ochrona Środowiska 2026, 28, 329-348. https://doi.org/10.54740/ros.2026.023
APA Style
Asmari A. (2026). Hybrid Experimental–Machine Learning Framework for Environmental Optimization of Bio-Modified Bitumen. Rocznik Ochrona Środowiska, 28, 329-348. https://doi.org/10.54740/ros.2026.023
ABNT Style
ASMARI A.. Hybrid Experimental–Machine Learning Framework for Environmental Optimization of Bio-Modified Bitumen. Rocznik Ochrona Środowiska, v. 28, p. 329-348, 2026. https://doi.org/10.54740/ros.2026.023
Chicago Style
Abdullah Faiz Al Asmari. 2026. "Hybrid Experimental–Machine Learning Framework for Environmental Optimization of Bio-Modified Bitumen". Rocznik Ochrona Środowiska 28, 329-348. https://doi.org/10.54740/ros.2026.023
Harvard Style
Asmari A. (2026) "Hybrid Experimental–Machine Learning Framework for Environmental Optimization of Bio-Modified Bitumen", Rocznik Ochrona Środowiska, 28, pp. 329-348. doi:https://doi.org/10.54740/ros.2026.023
IEEE Style
Asmari A., "Hybrid Experimental–Machine Learning Framework for Environmental Optimization of Bio-Modified Bitumen", RoczOchrSrod, vol 28, pp. 329-348. https://doi.org/10.54740/ros.2026.023