Stage-Dependent Responses in Sequential Dual-Well Supercritical CO2 Fracturing: A Sparse Second-Order Choquet Fracability Framework with Monte Carlo Uncertainty
Abstract
Deep tight-reservoir fracability depends on temperature, fluid transport, in situ stress, fracture development, and injection energy, so no single response can provide a defensible ranking of fracturing conditions. This study develops a sparse second-order Choquet–Monte Carlo fracability framework. Additionally, the CRITIC method is used independently to quantify the information contrast of the five evaluation indicators. This was done using true triaxial sequential dual-well experiments conducted at 25–200 °C, horizontal stress differences of 3–9 MPa, and with either supercritical CO2 (SC-CO2) or water. CT scanning, U-Net segmentation, and three-dimensional reconstruction supplied fracture volume and fractal dimension; two-stage breakdown pressures and hydraulic injection work described initiation accessibility and energy demand. The CRITIC information weights were 0.261 for stress compatibility, 0.235 for injection-work demand, 0.182 for fluid mobility, 0.172 for specimen temperature, and 0.150 for pressure accessibility. The fitted second-order model identified positive temperature–fluid (0.122) and temperature–stress (0.071) non-additive coefficients, while the pressure–work coefficient was close to zero (−0.008). Sequential injection produced a consistent stage-dependent difference in breakdown pressure, with the second-stage value exceeding the first-stage value in every test. This observation is interpreted descriptively within the fixed sequential injection protocol. Condition No. 5 (SC-CO2, 200 °C, 3 MPa) ranked first with a fracability index of 0.935 and a 95% Monte Carlo interval of 0.908–0.945. Leave-one-out cross-validation gave a Spearman coefficient of 0.830 and R2 = 0.860. Within the tested domain, temperature was associated with improved fracability, particularly under the low stress-difference condition where enhanced SC-CO2 mobility and thermal effects acted together. The framework therefore distinguishes coupled benefits, redundancies, and uncertainty that an additive score would conceal.