Neural network modeling of rutting performance for sustainable asphalt mixtures modified by industrial waste alumina

Abstract

Considerable efforts have been devoted to developing a sustainable approach to road construction. The use of recycled concrete aggregate (RCA) obtained from construction-demolished concrete as a substitute for natural aggregate (NA) has increased interest. However, the inferior performance of the resulting RCA mixtures requires adding other improvement materials. To build on the few existing works in this area, the present study investigated the role of waste alumina (Al) as a reinforcement agent when added at rates of 1 %, 2 %, and 3 % by mixture weight. The experimental program consisted of volumetric properties analysis accompanying the Marshall test, dynamic creep test, and wheel tracking test. This study revealed that RCA-mixtures reached their typical Marshall and volumetric properties at 4.5 % binder content recording an increasing increment of 10 % compared to NA-mixtures which needed only 4.1 % binder content. Overall, all the utilized Al dosages improved the rutting resistance; optimally, 2 % of Al decreased the rutting depth by 26 %. This dosage also increased the index of plastic deformation, resilient modulus, and flow number values by 30 %, 16 %, and 23 %, respectively. The artificial neural network statistical technique produced a robust model (R2 = 0.995) correlating the performed tests’ outcomes with rutting values.

Keywords

• Rutting
• ANN
• Waste alumina
• Resilient modulus
• Flow number
• Wheel tracking test