Injection of Supercritical CO2 at Citronelle Field, Mobile County, Alabama, for Carbon Utilization and Storage: Fluid-Rock Interactions and Preliminary Modeling Results

Donahoe, Rona J.; Donovan, Theodore; Weislogel, Amy

A pilot project was conducted as part of the SECARB phase III program to study supercritical CO2 injection as a means of EOR and carbon sequestration. The study site selected is Citronelle Field, a mature petroleum field which has produced oil from the Early Cretaceous Donovan Sand of the Rodessa Formation since its discovery in 1955. Although tertiary CO2 flood was tested on a very limited basis in the 1980s, large-scale CO2 flood has not been conducted. Injection of a 7500 ton slug of CO2 began without interruption in January 2010, ended in September 2010, and was followed by waterflood. The CUS pilot project afforded an opportunity to study the geochemistry of aqueous fluids during and after CO2 injection, and fluid-rock interactions induced by the injection of supercritical CO2. Water samples collected between June 2010 and February 2012 from 4 producing wells located spatially around the injection well were analyzed for major, minor and trace element concentrations, and the chemical data used for geochemical modeling of the observed spatial and temporal concentration trends. Increases in Ca, Br and Fe concentrations and pH decrease were measured during and after CO2 injection for most/all of the observation wells. Geochemical modeling of these samples using PHREEQC indicates that ion exchange reactions, rather than mineral dissolution reactions, provide the best explanation for the observed temporal element concentration trends. Precipitation of iron oxyhydroxides may explain the observed low injectivity to water following the injection of CO2. TOUGHREACT modeling efforts have included 2-D models and 1-D radial models utilizing water chemistry and approximating the reservoir's physical and chemical characteristics. A working physical model has not yet been obtained for this complex system, which is heterogeneous and appears to be dominated by fracture flow. Realistic breakthrough times could not be achieved within one order of magnitude. A 10 yr, simplified, 1-D radial model has demonstrated differences in concentration gradients between different ions. Input precipitation and dissolution kinetic data have not proven useful in model outputs because of the steep concentration gradients between injection water and saline reservoir brine predicted by the model. Interpretation of PHREEQC saturation index data, in conjunction with mineral weight percentages, will likely be more useful for interpreting fluid-rock interactions taking place in the reservoir.

AAPG Search and Discovery Article #90163©2013AAPG 2013 Annual Convention and Exhibition, Pittsburgh, Pennsylvania, May 19-22, 2013