Soil-rock mixtures with large particle size variations are often used as fill materials for expressway construction in mountainous areas. Conventional testing methods do not enable fast and nondestructive monitoring of real-time changes in the compaction quality of soil-rock filled subgrades. Selecting an appropriate evaluation method is the key to controlling the compaction quality of a soil-rock filled subgrade. In this study, three-dimensional DEM models of subgrade materials were reconstructed by a spherical harmonic series whose harmonization degree was fixed at 15. The macroscopic and mesoscopic behaviours and characteristics of the subgrade under vibratory rolling were analysed. The results showed that the porosity, contact force and coordination number of the subgrades tended to be stable in the last two passes. The subgrades with 4 filler combinations presented the similar mechanical anisotropy and meso-mechanical states. On-site monitoring of subgrades under vibratory rolling and settlement after construction was performed, and the results were considered. An evaluation method and criterion to control the compaction quality of the SRM subgrade was proposed, i.e., whether the average value of the vibration compaction value from the second to last pass differed by more than 2% from the average value in the last pass.

The ultrasonic pulse velocity (UPV) correlates significantly with the density and pore size of subgrade filling materials. This research conducts numerous Proctor and UPV tests to examine how moisture and rock content affect compaction quality. The study measures the changes in UPV across dry density and compaction characteristics. The compacted specimens exhibit distinct microstructures and mechanical properties along the dry and wet sides of the compaction curve, primarily influenced by internal water molecules. The maximum dry density exhibits a positive correlation with the rock content, while the optimal moisture content demonstrates an inverse relationship. As the rock content increases, the relative error of UPV measurement rises. The UPV follows a hump-shaped pattern with the initial moisture content. Three intelligent models are established to forecast dry density. The measure of UPV and PSO-BP-NN model quickly assesses compaction quality. (c) 2024 Institute of Rock and Soil Mechanics, Chinese Academy of Sciences. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/ 4.0/).