The fracturing design of horizontal wells in low-permeability reservoirs mostly relies on static data such as logging data, and insufficient utilization of while-drilling information from drilling and mud logging leads to an unreliable basis for staged fracturing and cluster design, making it difficult to achieve efficient reservoir stimulation. To address this issue, this paper takes mechanical specific energy (MSE) as the core engineering parameter. Based on the drilling and mud logging data from the horizontal section of an example well, a mechanical specific energy model is established after data preprocessing. The K-medoids unsupervised clustering algorithm is adopted to realize the fine division of fracturing stages, and the optimal number of stages is determined using the elbow method. A fracturing sweet spot index (FSI) is constructed using the random forest algorithm. This index integrates four key parameters: oil saturation, porosity, permeability, and mechanical specific energy to optimize perforation cluster placement. The proposed method is validated through finite element numerical simulation and field data comparison. The research results show that the optimal division of the example well is determined as 11 stages with six clusters per stage, and the matching degree between stage-cluster division and reservoir heterogeneity is significantly improved. Compared with the original scheme, the stimulated reservoir volume (SRV) of the optimized construction scheme is increased by 19.6%, and the treatment pressure exhibits an obvious overall reduction trend, as demonstrated by the construction pressure curves. The research indicates that this method can make full use of drilling while-drilling data to realize integrated geological and engineering evaluations of fracturing and optimization of staged fracturing and cluster designs, and can provide effective technical support for the fracturing optimization design of horizontal wells in low-permeability reservoirs lacking complete logging data.

Conventional fracturing design generally only considers the propagation of a single fracture in homogeneous storage, but the heterogeneity characteristics of the reservoir and the development of complex natural fractures will affect the fracturing effect. In order to study the horizontal well fracture propagation law under the synergistic effect of complex natural fracture reservoir heterogeneity, the globally embedded cohesive element finite model is established, and the heterogeneous reservoirs are built by Python programming. The influencing factors of multi-fracture propagation with heterogeneous reservoirs are analyzed, including the in situ stress difference, the injection rate of fracturing fluid, the perforation spacing, and the perforating clusters number. The results show that the heterogeneity and the natural fracture of the reservoir will affect the in situ stress distribution and propagation direction. When the in situ stress difference is low, the reservoir heterogeneity and natural fractures have a great influence on the fracture morphology. With the increase in injection rate, the fracture length and fracture width increase. With the in situ stress difference increase, the fracture deflection angle decreases. With the increase in fracturing fluid injection rate, the fracture length increases. The number of hydraulic fractures communicating with natural fractures increases with the natural fracture number and the number of perforation clusters under the synergistic effect of complex natural fracture and reservoir heterogeneity.