Shaped by long-term geological processes and human activities, rock masses contains a variety of discontinuous structures such as fractures, voids and weak interlayers of different scales [[1], [2], [3], [4], [5]]. Underground engineering activities disturb the initial state of stress equilibrium in geological structures, causing a redistribution of stress fields within rock masses. This, in turn, results in the initiation, propagation, and coalescence of cracks in the rock mass, ultimately leading to destabilizing failure [[6], [7], [8], [9], [10]]. The prevention and control of engineering disasters, as well as the scheme design for oil and gas exploitation, rely on accurate characterization of evolution processes of stress field and crack in rock masses [11]. However, these evolution processes hidden “black box” processes and challenging to directly elucidate using conventional methodologies. Therefore, how to directly and accurately describe the evolution of rock internal structure and stress field is the key problem to improve the safety and effectiveness of underground engineering.

The current theoretical analysis is constrained in accurately describing the stress field and crack evolution characteristics within rock interior due to the intricate nature of its structure [[12], [13], [14], [15]]. Traditional stress monitoring methods, including on-site monitoring [16] and laboratory tests utilizing stress-strain sensors [17], are limited in their capacity to provide comprehensive stress data due to the restricted number of available monitoring points. The rapid advancement of computer technology and numerical simulation algorithms has led to the widespread adoption of numerical simulation among scholars, enabling them to capture full-field stress characteristics and changes in rock structure [[18], [19], [20], [21]]. However, its accuracy is influenced by various factors such as the selection of material parameters, grid division, element definition, choice of constitutive model, boundary condition settings, necessitating experimental verification [22,23]. Photoelasticity, an optical method, provides a visual and quantitative way to characterize the full-field stress in models [[24], [25], [26], [27], [28]]. However, the traditional method of making physical models is time-consuming, labor-intensive, and inadequate for developing rock models with complex structures.

The emergence of advanced 3D printing technology offers novel possibilities for the fabrication of physical models of rock [[29], [30], [31], [32]]. Owing to its inherent advantages including high precision, efficiency, repeatability, batch production capability, and the ability to manufacture intricate structural models, 3D printing technology has found applications in the field of rock mechanics [29,33]. VeroClear, a printable photosensitive resin material, has been successfully utilized in reproducing rock models with intricate internal structures through the integration of CT scanning and 3D reconstruction techniques [34,35]. Notably, VeroClear exhibits excellent transparency, enabling direct visualization of internal structural changes. This characteristic proves beneficial for investigating deformation characteristics and crack evolution laws within the structure. Moreover, owing to VeroClear's remarkable birefringence effect, the photoelastic method can be employed to quantify and visualize full-field stress in printed models accurately [36,37]. The obtained results demonstrate good agreement with actual engineering data. However, VeroClear has a lower strength and higher ductility compared with natural rock. During the elastic stage, regardless of whether the mechanical properties of the printed model and rock model are identical, their stress distribution and deformation characteristics remain consistent as long as their geometries, boundary conditions, and applied loads are equivalent. However, beyond this stage, discrepancies in properties prevent accurate representation of the actual rock fracture process by the printed model. And it is of utmost importance to thoroughly comprehend the stress characteristics and failure modes of rock during the plastic deformation stage in order to precisely assess the safety of engineering projects and effectively exploit underground oil and gas resources. Hence, there is an imperative to develop materials that can precisely replicate the physical and mechanical properties of rocks while enabling visual representation of their physical processes as well as stress fields. Given this, we have endeavored to enhance the strength and stiffness of VeroClear through post-processing techniques and control over the printing orientation [38]. The properties of VeroClear models, however, still falls significantly below the level demonstrated by natural rock. Further, it has been discovered that reducing the ambient temperature can significantly improve the mechanical properties of VeroClear, and its mechanical parameters can reach the level of some natural rocks [39]. However, the fracture characteristics of the VeroClear model and the quantification effect of stress fields at low temperatures remain unknown, necessitating further investigation.

The present study investigates the influence of temperature treatment on the mechanical properties of VeroClear and analyzes the impact of temperature on the stress-optical properties of stress-sensitive materials. Utilizing 3D printed models with pre-existing cracks, this research obtains crack initiation conditions, propagation laws, and stress field distribution characteristics for stress-sensitive materials at different temperatures.