Search for a command to run...
Summary It is well-known that stress changes during fluid injection affect the permeability and porosity of coal. In addition, gas adsorption can further alter these properties over time. These dynamic coal properties can significantly affect injection well performance during carbon dioxide (CO2) sequestration operations targeting deep coal seams. However, few studies have attempted to isolate the effects of stress and adsorption using field data. In 2022, a field pilot was implemented in the deep (~4,900 ft) Mannville coals of Alberta to test the viability of CO2 injection and storage in deep coal seams, and mechanisms affecting CO2 injection, migration, and storage. Critical data collected for the pilot included pre-CO2 (water) injection/falloff data, performed over multiple cycles at variable injection rates, with injection pressures staying below the estimated fracturing pressure, as well as multicycle CO2 injection/falloff data collected at increasing injection rates. The pre-CO2 injection/falloff data were gathered to allow estimation of coal permeability as a function of injection pressure using a nonreacting fluid (water); the CO2 injection/falloff data were collected to evaluate the combined effects of injection pressure and CO2 adsorption. In this study, the results of the Mannville coal pilot pre-CO2 and CO2 injection/falloff test analysis are presented, and implications for CO2 injection into deep coals are discussed. Because coal fractures dilate during water injection, causing porosity/permeability to increase, and then close during the subsequent shut-in/pressure falloff, causing porosity/permeability to decrease, conventional well test solutions that do not account for dynamic properties must be modified for quantitative analysis. For pre-CO2 testing, modified pseudovariables (pseudopressure and pseudotime) are applied herein to correct for pressure dependence of porosity and permeability, allowing for accurate flow-regime identification (using log-log derivative plot). In addition, an analytical method is developed to estimate permeability with pressure during water injection and subsequent shut-in. For CO2 testing, modified pseudovariables are developed to account for multiphase flow, pressure-dependent properties, and adsorption/desorption. An analytical model is similarly developed to estimate coal permeability during CO2 testing. Pseudovariables and models developed for both pre-CO2 and CO2 tests are then verified using numerical simulation. Analysis of the Mannville pilot pre-CO2 data demonstrates that pressure-dependent properties of the coal do indeed distort the flow-regime signatures; after the application of pseudovariables in derivative calculations, a clear radial flow signature can be observed and used to estimate coal permeability from pressure falloff data. Permeability estimated at each time during falloff testing results in the observation that (1) peak permeability at the start of shut-in is significantly elevated above the permeability at initial reservoir pressure, (2) peak permeability is a clear function of injection rate, and (3) permeability returns to the initial (preinjection) permeability during the falloff period for each injection/falloff cycle. Analysis of CO2 data similarly demonstrates that pseudovariables corrected for coal dynamic properties allow for more confident flow-regime identification; further, permeability loss with cumulative CO2 injection can be quantified. Application of the proposed well-testing approach resulted in critical data that can be used to design long-term CO2 injection projects in the deep Mannville coal and provides a template for analogous projects globally.