Article pubs.acs.org/est
Permeability Reduction Produced by Grain Reorganization and Accumulation of Exsolved CO2 during Geologic Carbon Sequestration: A New CO2 Trapping Mechanism Andrew J. Luhmann,†,* Xiang-Zhao Kong, Benjamin M. Tutolo, Kang Ding, Martin O. Saar, and William E. Seyfried, Jr. †
Department of Earth Sciences, University of Minnesota, 310 Pillsbury Drive SE, Minneapolis, Minnesota 55455, United States S Supporting Information *
ABSTRACT: Carbon sequestration experiments were conducted on uncemented sediment and lithified rock from the Eau Claire Formation, which consisted primarily of K-feldspar and quartz. Cores were heated to accentuate reactivity between fluid and mineral grains and to force CO2 exsolution. Measured permeability of one sediment core ultimately reduced by 4 orders of magnitude as it was incrementally heated from 21 to 150 °C. Water-rock interaction produced some alteration, yielding sub-μm clay precipitation on Kfeldspar grains in the core’s upstream end. Experimental results also revealed abundant newly formed pore space in regions of the core, and in some cases pores that were several times larger than the average grain size of the sediment. These large pores likely formed from elevated localized pressure caused by rapid CO2 exsolution within the core and/or an accumulating CO2 phase capable of pushing out surrounding sediment. CO2 filled the pores and blocked flow pathways. Comparison with a similar experiment using a solid arkose core indicates that CO2 accumulation and grain reorganization mainly contributed to permeability reduction during the heated sediment core experiment. This suggests that CO2 injection into sediments may store more CO2 and cause additional permeability reduction than is possible in lithified rock due to grain reorganization.
■
amounts of CO2 (e.g., Kumar et al.4). Solubility trapping occurs when CO2 dissolves into brine at the CO2-brine interface. This form of trapping becomes more important with time, and dissolution rates can increase due to convection produced by differences in density between brine and CO2-saturated brine in permeable formations (e.g., Ennis-King and Paterson5). Mineral trapping incorporates CO2 into minerals through fluid-mineral reactions and provides the greatest storage security. Mineral trapping can occur relatively quickly in basalt or ultramafic rock (e.g., McGrail et al.;6 Matter and Kelemen7) or over much longer time scales in siliciclastics (e.g., Xu et al.8). In general, any process that reduces permeability following injection helps prevent CO2 from escaping to the atmosphere. Permeability varies by orders of magnitude in nature and is one of the most important parameters in terms of injectivity, storage, and escape of CO2 during carbon sequestration. Thus, we need to understand permeability changes that occur during
INTRODUCTION Atmospheric CO2 has now surpassed 390 ppmv1 from a preindustrial concentration of 280 ppmv. CO2 has a particularly long lifetime in the atmosphere, and elevated levels could persist and thus affect climate for more than 100 000 years.2 Geological carbon sequestration involves injecting CO2 into geological formations to reduce anthropogenic CO2 emissions, and thus limit the currently increasing atmospheric CO2 concentrations. One sequestration option involves capturing CO2 from stationary power plants and injecting it into subsurface, high-permeability saline formations below lowpermeability caprocks in deep (>800 m) sedimentary basins. Ideally, permeability near injection wells remains high during injection to facilitate the transfer of large amounts of CO2 into the subsurface. Once underground, CO2 optimally remains isolated from the atmosphere via a series of trapping mechanisms that start with structural and stratigraphic as well as capillary trapping, continue with solubility trapping, and eventually reach mineral trapping.3 Structural and stratigraphic trapping is ultimately dependent on the integrity of the caprock and nearby boreholes. Capillary trapping occurs when disconnected CO2 ganglia become immobile due to capillary forces depending on CO2-brine surface tension and CO2mineral wettability, and this mechanism can store significant © 2012 American Chemical Society
Special Issue: Carbon Sequestration Received: Revised: Accepted: Published: 242
July 31, 2012 October 11, 2012 November 9, 2012 November 10, 2012 dx.doi.org/10.1021/es3031209 | Environ. Sci. Technol. 2013, 47, 242−251
Environmental Science & Technology
Article
Figure 1. (a) Hydrothermal flow system and (b,c) flow-through reaction cell, shown both (b) with and (c) without the pressure vessel. One end of the delivery and exit lines in (b) or (c) is inserted into the Teflon sheath holding the sample, and the other end is inserted into one of the pressure vessel covers. O-rings on both ends of the delivery and exit lines facilitate axial pressure on the sample.
accumulations of uncemented material remain (e.g., Land et al.;33 Gluyas and Cade34) that could potentially serve as CO2 storage reservoirs. Furthermore, several researchers35−38 have demonstrated advantages of CO2 injection into offshore sediments. We conducted experiments to simulate geologic carbon sequestration in uncemented material. More specifically, we heated CO2-saturated fluid flowing through sediment cores to assess how fluid-mineral grain interaction and CO2 exsolution affect core permeability. Previous CO2 exsolution experiments have been conducted on rock cores,39,40 and Zuo et al.40 observed no statistical correlation between exsolved CO2 saturation and porosity following exsolution with their nonflowing experimental design. In this study, we exsolve CO2 by increasing system temperatures to produce two phases in the core. Temperature-induced exsolution of CO2 could occur during lateral or vertical migration of fluids with dissolved CO2, but our results are equally applicable to exsolution that occurs via pressure or salinity changes.19,20 We also conducted a similar experiment on a solid rock core to identify differences produced by the lack or presence of cementation.
CO2 injection produced from both physical and chemical processes. CO2 injection studies in siliciclastic reservoirs have produced rapid changes in water chemistry from mineral dissolution,9−11 unless carbonates and feldspars are rare components.12 Carbon sequestration experiments using common siliciclastic minerals have generally shown dissolution of carbonates, feldspars, mica, quartz, and clays and/or precipitation of clays and carbonates.13−17 Luquot et al.18 document dissolution of feldspars, laumontite, and chamosite, and precipitation of kaolinite, silica, ankerite, siderite, and amorphous carbon during carbon sequestration experiments through a chlorite/zeolite-rich sandstone, reducing permeability by 1 order of magnitude. In subsurface CO2 storage reservoirs, brine and CO2 will frequently exist in close proximity as two separate phases. Even if CO2 fully dissolves into the brine, slight changes in pressure, temperature, or salinity may cause CO2 exsolution and the reemergence of two-phase systems.19,20 CO2 existence in pores is directly correlated with porosity in multiphase (water and CO2) flow experiments,21,22 and experimental work has shown that CO2/water flow in siliciclastic rocks is generally waterwetting and CO2-nonwetting and thus capable of trapping significant amounts of CO2 as capillary trapping (e.g., Pentland et al.;23 Krevor et al.24). Furthermore, CO2 plumes larger than capillary trapping ganglia have been shown to accumulate below or behind reservoir permeability heterogeneities in both numerical25 and experimental studies.26 Zoback and Gorelick27 have recently argued that carbon capture and storage to reduce greenhouse gas emissions is risky because seismic events will likely follow large CO2 injection into brittle rocks, compromising seal integrity. However, these investigators27 assert that sequestration may be safe and successful in formations like the Sleipner gas field in the North Sea, where CO2 has been injected into the largely uncemented28 and unconsolidated29 Utsira Sand since 1996. Brittle faulting is unlikely in this type of reservoir.27 Thick sediment accumulations can be prevalent along continental margins and in other basins.30,31 At an extreme, sedimentary fill of the South Caspian Basin approaches ∼30 km in thickness.32 While diagenesis has lithified some of these deposits, large
■
EXPERIMENTAL SYSTEM AND METHODS A schematic of our hydrothermal flow system is shown in Figure 1a. CO2−saturated water at room temperature (21 °C) is maintained by keeping pressurized CO2 continuously in contact with the experimental fluid in a separator. A key element of the experimental system is the use of computercontrolled Teledyne ISCO syringe pumps. Pumps A, B, and C are model 260D pumps, and pump D is a model 500D pump. Pumps A and B operate in tandem to provide defined flow rates through the core sample. Pump C is used to define the pressure at the core outlet, while Pump D provides confinement pressure around the sample. The sample core is held within a stainless steel pressure vessel, surrounded by four independently controlled Watlow band heaters for temperature control. Because the experimental design facilitated CO2-saturated water at 21 °C in the separator, exsolution of CO2 in the core and the nearby area occurs when the pressure vessel is heated. Two high-precision Heise pressure transducers near the 243
dx.doi.org/10.1021/es3031209 | Environ. Sci. Technol. 2013, 47, 242−251
Environmental Science & Technology
Article
grains indicates normalized feldspar mineralogy that consists of 95.8% K-feldspar, 4.0% albite, and 0.2% anorthite. Minor micas (likely muscovite, glauconite, and biotite) are present in thin section, and minor rutile was identified during EMP analyses. The first sediment experiment (Experiment 1) was run for 55 days using fluid flow rates that ranged from 0.01 to 1.5 mL/ min. Fluid flowed through the CO2 liquid in the upper portion of the glass separator, facilitating CO2 saturation at 21 °C (see Supporting Information (SI) for additional fluid details). Flow rates were generally changed to confirm that flow was laminar. Flow rates