Objectives and Methods Deep coalbed methane (CBM) reservoirs commonly exhibit well-developed beddings, strong mechanical heterogeneity, and high in-situ stress gradients. These characteristics result in pronounced nonlinear fracture propagation and strong multi-field coupling effects during hydraulic fracturing. Consequently, it is challenging to accurately describe the mechanisms governing fracture complexity in deep coal reservoirs using conventional mechanical models for fractures. Using a super-large true triaxial system with dimensions of 2.0 m × 2.0 m × 1.0 m, this study conducted physical simulation experiments on hydraulic fracturing under varying injection rates and viscosities of fracturing fluids. In combination with fracture mechanics and energy conservation theory, this study established an energy balance equation for fracture propagation, a convection-diffusion equation for proppant transport and settling, and a model for the coupling relationships among fracture complexity and the injection rate and viscosity of fracturing fluids. Accordingly, both the dynamic mechanisms behind fracture evolution and the pattern governing the fracture network complexity were systematically elucidated.

Results  The results indicate that fracture propagation is jointly controlled by the in-situ stress field, fluid pressure field, and bedding structures, representing a unsteady energy conversion process. The fracture propagation rate exhibits a power-law relationship with the energy release rate. The injection rate of fracturing fluids primarily determines the energy input rate and fracture propagation velocity. A high injection rate results in energy concentration in the front of the primary fracture, promoting fracture interconnectivity while suppressing branch development. Accordingly, fracture complexity is reduced. In contrast, a low injection rate corresponds to a more uniform energy distribution, enhancing the accumulation and lateral diffusion of energy. This facilitates multi-point initial cracking and fracture branching, increasing fracture complexity by approximately 25%–35%. Fracturing fluid viscosity significantly influences the energy transfer between fluids and solids, as well as proppant settling behavior. A high viscosity (45 mPa·s) is associated with a significant decrease in the proppant settling velocity. Compared to a low viscosity of 15 mPa·s, the high viscosity increases the proppant transport capacity by approximately 40%, promoting more uniform proppant placement in far-wellbore zones and creating favorable conditions for the formation of continuous hydraulically conductive pathways.

Conclusions Empirical relationships derived from experiments and fitting indicate that the fracture complexity exhibits power-law coupling relationships with the injection rate and viscosity of fracturing fluids. Notably, the low-injection-rate and high-viscosity combination is more favorable for the development of 3D fracture networks, with a fractal dimension reaching up to 1.46. The proposed theoretical-experimental coupling framework reveals the energy transfer mechanisms governing fracture propagation and proppant transport in deep coal reservoirs, providing a quantitative theoretical basis for optimizing hydraulic fracturing parameters and predicting fracture complexity in deep unconventional reservoirs.