油页岩
机械
断裂(地质)
粘度
边值问题
地质学
水力压裂
床
流体力学
石油工程
地温梯度
超临界流体
材料科学
岩土工程
热力学
复合材料
地球物理学
物理
各向异性
古生物学
量子力学
作者
Jie Yang,Hamdi A. Tchelepi,Anthony R. Kovscek
标识
DOI:10.1016/j.jgsce.2023.205109
摘要
Supercritical carbon dioxide (sc-CO2) is an alternative to water for stimulation of low permeability systems such as shale gas and geothermal resources. Previously core-scale experimental studies have compared the behavior of CO2 to water injection for sample breakdown. Due to differences in experimental setup and core sample preparation, inconsistent or even apparently contradictory conclusions have resulted. To reconcile this contradiction, a phase-field numerical model is applied to understand hydraulic fracturing experiments using Green River shale found in the literature. The finite element numerical model incorporates a rate-dependent phase-field fracture model developed separately to describe fracture initiation and growth. We investigate the impact of various material and fluid properties on the resulting fractures. Most importantly, we study the effect of fluid properties and boundary conditions on the breakdown pressure, including the direction of the resulting fracture plane. Model results predict that (1) sc-CO2 injection in the laboratory may result in greater breakdown pressure than that of water under no-flow boundary conditions because lower viscosity sc-CO2 may result in pressure build up at the core boundary that opposes fracture initiation and (2) lower viscosity sc-CO2 also produces fast-propagating fractures that are less influenced by the bedding plane on their resulting fracture topology. Our model offers a straightforward explanation and reconciliation of existing experimental observations, as well as a means to extrapolate to new conditions. Exploration of field-scale conditions suggests less pronounced or no elevation in breakdown pressure when sc-CO2 is injected because the pressure build up effect at the system boundary is significantly less or absent at field length scales.
科研通智能强力驱动
Strongly Powered by AbleSci AI