ML-Based Co-Optimization Framework Improves CO2 Sequestration and Oil Recovery

_ This article, written by JPT Technology Editor Chris Carpenter, contains highlights of paper SPE 224150, “A Machine-Learning-Based Co-Optimization Framework Improves CO2 Sequestration and Oil Recovery,” by Kassem Alokla, SPE, James Omeke, SPE, and Esuru Rita Okoroafor, SPE, Texas A&M University, et al. The paper has not been peer reviewed. _ This work combines CO2-enhanced oil recovery (EOR) methods with subsurface containment strategies to permanently store CO2 while simultaneously increasing cost-effective oil production from reservoirs. The study develops a novel workflow with multiobjective optimization techniques to assess the integration of pressure-management methodologies for permanent geological CO2 storage in saline aquifers. Methodology Reservoir Model Description. A homogeneous reservoir model, representative of typical Gulf of Mexico formations, was constructed using a nonisothermal modeling code for the purpose of the project. The model features a shale layer separating an oil reservoir from an aquifer, allowing the assessment of simultaneous CO2-EOR and CO2 sequestration. The model was designed under a compositional multicomponent system using the EOS-PVT E300 simulator. Pure CO2 was selected as the injection fluid for the simulation. The relative permeability functions for water/oil and gas/oil systems were calculated based on Stone’s approach to define the related coefficients. Simulation Approach. The static model consisted of 42,966 grid cells with 36, 62, and 21 grids in the I, J, and K directions, respectively. For equilibrium initialization, this model used a datum depth of 6,000 ft and a pressure of 3,375 psi. For the base case, the average permeability and porosity for the aquifer and reservoir were 106 md and 17%, respectively. Permeability was considered as a function of porosity. The shale formation was assumed to be completely impermeable, and the anisotropy ratio of modeled permeability in both reservoir and aquifer was 0.1. The base case was developed to test the performance of the system and investigate the ranges of the parameters needed for the optimization process. For that purpose, the wells considered for the aquifer were a CO2 injector and two water producers. For the reservoir, one oil producer (PRO1) and two injectors for water-alternating-gas (WAG) purposes (CO2_Oil and WAG_INJ) were used. For sequestration into the aquifer, a total of 32.787 million scf/D of CO2 was injected through CO2_AQ for 10 years. To balance the pressure and mitigate pressure buildup above the fracture gradient, water was produced through the two water producers. This water was reused for water injection into the reservoir. As part of the base case, the WAG process was designed with cycles of 8-month CO2 injection followed by 8-month water injection for a total of 10 years. The injection rates for both were 32.787 million scf/D and 11,000 STB/D, respectively.