COMPACTION OPTIMIZATION OF SILTY SANDS USING FACTORIAL DESIGN AND REGRESSION MODELING IN PATHUM THANI

Abstract

This study establishes an optimization-based framework for improving the compaction performance of natural and modified silty sands widely used in earthwork applications in Pathum Thani, Thailand. A 3 × 3 × 3 full-factorial design comprising 27 Modified Proctor tests was conducted on natural silty sand (SM-N), cement-modified silty sand (SM-C), and polymer-modified silty sand (SM-P). The results reveal clear soil-dependent compaction behavior, with maximum dry density (MDD) values of 1.81 g/cm³ for SM-N, 1.65 g/cm³ for SM-C, and 1.788 g/cm³ for SM-P under their respective optimal conditions. The highest densities were consistently achieved at 1.5× compaction energy, while optimum moisture levels varied by soil type OMC for SM-N and SM-P, and OMC + 2% for SM-C. Increasing compaction effort from 0.5× to 1.5× yielded up to a 12.4% improvement in MDD, whereas moisture deviations of ±2% from OMC resulted in 4–9% reductions in density depending on modification type. ANOVA confirmed that all main factors and interaction terms were statistically significant (p < 0.05), emphasizing the multi-variable nature of silty sand compaction. A second-order regression model demonstrated strong predictive accuracy (R² = 0.968), enabling reliable estimation of density across varying field conditions. The integration of factorial analysis and predictive modeling produced a soil-specific compaction optimization matrix that highlights the superior densification potential of polymer-modified silty sand and the moisture-sensitive behavior of cement-treated material. Overall, the proposed framework provides a systematic, data-driven alternative to conventional trial-based compaction practices and enhances earthwork reliability in moisture-sensitive silty sand deposits.