Old cement pavement directly overlaid with an asphalt layer produces many reflection cracks. Using microcrack homogenization technology to treat old cement pavement can effectively reduce the occurrence of reflection cracks. Micro-crack homogenization is relatively mature at the technical level; however, there is relatively little research on its mechanism. To evaluate the microcrack effect of old cement pavement, we conducted core sampling on a project road section after the appearance of microcracks. The core samples were sliced by X-ray tomography (X-ray CT). The mesostructure of the core samples was obtained. The core sample was further divided into microscopic cracks and macroscopic cracks according to the length and width of the crack. The development characteristics of cracks were subdivided into type I-III microcracks and type I-IV cracks. The core-drilling sample was divided into 5652 CT images. The statistical results showed that there were 3582 type-I microcracks, 3197 type-II microcracks, and 1835 type-III microcracks. Among the specimens, the minor proportion of microcracks was 32.87%, the most significant proportion was 100.00%, and the average ratio was 47.51%. Furthermore, the cracks development law and formation mechanism were analyzed based on CT images and cracking statistics. The results showed: (1) The microcrack homogenization process produced many microcracks in the test section and achieved a specific microcrack effect. (2) The cracks produced by the microcrack homogenization process tended to develop along the transition zone of the aggregate–cement mortar interface. The development of microcracks was mainly related to the aggregates’ shape and gradation, as well as the energy required to generate cracks. The research conclusions of this paper can be used as a theoretical basis for the optimization and improvement of the microcrack homogenization process.

In northwestern China, loess is abundant whereas suitable aggregates for rural pavement bases are scarce, yet the transfer from laboratory stiffness of stabilized loess to field support and cement concrete slab response remains unclear. This study aims to establish an E_i–E_t–pavement-response framework for cement–curing agent composite-stabilized loess and loess–sand mixtures. Compaction, unconfined compressive strength and uniaxial compression modulus tests were conducted; E_t was calculated using a specification-based method, checked by falling weight deflectometer (FWD) back-calculation, and introduced into a three-dimensional finite element model. Composite stabilization markedly improved base stiffness because cementation, curing agent-assisted bonding, gradation optimization and skeleton–filling effects produced a denser load-bearing structure. After 90 days, the compressive modulus increased from 225 MPa for 6% cement-treated loess to 570 MPa for the 45% loess–sand mixture. The calculated E_t agreed well with FWD back-calculated field values (R² = 0.91). Increasing E_t from 98.8 to 155.7 MPa reduced slab-bottom stress from 1.7449 to 1.1095 MPa and vertical displacement from 1.7102 to 0.2654 mm. Field engineers can use E_t to select local loess-based base materials and coordinate base and slab thickness, provided that compaction quality, water stability and long-term durability are verified.