Coupled Large Scale Hydromechanical Modelling for Caprock Failure Risk Assessment of Gas Storage in Aquifer

Abstract

The rapidly increasing demand for the consumption of natural gas has attracted the interests to store natural gas in aquifer reservoir. However, natural gas injected into the aquifer reservoir, which could cause ground surface deformation and mechanical integrity destruction of caprock. Taking the aquifer gas storage of S trap as the research object, according to the geological structure and hydrogeological information, a coupling large-scale hydromechanical model is established to evaluate the damage risk of the gas reservoir in S aquifer. The proposed methodology is based on the development of fluid-solid coupling and application of FEM. The different failure mechanisms of S aquifer gas storage caprock can be evaluated on the basis of the tensile failure criterion and Mohr-Coulomb shear failure criterion. To analyze the change of caprock in gas injection and production process more clearly, a reference model is defined as an ideal calculation condition to discuss the mechanical response, pore pressure variation, and surface deformation characteristics of the caprock during injection and production. On this basis, the second scheme of sensitivity analysis is defined. The pressure injection rate, reservoir parameters, in situ stress, and other factors are considered, respectively, and the influence of different input parameters on mechanical stability and surface deformation of caprock is analyzed. Finally, the mechanical stability is analyzed and combined the above two criteria to predict the upper limit injection pressure of S. Simulation results show that the permeability and in situ stress have a significant influence on ground surface deformation and mechanical integrity of caprock, but Young’s modulus and Poisson’s ratio can be ignored; the upper limit pressure coefficient of S is 1.908.

Simulation and Prediction of Injection Pressure in CO2 Geological Sequestration Based on Improved LSO Algorithm and BP Neural Networks

“…The traditional BPNN is a local search optimization and the network weights are gradually adjusted by following the direction of local improvement, which easily leads the algorithm to fall into local extremes, thus leading to the failure of training. In addition, because the search process of the BPNN uses the gradient descent method, when the objective function to be is very complex, the algorithm will be 16 Geofluids inefficient and the convergence speed is often slow. Therefore, many researchers optimize BPNN by swarm intelligent algorithms in order to improve their convergence speed and global search capability and avoid falling into local optima.…”

“…Thus, it is necessary to quantitatively characterize the extent of the caprock damage during the injection. Many complex factors affect the caprock integ-rity in CO 2 sequestration projects, such as the injection rate and material properties [3][4][5]. Among them, the injection pressure is the key external factor triggering the geomechanical effect, whereas the mechanical properties of rocks are the objective internal factor affecting the sealing performance.…”

“…The maximum safe injection pressure in the traditional method is obtained by multiplying the formation breakthrough pressure by a safety factor (0.8-0.9). However, this method can only roughly determine a range and cannot accurately predict whether the injection pressure will exceed the critical value and lead to the risk of rupture [15,16]. In addition, the mechanical and permeability properties of reservoir caprock are obtained from the core test data, and porosity and permeability, as important parameters affecting the sequestration performance, present a complex multiparameter, nonlinear mapping relationship with the maximum injection pressure [17].…”