Response of Heterogeneous and Fractured Carbonate Samples to CO2-Brine Exposure

Monday, 15 December 2014: 11:05 AM
Harris E Mason, Megan M Smith, Yue Hao and Susan Carroll, Lawrence Livermore Nat'l Lab, Livermore, CA, United States
Carbonate rock units are often considered as candidate sites for storage of carbon dioxide (CO2), whether as stand-alone reservoirs or coupled with enhanced oil recovery efforts. In order to accept injected carbon dioxide, carbonate reservoirs must either possess sufficient preexisting connected void space, or react with CO2-acidified fluids to produce more pore space and improve permeability. However, upward migration of CO2 through barrier zones or seal layers must be minimized for effective safe storage. Therefore, prediction of the changes to porosity and permeability in these systems over time is a key component of reservoir management. Towards this goal, we present the results of several experiments on carbonate core samples from the Wellington, Kansas 1-32 well, conducted under reservoir temperature, pressure, and CO2 conditions. These samples were imaged by X-ray computed tomography (XRCT) and analyzed with nuclear magnetic resonance (NMR) spectroscopy both prior to and after reaction with CO2-enriched brines. The carbonate samples each displayed distinct responses to CO2 exposure in terms of permeability change with time and relative abundance of calcite versus dolomite dissolution. The measured permeability of each sample was also much lower than that estimated by downhole NMR logging, with samples with larger fractured regions possessing higher permeability values. We present also our modeling approach and preliminary simulation results for a specific sample from the targeted injection zone. The heterogeneous composition as well as the presence of large fractured zones within the rock necessitated the use of a nested three-region approach to represent the range of void space observed via tomography. Currently, the physical response to CO2-brine flow (i.e., pressure declines with time) is reproduced well but the extent of chemical reaction is overestimated by the model.