Evaluating and enhancing the geotechnical properties of clay–wood ash and clay–sawdust composites using RSM and ANN

This study examines the mechanical and geotechnical performance of expansive clayey soil stabilized with sustainable agro-industrial by-products. The experimental program evaluates the effects of incorporating wood sawdust (WS) and wood ash (WA) at dosages of 2% to 10%, both individually and in combination with a fixed 2% addition of Ordinary Portland Cement (OPC). Comprehensive laboratory characterization included Atterberg limits, methylene blue value, hydrometer analysis, compaction characteristics determined by the Standard Proctor test. Direct shear testing was conducted to determine cohesion (c) and the internal friction angle Φ), while California bearing ratio (CBR) and unconfined compressive strength (UCS) were evaluated as key mechanical properties. Soil responses were modeled using response surface methodology (RSM) and artificial neural networks (ANNs). The mixture containing 8% WA and 2% OPC showed the greatest mechanical improvement relative to the untreated natural clay baseline, increasing CBR by 389.34% and UCS by 335.50%. while significantly enhancing shear strength parameters. Both ANN/Genetic Algorithm (GA) and RSM/Desirability function effectively identified mixes that maximized CBR and UCS, although ANN/GA provided superior prediction accuracy and closer alignment with the model-based optimum. These findings support the use of optimized clay-based biocomposites incorporating WA, WS, and OPC for sustainable foundation engineering, soil stabilization, and pavement subgrade enhancement.