Effects of methane-carbon dioxide replacement on the mechanical properties of natural gas hydrate reservoirs

While recovering methane (CH4), the exploitation of natural gas hydrates (NGHs) with replacement by carbon dioxide (CO2) can result in sequestering of the greenhouse gas CO2 underground and the generation of CO2 hydrates in reservoirs. This replacement process enhances the mechanical properties while preventing the reservoirs from becoming unstable. Methane hydrate-bearing sediment (MHBS) samples with different initial hydrate saturations were prepared by the in situ formation method. By carrying out replacement experiments with CO2 and triaxial compression tests of separate replacement degrees, this research determined the mechanical properties of the samples after replacement and established the strength criterion of NGH reservoirs under CO2 replacement conditions. The results demonstrate that the stress-strain curves of the hydrate-bearing sediment (HBS) samples before and after replacement are hyperbolic and show strain-hardening characteristics, and plastic failure without clear peak strength occurs in all samples. Within the replacement duration of 0–20 h, the replacement ratio has a change whose absolute value less than 1.65% affected by effective confining pressure, and decreases by 8.70%–21.15% with the increase of initial hydrate saturation from 13% to 38%. As the effective confining pressure and the replacement ratio rise, the mechanical parameters of HBS samples change approximately linearly; For the MHBS sample with effective confining pressure of 1 MPa and initial hydrate saturation of 13%, after the replacement ratio reaches 0.69, the stiffness and strength of the sample increase by 29.44% and 30.84% respectively, and the Poisson's ratio decreases by 10.51%. The initial hydrate saturation and replacement ratio mainly influence the failure strength by affecting the cohesion of the HBS samples. The calculated curves based on the established strength criterion are in good agreement with the experimental results, with a relative error of 0.29%–6.18%.