This dissertation develops a coherent suite of numerical modeling components for pavement engineering that spans particle, mesoscale, and equipment–soil representations, with the shared goal of improving model fidelity and interpretability in computational studies relevant to construction processes. The work advances (i) adhesive contact modeling for discrete element analyses, (ii) virtual aggregate generation with independently controlled morphology for mesoscale mixture models, and (iii) dynamic modeling of the compactor–soil system with explicit treatment of mechanical inertia and delayed feedback control. Each development is verified against targeted benchmarks before being exercised in application-style simulations, ensuring that the resulting insights rest on demonstrably stable and transparent numerical formulations. At the particle scale, a Johnson–Kendall–Roberts (JKR)–based contact formulation is introduced in which the surface energy parameter is prescribed as a function of time, enabling controlled within-run evolution while preserving the analytical structure of the classical contact model. Canonical tests (ball–ball pull-off and gravity loading) confirm force–displacement behavior and energy conservation, after which the contact model is applied to pre-compaction and rotating drum scenarios. Relative to a constant parameter baseline, compaction impulses are increased by 15.57%/14.54% (tamper–particle, SMA-11/AC-11) and 13.04%/14.87% (screed–particle, SMA-11/AC-11), demonstrating the quantitative sensitivity of predicted effort to adhesion evolution. In rotating drum sequences, the angle of repose decreases from 48.4° to 37.7° over 5 s under a prescribed parameter reduction path, reflecting enhanced flowability captured by the time-parameterized law. At the mesoscale, a multi-scale algorithm generates virtual aggregates with independent controls for shape (coarse scale), angularity (medium scale), and surface texture (fine scale). Validation against a 3D-scanned database demonstrates high statistical fidelity, with Bhattacharyya coefficients of 0.9710 (true sphericity), 0.9432 (angularity), and 0.9499 (arithmetic-mean roughness). The generated skeletons are then assembled into asphalt concrete mixture models for dynamic simulations, providing evidence that the morphologically realistic, parameter-controllable mesostructures are mechanically reliable for use in mixture-level computations. At the equipment–soil level, a three-degree-of-freedom compactor–soil coupling model is established that explicitly includes the mechanical inertia of the suspension. Accounting for inertia markedly prolongs predicted transients (steady-state times increase from 8.03 s/4.57 s for frame/drum to 55.78 s/27.06 s), while a displacement-based delayed feedback (DF) active suspension reduces these to 14.31 s/7.67 s under representative gains. These results clarify the distinct numerical roles of inertia (response prolongation) and delay-based control (response contraction) within the same modeling framework. Taken together, the thesis consolidates a robust computational foundation for pavement engineering analysis: particle-scale contact is modeled with greater fidelity within the discrete element method; mesoscale aggregate morphology is generated under explicit, validated control; and equipment–soil interaction is expressed with the dynamic features required for construction-stage simulations. Verification through targeted benchmarks and application-style studies establishes the reliability of the overall framework for realistic use.
Dong Feng (Thu,) studied this question.