Objective and Method To achieve rapid coalbed methane (CBM) extraction from broken-soft coal seams using surface horizontal wells, this study proposed a volume fracturing technology for coal seam roof using horizontal wells. Focusing coal seam 4-1 in a mining area, this study established a numerical model using finite element analysis (FEA) software Abaqus Using this model, the impacts of in-situ stress on the cross-layer propagation characteristics of multiple fractures were investigated. Through true triaxial hydraulic fracturing experiments, the multi-fracture propagation patterns under varying injection rates of fracturing fluids, perforation cluster numbers, and cluster spacings were analyzed. Using 3D fracturing design software Mshale, key parameters for volume fracturing, including perforation cluster number, cluster spacing, injection rate of fracturing fluids, average proppant ratio, and fracturing fluid volume, were optimized. Finally, the proposed technology with optimal parameters was applied to practical engineering.

Results and Conclusions The in-situ stress state determined the morphology of fracture networks induced by volume fracturing. Specifically, when the vertical stress difference coefficient (k) ≤ 0, multiple fractures were prone to propagate along the direction of the minimum horizontal principal stress, ultimately forming horizontal fractures. In contrast, in the case of k ≥ 0.25, multiple fractures propagated rapidly across layers to form vertical fractures, with the cross-layer velocity increasing with k. Increasing interlayer stress difference increased the fracture length within the coal seam but decreased the fracture length within the coal seam roof. Meanwhile, the length difference between the middle fracture and fractures besides, as well as the average pressure for initial cracking, increased with the interlayer stress difference. Fractures in specimens preferentially propagated within the coal seam roof after initial cracking, with propagation patterns varying with the injection rate of fracturing fluids. A low injection rate caused insufficient fracture heights in the coal seam; a medium injection rate promoted fracture propagation in the coal seam, leading to the formation of interconnected fracture networks; a high injection rate, despite enabling fractures to extend to the specimen boundaries, increased the initial cracking. Regarding the impacts of perforation cluster number, two clusters generated I-shaped fractures, leading to uneven reservoir stimulation, while three clusters enabled fractures to interconnect within the coal seam, resulting in the formation of V-shaped fracture networks. In contrast, four clusters induced intense inter-fracture interference, leading to the formation of “凵”-shaped fracture networks within the roof. In this case, horizontal fractures prevented vertical fractures from propagating across different layers. Under excessively large cluster spacing, perforation clusters near the wellhead induced initial cracking first, leading to the formation of dominant channels. In contrast, it proved difficult for the remaining clusters to cause initial cracking. Therefore, excessively large cluster spacing is unfavorable for the formation of fracture networks. Furthermore, an excessive number of perforation clusters or too small cluster spacing can intensify inter-fracture interference, thereby inhibiting fracture propagation. Increasing the injection rate of fracturing fluids can enhance the net pressure within fractures, thus promoting the synchronous development of fracture length, width, and height. In contrast, an increase in fracturing fluid volume preferentially improves fracture length. Controlling the average proppant ratio within a range of 16%‒24% can avoid near-wellbore sand plugging while also ensuring effective far-field proppant placement. Accordingly, the optimal fracturing parameters were determined, including a fracturing section length of 60 m, a perforation cluster number of 3, a cluster spacing of 20 m, an injection rate of fracturing fluids of 20 m³/min, an average proppant ratio of 16% (maximum proppant ratio ≤ 24%), and a fracturing fluid volume of 3 500 m³. The proposed technology performed well in the field application. The field microseismic monitoring results indicate that the resulting fracture networks exhibited fracture half-lengths ranging from 212.5 m to 225.5 m (average: 219.0 m) and fracture heights from 25 m to 40 m (average: 32.5 m). The fracture networks penetrated the coal seam and extended into the coal seam floor, with the fracture propagation characteristics consistent with the theoretical research results. Within the same mining area, the expansion scale of the fracture networks created using this technology was significantly larger than that produced by the conventional multistage fracturing of coal seam floor using horizontal wells. The results of this study will provide technical support for efficient hydraulic-fracturing stimulation of CBM reservoirs in broken soft coal seams.