Grantee Research Project Results

 

 

Understanding Eruption Mechanisms of CO₂-Saturated Cold Geyser:  Analogues to Failed CO₂ Sequestration Sites

 

The purpose of work associated with Subtask 1.1 is to evaluate the processes that may cause or otherwise influence leakage of CO₂ from target reservoirs to USDWs via abandoned wells or other high-permeability conduits. Rather than speculate regarding conceptual models, we started with analysis of a known CO₂ leakage site located in central Utah, just a 3-hour drive from the University of Utah. The Crystal Geyser, located near the town of Green River, Utah, emits low levels of CO₂ continuously, and also “erupts” with a significantly larger emission of CO₂ daily. Whether this system is representative of a failed CCS site is unclear, and thus we made characterization and simulation analysis of this system as a starting point for the Aquifer Risk Assessment Framework (ARAF).

 

To characterize the periodicity and eruption mechanisms of the Crystal Geyser, we collected the time-series changes of pressure and temperature by installing Geokon Model 4500 vibrating-wire piezometers during the two different periods from August 4-6 and September 1-5. The piezometer uses a pressure-sensitive stainless-steel diaphragm to which a wire is connected, the other end being attached to a fixed position in the body of the instrument. Pressure variations in the fluid deflect the diaphragm, causing it to change the tension in the wire. The wire is regularly “plucked” by an electromagnet and natural frequency of vibration recorded. This frequency then is interpreted in terms of diaphragm movement and hence pressure change.

 

 

Figure 3. SChematic diagram of well configuraten and pizometer installation.

 

 

During the early trip from August 4-6, we first measured the maximum depth where the piezometer could be installed and found out that there is an unknown obstacle at the 16 m depth (Figure 3). We were not able to determine the exact reason why the piezometer does not go deeper than 16 m in this study; however, anecdotal reports and discussions with several local people suggested that it had been dynamited to enhance the eruption ability of the adjacent geyser and railroad ties had been dumped there earlier to cap the geyser (Shipton, et al., 2004; Waltham, 2001). Likewise, during our two field-trip periods, we observed a few groups of tourists dropping by the Crystal Geyser and even tossing small rocks in it. Interestingly, the researcher who studied the CO₂ exsolution mechanism in the Crystal Geyser 20 years ago (Mayo, et al., 1991) indicated that he had no problem inserting a video camera several tens of meters or more at that time (Personal Communication, Mayo, 2010).

 

Due to the presence of the unknown obstacle, one piezometer was installed at the approximate depth of 16 m immediately above the obstacle and measured fluid pressure and temperature every 20 seconds. The piezometer was installed at 14:10 pm on August 4, and the observed static pressure and temperature was close to 0.16 MPa and the temperature was 17.3°C at that time (Figure 4). After the major eruption continued about an hour from 7:30 am to 8:40 am on August 5, we noticed that the static pressure and temperature changed to approximately 0.08 MPa and 17.0°C. This was because the groundwater discharge throughout the 0.39 m diameter well continuously for 1 hour was so powerful that the installed piezometer at 16 m depth was pushed up to an 8 m depth. Therefore, we stopped the monitoring at 12:40 pm immediately after the recharge period and began to record manually. In the second trip from September 1-5, we installed a 1.9 cm-diameter PVC pipe from the top of the well-casing to immediately above the unknown obstacle (Figure 3) and attached two piezometers to PVC pipe at the depths of 6 m and 13 m, respectively.  This configuration was designed to prevent the abrupt movement of the piezometers during the large-scale eruption. We began taking measurements at 13:10 pm on September 1, at 20 second intervals, and stopped at 10:30 am on September 5.

 

Figure 4 shows the time series changes of pressure and temperature measured from 14:10 pm on August 4 to 12:40 pm on August 5, explicitly conceptualizing the disturbance of fluid pressure and temperature during the eruption period. The piezometer installed at a depth of 16 m indicated that both pressure and temperature regularly drops during the eruption period only. The pressure and temperature remained at consisten leves (0.15 MPa and 17.3°C) when there was no eruption activity. 

Figure 4. Time-series changes of pressure and temperature collected August 4-5 2010.

 

Two distinctive patterns of eruptions (small and large eruptions) were identified (Figure 4). There was a total of 26 small-scale eruptions with an average eruption duration of 0.12 hour (7 minutes) that occurred during the 16-hour period from 15:30 pm on August 4 to 7:30 am on August 5 (Table 2) prior to the single large-scale intensive eruption, which continued about an hour. Photographs of the small- and large-scale eruptions are shown in Event B and Event D of Figure 4, respectively.

 

In general, we observed that CO₂ bubbles were actively emitted within the geyser (Event A in Figure 4) 5 to 10 minutes prior to every small-scale eruption, indicating that CO₂-saturated brine reached the boiling point and started to exsolve CO₂ gas from CO₂-dissolved brine. Despite the notable change inside the well, the water pool around the geyser stayed calm during this period. Five to 10 minutes later, the notable small-scale eruption began through the hole at the contact between the well and the travertine surface. Because the intensity of geyser eruption was still small, water discharge did not reach the casing top but flew through the hole at the contact between the well and the travertine surface. Soon, the discharge reached the casing top (Event B of Figure 4) and the eruption continued approximately 0.12 hour (7 minutes). When the small-scale eruption activity stopped, water did not bubble anymore but a notable amount of CO₂ gas was emitted instantaneously at the top of the well casing with hissing sounds (Event C of Figure 4). Interestingly, our field observations addressing the sequence of CO₂-driven cold geyser eruption were similar to the field observations and numerical experiments of temperature-driven hot geyser eruption (Ingebritsen and Rojstaczer, 1993; 1996; White, 1967). In the hot geyser, there was a period of liquid-only discharge at the beginning of the eruption, and then the condition changed to a two-phase flow in which both water and steam were emitted together. At the final stage, the water discharge stopped and only steam was emitted.

 

Table 2. Description of the eruption and recharge sequences during two field-trip periods

Field Trip	Starting Time	Ending Time	Eruption Type	Duration	Comment
8/4 - 8-6	8/4 - 15:30 pm	8/5 - 7:30 am	A-type	16 hours
	8/5 - 7:30 am	8/5 - 8:40 am	B-type	1.2 hours
	8/5 - 8:40 am	8/5 - 12:40 pm	Recharge	4 hour	Because the levels of static pressure and temperature were changed, we stopped the recording at the pizometer.
	8/5 - 12:40 pm	8/5 - 18:15 pm	C-type	5.6 hour	We began to record manually
	8/5 - 18:15 pm	8/6 - over 1:00 am (?)	Probably D-type	Over 4.8 hour (?)	Eruptions continued ov r1:00 am on 8/6. We do not know the exact time it stopped
	8/6 - over 1:00 am (?)	8/6 - 10:00 am	Recharge		At 10:00 am on August 6, the geyser was completely drained as shows in Event D of Figure 4
9/1 - 9/5	9/1 - 16:50 pm	9/2 - 6:10 am	A-type	13.3 hour
	9/2 - 6:10 am	9/2 - 7:10 am	B-type	1 hour
	9/2 - 7:10 am	9/2 - 10:00 am	Recharge	2.8 hour
	9/2 - 10:00 am	9/2 - 16:20 pm	C-type	4.3 hour
	9/2 - 16:20 pm	9/2 - 21:00 pm	D-type	4.7 hour	PVC pipe was attached to pizometers are broken. The level of static pressure and temperature was changed slightly
	9/2 - 21:00 pm	9/3 - 7:10 am	Recharge	10.2 hours
	9/3 - 7:10 am	9/3 - 18:40 pm	A-type	11.5 hour
	9/3 - 18:40 pm	9/3 - 19:30 pm	B-type	0.8 hour
	9/3 - 19:30 pm	9/3 - 21:50 pm	Recharge	2.3 hours
	9/3 - 21:50 pm	9/4 - 3:50 am	C-type	6 hours
	9/4 - 3:50 am	9/4 - 9:10 am	D-type	5.3 hours
	9/4 - 9:10 am	9/4 - 19:50 pm	Recharge	7.7 hours
	9/4 - 16:50 pm	9/5 - 6:39 am	A-type	10.7 hours
	9/5 - 6:30 am	9/5 - 7:30 am	B-type	1 hours

 

 

 

Single Large-Scale Eruption:  A single large-scale eruption, characterized by large reduction of both pressure and temperature and, at the same time, longer eruption duration, occurred from 7:30 am to 8:40 am on August 5 after a series of multiple small-scale eruptions (Figure 4). We differentiated between large- and small-scale eruptions based on eruption intensity, duration, and drainage period. First, the intensity of the large-scale eruption was strong and, due to the power of instantaneous CO₂ gas emission, the water column reached more than 10 m above the surface (Event D in Figure 4). The water only reached the well casing top in the small-scale eruption (Event B in Figure 4). The reduction of both pressure and temperature was significantly large in large-scale eruption compared to the small- scale eruption. Second, the average duration of small-scale eruptions was approximately 0.12 hour (7.2 minutes) but that of large-scale eruption was 1.2 hours. Finally, we did not observe any change of height in the water pool around the geyser immediately after the small-scale eruption although there was a slight decrease of hydrostatic pressure. In contrast, after the large-scale eruption ended (lasting 1.2 hours), the water pool completely drained into the geyser (Event E in Figure 4). Even the water level inside the geyser well was approximately 2.5 to 3 m below the surface.

 

Following the intensive large-scale eruption, the recharge period began at 8:40 am and ended at 12:40 pm on August 5, with the first occurrence of a small-scale eruption. We discovered that the pressure and temperature changed at this point so we stopped the piezometer measurement and recorded manually until 10:00 am on August 6. After the recharge period ended at 12:40 pm on August 5, we observed a series of small-scale eruptions (Table 2). The general characteristics of these small-scale eruptions were similar to those observed on August 4 but the duration of these small-scale eruptions prior to the single large-scale eruption was only 5.6 hours, which was approximately three times shorter than the previously measured period (which was 16 hours). The single large-scale eruption began at 18:15 pm on August 5 and continued until 1:00 am on August 6, a duration of more than 6.8 hours. Although we did not observe the exact time when this large-scale eruption stopped, the well was completely dry when we visited it at 7:00 am the next morning (August 6).

 

Identification of Single- and Large-Scale Eruptions:  From the observations made during the August 4-6 field trip, we identified four different eruption patterns (A, B, C, and D) and two recharge periods in the Crystal Geyser (Table 2). A-type is characterized as multiple, small-scale eruptions that continued for 16 hours. The B-type eruption occurred immediately after the A-type eruption period. The B-type eruption was characterized by a single large-scale eruption that continued for 1.2 hours. At the end of the B-type eruption, the water level inside the well was 2-3 m below the surface. There was a recharge period of approximately 4 hours before the C-type eruptions began. The C-type eruptions were multiple, small-scale eruptions similar to A-type eruptions but the duration of the C-type eruptions was shorter (5.6 hours). After the end of C-type eruptions, there was a single, large-scale eruption (D-type eruption), which continued for more than 6.8 hours. This eruption type pattern was confirmed during the second field trip conducted from September 1 to 5 and will be discussed further.

 

Coupled Nature of CO₂-Driven Geyser Activity: Based on our observation of the multiple, small-scale eruptions, we found that these eruptions had three regular periods:  pre-eruption, eruption, and recharge (Figure 5). In addition, the changes in pressure and temperature during the eruption period were closely correlated, similar to observations of the other CO₂-driven cold geyser in New Zealand (Lu, et al., 2005).

 

During the pre-eruption period, the pressure was consistent but a 0.1°C increase in temperature was observed (Figure 5). The slight temperature increase was considered to be caused by the increase in kinetic energy due to CO₂ bubbling in the system and the recharge of relatively hot groundwater from deeper formations. There is no CO₂ bubbling (as shown in Event A of Figure 4) observed within the geyser prior to the pre-eruption period, which implies that the slight increase of temperature is somehow related to CO₂ bubbling.

 

During the eruption period, we observed that both pressure and temperature dropped instantaneously (Figure 5). The reduction of hydrostatic pressure indicates that a certain portion of the water column above the piezometer is displaced with CO₂ bubbles. In addition, the instantaneous temperature drop is caused primarily by the cooling contribution from the Joule-Thomson effect and endothermic CO₂ exsolution. The Joule-Thomson effect accounted for the temperature drop due to the instantaneous expansion of CO₂ gas volume induced by a pressure reduction during the isenthalpic condition (enthalpy change is zero). This effect has been studied previously (Han et al., 2010; Katz and Lee, 1990; Oldenburg, 2007). In the Crystal Geyser, CO₂ gas is exsolved from CO₂-saturated brine at a certain depth below the surface. Immediately after the nucleation of the CO₂ gas bubbles, the CO₂ gas will migrate vertically through the well's water column because of their buoyancy. As the CO₂ gas migrates closer to the surface, the hydrostatic pressure falls and the volume of CO₂ gas increases. If the series of processes occurs instantaneously, we would expect a substantial temperature decrease. Further, it is well known that the CO₂ dissolution process is an exothermic reaction within the appropriate range of enthalpy variations from -440 to -200 kJ/kg at temperatures from 25 to 100°C (Carroll, et al., 1991; Duan and Sun, 2003; Ellis, 1959; Ellis and Golding, 1963; Koschel, et al., 2006]. Because the reaction is exothermic, heat evolves as the CO₂ dissolves in the brine. The CO₂ exsolution from liquid is simply the opposite process of CO₂ dissolution in liquid. In principle, the corresponding enthalpy change in CO₂ exsolution is going to be the same as that with CO₂ dissolution except that it is defined as an endothermic reaction; the brine cools as CO₂ gas exsolves from the fluid. The slight temperature decrease observed during the eruption period occurs because of the cooling contribution from both the Joule-Thomson effect and CO₂ exsolution. Interestingly, we found that the temperature decreased again after the recharge period. We think this delay is because the temperature response in the system is slower than the pressure change.

 

Figure 5. Time-series changes of pressure and temperature

 

 

General Characteristic of Eruptions—Intensity and Duration: The observation from the second field trip from September 1-5 confirmed the eruption cycle pattern observed during the first field trip and, thus, we were able to construct a clearer eruption cycle in the Crystal Geyser (Figure 6). As discussed in the previous section, “Identification of Single- and Large-Scale Eruptions,” the data from the September field trip also confirmed that the eruption cycle is composed of four different eruption types (A-, B-, C-, and D-types) and two recharge periods (Table 2 and Figure 6). Two eruption cycles were captured during this field trip and the duration of these two cycles was 38.3 hours and 33.6 hours, respectively.

 

 

Figure 6. Time-series changes of pressure and temperature collected at September 1-5 2010

 

 

Within the eruption cycle, both B- and D-type eruptions are characterized as large-scale eruptions, which occurred immediately after the end of the A- and C-type eruptions. Interestingly, there were distinctive differences between B- and D-type eruptions, including eruption duration, recharge period, and the characteristics of the multiple small-scale eruptions that occur prior to the large-scale eruption. First, the D-type eruption was always longer than the B-type eruption (Figure 6 and Table 2). Within the first eruption cycle in Figure 6, B- and D-type eruptions lasted for 0.8 and 4.7 hours, respectively. Second, D-type eruptions required a longer recharge period before the first small-scale eruption occurred. Finally, prior to each of these two large-scale eruptions (B- and D-types), multiple small-scale eruptions (A- and C-types) occurred but the number of eruptions for A- and C-types differed. As seen in Figure 7, the number of A-type eruptions was less than the number of C-type eruptions. The three characteristics mentioned above were consistently observed during both the August and September field trips. This finding differs from the general understanding that the periodic eruption of the geyser shows a bimodal pattern (Azzalini and Bowman, 1990; Rinehart, 1980). Despite this finding, however, we were not able to quantify the exact reason why the durations of the two single large-scale eruptions were different.

 

The previous study of a temperature-driven hot geyser discovered that a change in barometric pressure could affect eruption cycles (Rinehart, 1972; Rojstaczer, et al., 2003; White, 1967). In general, although the boiling temperature of pure water at atmospheric conditions is 100°C, it varies with different pressure condition and the amount of dissolved gas content. Specific to the pressure variation, boiling temperature increases with depth as the hydrostatic pressure increases but it decreases as the atmospheric pressure is reduced. Therefore, the change of barometric pressure could affect the boiling temperature in the hot geyser, resulting in a change of the eruption characteristics of the geyser. For example, previous studies showed that, as the barometric pressure decreased, the discharge amount from the geyser is increased and the eruption interval becomes shorter (Rinehart, 1972; White, 1967). To investigate the effect of barometric pressure change, we gathered barometric pressure data from the closest weather station located at Moab, Utah (Figure 7c). During the 5 days from September 1-5, the barometric pressure reached the maximum (> 0.016 MPa) the morning of September 3, and decreased continuously until 13:00 pm on September 5. Although there was a change in barometric pressure during the field trip, we found that this external force did not control the interval between the multiple, small-scale eruptions, which became smaller as the time for the single large-scale eruption neared (Figure 7a and 7b). This specific trend, which was observed only at the Crystal Geyser, confirmed that the small-scale eruptions caused the reduction of hydrostatic pressure, which caused more CO₂ gas to exsolve from the CO₂-saturated brine. As a result, a series of multiple, small-scale eruptions (A- and C-types) could trigger a single large-scale eruption (B- and D-types).

 

Specific Characteristics (I, II, III, and IV) of the Eruption Cycle:  Four different characteristics (I, II, III, and IV), addressed in Figure 6, were found in the eruption cycles. Both characteristics I and II represent the systematic coupled-nature of pressure and temperature due to CO₂-related chemical reactions such as Joule-Thomson cooling and endothermic CO₂ exsolution. Typically, larger scale eruptions caused more significant reduction of both pressure and temperature than smaller scale eruptions (comparison of I and II). In addition, although a temperature change during the large scale-eruption appeared at both 6 m and 13 m, we found that the thermal effect is relatively small at the shallow depth. This is because the CO₂ bubbles, which were cooler than the surrounding fluid because of Joule-Thomson cooling and endothermic CO₂ exsolution, lost their heat due to thermal conduction while they migrate vertically. For this reason, thermal disturbance caused by multiple small-scale eruptions was distinctive at 16 m but not evident at 6 m. Characteristic III was specific to the process at the end of the large-scale eruptions, where the measured hydrostatic pressure increased but the temperature reached its minimum. At the end of the large-scale eruption, the temperature of the fluid in the geyser becomes cooler due to the continuous CO₂ exsolution and the adiabatic cooling processes. The decrease in temperature caused an increase in brine density, which then resulted in an increase of the hydrostatic pressure in the geyser. Finally, we observed a slight increase of temperature immediately before the multiple, small-scale eruptions (Characteristic IV in Figure 6). Again, as discussed earlier, (Figure 5), the slight increase of temperature was attributed to the increase of kinetic energy caused by CO₂ bubbling in the system and the recharge of relatively hot groundwater from deeper formations.

 

Conceptual Description of CO₂-Driven Cold Geyser Eruption Mechanisms

 

CO₂ exsolution results in the decrease of the hydrostatic pressure and also reduces the temperature during both small and large eruption periods. Based on the observations, we developed a conceptual model describing the physical and chemical mechanisms causing cyclic eruptions from the CO₂-driven cold geyser. Stage A in Figure 8 indicates the start of the geysering cycle, when the water reaches the top of the well and overflows. During the overflow process, CO₂ exsolves from the CO₂-saturated water at the depth above the critical exsolution pressure point and CO₂ bubbles begin to appear in the well. At Stage B, the well is refilled at the bottom by the recharging aquifer, which causes the water level to rise in the well. When the overflow starts, the bubbles rising with the fluid begin to decrease the hydrostatic pressure at the top of the well (see the hydrostatic line in Stage B). The water level is fixed at the wellhead, so any slight density change progresses down the well as gaseous CO₂ is released from solution and the rate of bubble growth increases. CO₂ bubbles keep coalescing and the critical CO₂ exsolution pressure point gets deeper because more CO₂ bubbles are exsolved, resulting in further decrease of the hydrostatic pressure. At Stage C, eruption begins once the depth of the critical CO₂ exsolution pressure point reaches the deepest point. Further, the hydrostatic pressure reaches its lowest point at this stage. Once the eruption is over at Stage D, CO₂ bubbles escape to the atmosphere, causing the water level within the well to decrease. In addition, the depth of CO₂ exsolution pressure point becomes shallower. Finally, CO₂-saturated brine is recharged from the bottom of well and Stage A begins again.

 

 

 

 

 

The ARAF team is evaluating multiple case studies to analyze the risks at various sites at different stages of carbon capture and storage (CCS) operation, including the following three sites:

 

The Gordon Creek field in north-central Utah, just 25 miles from the original Farnham Dome site (and thus our baseline geologic and hydrologic knowledge is very solid). This site has never been subjected to CO₂ injection, and injection will not begin for another year or more.

 

Our Southwest Regional Partnership on Carbon Sequestration (SWP) (“Phase 2”) field site in the Permian Basin in western Texas; we started CO₂ injection at this site in October 2008, and injection is not slated to cease until 2010 at the earliest.

 

Our SWP (“Phase 2”) field site in the San Juan Basin in northern New Mexico; we started CO₂ injection at this site in July 2008, and injection ceased the week of August 17, 2009.

 

We plan to include all three of these field sites in the ARAF STAR project to some extent, but our priority is the Gordon Creek field, and thus it will receive most of our time and effort. These three sites represent the three primary permutations of a CCS geologic site, to facilitate evaluation of different stages in development of the ARAF. Specifically, Gordon Creek is a new project area slated for future CO₂ injection, and thus it represents a site with no baseline CCS analysis or data at the onset of the project. The Permian Basin site represents an active CO₂ injection and monitoring site with a relatively mature set of fundamental geologic evaluation, monitoring and simulation data to support ARAF development. Finally, San Juan represents a site that has run the full course of CO₂ injection from drilling to well closure (albeit a medium-scale demonstration).

 

The Gordon Creek field was a primary focus for the first year of the ARAF project because it represented a site in pre-injection mode.

 

 

 

The Gordon Creek project area is one of the three field sites proposed in the EPA-sponsored ARAF STAR project. Gordon Creek field is a proposed large-scale CO₂ injection site of the SWP. Gordon Creek field is a natural gas field located in T14S, R7-8E, Carbon County, Utah, that produces methane at an estimated rate of 8,900 million cubic feet per day with a CO₂ concentration of 98.82% (Morgan and Chidsey, 1991).

 

Gordon Creek field is located on the eastern flank of the Wasatch Plateau. The closed anticline structure trending north-south in the field is approximately 9 miles long and 5 miles wide and serves as a structural trap for methane gas in the Ferron Sandstone and possibly CO₂ in the White Rim. The geology of the proposed site exhibits a stacked system of alternating reservoirs and seals, which provides optimal conditions for the geologic storage of CO₂. The target CO₂ injection formations are the Jurassic Navajo and Entrada formations due to their intermediate depth and current use of salt water disposal into the Navajo.

 

 

Figure 9. Location map of Gordon Creek project area, Carbon County, Utah.

 

 

In this task, we focused on developing a site-specific static model based on the data collected specific to the Gordon Creek site. The location of the Gordon Creek site is illustrated in Figure 9. We gathered the stratigraphic formation top picks, well information, and well log images available from the project site. The distribution of wells in the site is shown in Figure 10. The well log images were digitized into the digital format (LAS).

 

 

 

 

For the static geologic model development, we utilized the formation tops data tied to isochore surfaces. The resulting static model within the model boundary is shown in Figure 11. The model domain covers 8.2 km by 9.2 km in the x and y direction, centered at the proposed injection well. The constructed model contains 13 formations starting with the Cretaceous Mancos shale on top to the Kaibab formation at the bottom.

 

Figure 12 illustrates the grid setting of our target formations, Navajo and Entrada. Currently, the grid configuration is 41 x 46 x 65 cells in the x, y, and z direction, respectively, with a cell dimension of 200 m x 200 m. Five cells were equally spaced vertically in each formation. The average thickness of target formations within the model domain is found to be approximately 19.2 m and 63 m, respectively for Navajo and Entrada formations.

 

 

 

 

 

 

Using the initial Gordon Creek 3-D grid, we evaluated conditions under which overpressures might form in the Navajo Sandstone, the primary storage target reservoir. Because we have very little data (e.g., permeability, porosity, etc.) for the area to date, this initial analysis is a sensitivity study by definition.

 

For this area, the fracture gradient approaches 80% of lithostatic pressure, and thus these simulations provide an initial glimpse regarding conditions that will lead to seal failure via fractures. This initial analysis will serve as a starting point for development of probability density functions for this pre-injection site. Figure 13 suggests that for the specific thickness and geometry of the preliminary model, overpressures may develop only if permeabilities are less than approximately 5 x 10^-17 m², for any significant CO₂ injection rates (greater than 1000 tons per year, for example).

 

Likewise, once CO₂ is injected in the reservoir, the viscosity of the water/brine/CO₂ mixture changes as composition changes. While the viscosity of pure (liquid) CO₂ at storage-reservoir conditions will not exceed 0.1 cP, water/brine viscosities can approach 1 cP, even higher if oil is in place. Figure 14 indicates that fluid mixtures must approach relatively high viscosities (> approximately 1.5 to 2 cP) to favor overpressure development. These general results are corroborated by re-plotting the results of normalized overpressure as a function of viscosity and permeability (Figure 15).

 

 

 

 

 

 

 

 

 

 

 

Analysis of CO₂ Leakage into Overlying Formations

 

Given the presence of a major fault located 500 m from the planned injection site, we will develop PDFs for risk of (a) leakage through this fault, and (b) induced slip on this fault, following CO₂ injection; we will deliver these PDFs and a detailed description of the process and algorithms used to develop these PDFs. Thus, the Subtask 3.2 activities during this first year consisted of two main components:

A generalized experimental design approach was developed to generate response surfaces and subsequently PDFs of important outcomes (overpressure, free CO₂ concentration, etc.) as functions of significant input variables (porosity, permeability, anisotropy, etc.). Detailed methodology with results will be discussed in the next progress report. In this report, we provide results of the effect of fault structure on CO₂ leakage and possible consequences to USDWs. More details can be found in Pasala (2010).

 

Faults and fractures provide one of the primary risk factors for leakage in an engineered system. Injected CO₂ may find its way out of the repository and into USDWs. In this part of the project, faults in various settings were examined to determine their influence on CO₂ flow with reservoir rock and fault-affected rock. Fluid flow simulation results in this study help to constrain how engineered sites might best be designed to reduce the risk of leakage, and to examine the possible consequences of a leaking system.

 

The Navajo sandstone is a good analog for reservoir simulation and study. Porosity values of the Navajo sandstone can be as large as 20 to 30%. Permeability values measured in the Navajo sandstone typically range from 100 to 1000 millidarcies (md), with values as large as 8000 md (Freethy, et al., 1991). Average permeability values for Navajo sandstone of 450 to 860 md also have been reported. Joints are features that can either enhance permeability or reduce it.

 

Joints are extension-induced mode I fractures that exhibit only opening displacement (white features in Figure 16). Deformation bands are planar, light-colored, vein-like features formed of ground and crushed sandstone grains and fine powder (dark features in Figure 16). Both joints and deformation bands are found as single features, or in networks, aligned sub-parallel to fault zones. Joints or deformation bands may be found alone, or together, leading to the possibility of a wide range of deformation-induced permeability variation within and near faults.

 

 

 

 

Figure 16. Typical arrangements of deformation-induced features in faulted sandstone. Arrows show direction of fault slip. Black features are individual/ amalgamated deformation bands. White features are open fractures.

 

A sealed case with a high permeability fault and its impact on the CO₂ leakage is being investigated. The model volume used in all of the above simulations was 600.5 feet by 420 feet in plan and 280 feet thick. Vertical, impermeable faults are assumed to surround a sandstone aquifer that is sealed by low permeability rocks both above and below the aquifer. A 35 x 60 x 40 (84,000 grid blocks) Cartesian grid was used for all simulations. The 35 grid blocks in the x direction are of lengths 16 x 25, 3 x 15, 7.5, 5.0, 3.5, 2.25, 1.5, 1.0, 1.5, 2.25, 3.5, 5.0, 7.5, 15, and 4 x 25 feet. The 60 grid blocks in the y direction are of lengths of 7 feet each. This makes the areal dimension of the field 600.5 x 420 feet. In the vertical (z) direction, there were 40 layers of 7 feet each. A vertical, north-south trending fault is located 100 feet west of the injection well and fully penetrates the reservoir/aquifer. The fault is 15.5 feet wide. Fault grid blocks with sides as small as 1 foot are included in the model. Such fine grid blocks resolutions are needed to capture the small physical size of fault zone details.

 

The reservoir is basically divided into three sections, upper reservoir, sealed middle section, and lower reservoir, as shown in Figure 17. The thicknesses of the upper reservoir and middle seal were 70 feet each.

 

The injection well used a rate constraint of 100 MSCF/day. Injection was uniform and was selected to occur over the entire thickness of 140 feet (of the reservoir) for 500 days, which means the total amount of gas injected is 50,000 MSCF (50 MMSCF). The basic idea was to observe CO₂ flow paths and was to investigate a long-term CO₂ injection study.

 

The success of CO₂ sequestration by deep injection depends on the ability of upper and lateral seals to confine the over-pressured fluid. One potentially serious problem associated with injection into underground formations is the possible leakage of injected CO₂ through or along faults into USDWs. Over long time scales, these faults/fractures may serve as short-circuit pathways for ultimate leakage into upper formations. This is of great concern because leakage of CO₂ in large quantities would compromise the integrity of USDWs.

 

 

Figure 17. Front view of the reservoir simulator set-up in the reservoir with horizontal seal case.

 

 

Permeable fault-related fractures in sandstone can focus CO₂ flow to leakage points where permeable faults breach reservoir-sealing units and enable injected CO₂ to leak into overlying USDW formations. On the other hand, low-permeability faults can restrict access to fault-bounded CO₂ sequestration volumes. The total amount of gas injected for 500 days was 50,000 MSCF (50 MMSCF). The top and the bottom layers have the same reservoir properties.

 

Depending on whether the fault zone is characterized by the deformation band or the joint, the permeability of the fault zone could vary by as much as six orders of magnitude (Figure 18).

 

The sealed middle layer has a permeability and porosity of 0.01 md and 7%. The two end member cases for fault permeability structure as discussed earlier were studied, the high-permeability fault containing open fractures and the low permeability fault containing deformation bands.

 

Figures 19 and 20 show vertical views of dissolved CO₂ distributions for a deformation band and fractured fault case after 100 and 500 days. The injected gas spreads out in the bottom layers and does not escape into the top reservoir due to reservoir seal. A low permeability fault even though cut across the entire reservoir does not allow any leakage of CO₂ into the overlying formations.

 

Figure 21 shows the volume of CO₂ sequestered in the three layers. In the fractured fault case, CO₂ has leaked to the top layer as opposed to in the deformation band fault. The fractured fault broke the seal and migrated into the top layer. About 22.75% of the injected CO₂ has leaked into the upper layer in the fractured fault case. On the other hand, in the deformation band (DB) fault, 87.3% of the CO₂ resided in the bottom layer. Basically, there is negligible CO₂ migration into the top layer in this case. Hence, the DB fault has almost no leakage into the USDWs compared to the fractured fault.

 

 

 

Figure 18. Variation in equivalent porosity and permeability of a 1m³ rock block for two different host rocks as low-k deformation bands (DBs: _ = 0.07, k = 500 mD, 1 mm thick) and high-k fractures (Fracs: aperture = 0.1 mm) are added. The upper line is for host k of 500 mD and porosity of 25%. The lower line is for host k of 50 mD and _ = 0.15. ‘'DB'’ identifies deformation bands and 'Frac'’ identifies fractures.

 

 

 

Figure 19: Dissolved CO₂ distribution in a stacked reservoir system after 500 days for a deformation band fault case.

 

 

 

Figure 20: Dissolved CO₂ distribution in a stacked reservoir system after 500 days for a fractured fault case.

 

 

 

Figure 21: Volume of CO₂ sequestered in the top, middle, and bottom layers in a deformation band, and a fractured fault case.

 

 

Task 4 -- Hydrodynamic Analysis of Tracer, Microbiologic, Chemistry and Physical Field Data and Use of These Data to Calibrate Quantitative Risk Assessment (e.g., probability of USDW contamination based on a certain event or combination of events)

 

 

 

The purpose of Subtask 4.1 (noble gas portion of the project) is to measure the ⁴He content of minerals obtained from drill core or cuttings, relate this to pore water concentrations of ⁴He via solubility relationships, and then utilize the derived pore-water concentrations to estimate natural rates of fluid flow. We will evaluate our results by comparing them with more traditional estimates of fluid flow (e.g., based on Darcy’'s Law), along with results from numerical simulations and with other environmental tracers such as stable isotopes of oxygen and hydrogen in water.

 

Our objective for year 1 of the project was to develop a methodology for (1) measuring the helium content of quartz, (2) determining a sample specific partition coefficient, and (3) calculating the in-situ concentration of helium in pore waters. The first step in this process was to develop a method for separating appropriately sized quartz grains from bulk caprock samples.

 

We have successfully obtained high-purity quartz separates from shales using the following procedure.

 

Approximately 50 grams of each core was disaggregated using a shatterbox for 15 seconds. The samples then were repeatedly rinsed in deionized (DI) water, allowed to settle, and decanted to remove clay particles followed by drying at ambient temperatures. Samples then were dry sieved at 850, 150, and 43 µm. The > 850 µm fraction was disaggregated further with a porcelain mortar and pestle before re-sieving. The 43-150 µm fraction (~10 g) then was treated with 10% nitric acid in an ultrasonic bath for 5 minutes to dissolve carbonate minerals. This was followed by rinsing in DI water, settling, and decanting to remove fine particles. The samples then were treated for 5 minutes with 5% hydrofluoric acid (HF) in an ultrasonic bath and rinsed with DI water, settled, and decanted. This was repeated two to three times. The HF selectively dissolves clays minerals with minimal reaction with quartz. The samples then were dried at ~75°C before magnetic separation using a Frantz machine to remove mafic minerals (those rich in magnesium and iron) such as amphiboles and pyroxenes. This step is particularly important as mafic minerals could retain internally produced ⁴He and are not likely to be in solubility equilibrium with pore fluids.

 

The above procedure generally yielded between 1.0 and 1.5 g of purified sample of ~95% quartz and the remainder being feldspars with trace amounts of clay. Figure 22 expresses a photomicrograph from a petrographic microscope of a typical sample. The field of view is approximately 500 microns (left side to right side) and the mounting media is clove oil with a refractive index of 1.543. The upper image is with the polarizer crossed while the lower image is with plain light with a single polarizing filter. The dark spots on some of the grains (when viewed under plain polarized light) are trace amounts of clay minerals that could be removed with additional treatments in HF; however, they are unlikely to change either the measurement of ⁴He dissolved in the quartz under in-situ conditions, or our determination of the helium partition coefficient.

 

 

 

 

Figure 22. Photomicrographs of quartz separates obtained from cores of shales (caprock). The upper image was taken with the polarizer crossed while the lower image was with plain light with a single polarizing filter. The field of view is approximately 500 microns from the left to the right side.

 

The ⁴He content of quartz separates was measured by heating the samples at 290^oC in a custom high-vacuum vessel designed specifically for this project (Figure 23). The vessels were constructed by boring a 1/4” outer diameter (OD) copper tube to an inner diameter (ID) of 0.225”. A stainless steel frit with a pore-diameter of 10 microns was installed inside the copper tube, and then sealed using a swaging tool. After the sample was placed inside the vessel, the end was sealed using a cold welding tool. The other end of the copper tube was flared to 37^o in order to make a high-vacuum seal when connected to the helium analytical system. The samples are evacuated to high vacuum (< 10–8 torr) for more than 48 hours to remove any residual water and noble gases that are either dissolved in the water or trapped as a gas phase by the residual water. The stainless steel frit is particularly important for this evacuation as it prevents loss of sample and protects the high-vacuum system from particulates.

 

 

 

Figure 23. High vacuum container for measuring helium in quartz. The sample is located in the right side of the container. A stainless steel frit with a pore size of 10 microns (shown above the sampler for illustration) has been swaged into the copper tube. The frit retains the sample within the tube especially during initial pumpout. The metal pinch clamp can be opened and closed numerous times and still achieve a high-vacuum seal. The left end is flared to produce a high-vacuum seal when connected to the helium analytical system. The container allows both the initial measurement of in-situ helium in the quartz and the impregnation of tank helium to be made in the same container. Also, because of the potential for impregnating the container with trace amounts of helium, the sample is removed after impregnation and placed in a new container before measuring the amount of impregnated helium.

 

One sample has completed the entire measurement cycle with the results shown in Table 3.

 

Table 3: Results of helium and neon measurements on sample B-1-2698.2

Step	4He (ccSTP/g)	20Ne (ccSTP/g	R/Ra	Comments
Initial heating at 290 oC for 4.5 days	1.59 X 10-7	2.45 X 10-8	0.108	Total sample mass = 0.5390 g
Additional heating at 290 oC for 7 days	1.41 X 10-8	1.99 X 10-8	0.148	Result shows that approximately 90% of He was removed during initial heating
Filled to 3.596 Torr with pure helium, heated to 290 oC for 11 days. Placed in new container and heated at 290 oC for 23 days.	8.76 X 10-7	1.36 X 10-8	0.169	Mass placed in new container = 0.3068 g

 

Following the approach of Lehmann, et al. (2003) the partitioning coefficient can be expressed as the helium-accessible volume (V2) of quartz by applying the ideal gas law as:

 

V₂=V₁(p₁/p₂) (T₂/T₁) where,
V₁ is volume of helium (at STP) per cm3 fo quartz tested (determined by measuring mass and a standards density of 2.67 g/cc for quartz
p₁ = standard pressure for gasses (1 atm)
p₂ = helium pressure in the vessel
T₁ = standard temperature fo rgases (273.15 K)
T₂ = inpregnation temperature (563 K)

 

The helium-accessible volume (V₂) will be calculated for each sample used in the impregnation tests. Because the solubility of helium in quartz is influenced by microstructures, we expect this value to differ from sample to sample, even if the samples come from the same location in the caprock. However, V₂ will subsequently be used to estimate the helium concentration in pore water (C_w) as

 

S = solubility of pure helium in water
C_qtz = concentration of helium in quartz (before the impregnation test)
V₂ = helium accessible volume
T_frm = temperature in the caprock formation at the location of the sample (estimated using the geothermal gradient at the site along with field measurements)
T₁ = standard temperature for gases (273.15 K)

 

The helium accessible volume calculated using Eq. 1 is 0.00102 ccSTP/cm³_qtz. This value is similar to the values determined by Lehman, et al. (2003) even though the helium impregnation pressure was approximately 10,000 less. The pore water concentration calculated using Eq. 2 (with a helium solubility of 9.5 X 10^–3 ccSTP/cm³) is 5.4 X 10^–6 ccSTP/cm³_qtz. This value is approximately 5 times lower than the estimate obtained using the Osenbrück, et al. (1998) method. These two values are still relatively similar given that helium concentrations in natural pore waters varies by more than a factor of 100,000. Nevertheless, this difference will be investigated in the coming months by impregnating quartz at various helium pressures, and by using various methods for obtaining the quartz separates. One possibility is that the initial crushing of the bulk shale samples has released some helium trapped in micro bubbles.

 

 

There was no work on Task 5 completed this year.

 

 

There was no work on Task 6 completed this year (just the “building blocks” are taking form).

 

 

 

 

The following tasks are proposed for eduation and outreach: 

 

The intent of the ARAF is a formalized, practical methodology for characterizing risks to USDWs and mechanisms (in the form of optimizing engineering injection conditions) for minimizing those risks. The ARAF will take the form of a documented protocol for USDW risk assessment published in a summary report. We also will develop a set of companion Web pages to provide practical access to the ARAF. At the end of the project, we will deliver this report and these Web pages to EPA, but we plan to continue updating the framework over the long term.

 

Task 1:  Prepare a summary report of leakage mechanisms and processes for natural analogue leakage sites and a summary report of characterization of a natural CO₂ storage (non-leaking) site. The planned site for this analysis and report is the Farnham CO₂ Dome in central Utah.

 

Task 2:  We will use these general results to limit the “parameter space” required for development of PDFs for induced seismicity. We will begin analysis of risk of induced seismicity for the Gordon Creek site during this next quarter. We will prepare a summary report of Gordon Creek model simulation analyses for basic CO₂ migration and leakage assessment (these simulations will be the primary tool for developing PDFs associated with selected risks). If time affords, we also will develop summary reports of analyses of the other two field sites (San Juan Basin, NM, and SACROC field, TX).

 

Task 3:  Given the presence of a major fault located 500 meters from the planned injection site, we will develop PDFs for risk of (a) leakage through this fault, and (b) induced slip on this fault, following CO₂ injection. We will deliver these PDFs and a detailed description of the process and algorithms used to develop these PDFs.

 

Generalized Risk Registry (tabulation of risks) for sites with USDWs present.

Detailed (specific) risk registry for the Gordon Creek field site.

 

Task 4:  (1) Prepare a summary report of methods and associated validation approach for tracer, microbiologic, chemical and physical field data used for calibration of quantitative risk assessment; (2) prepare a brief summary report of the approach for application of tracer, microbiologic, chemical and physical field data in reservoir models; (3) prepare a brief summary report of approach for the application of tracer, microbiologic, chemical and physical field in developing risk estimates (in the form of PDFs).

 

Task 5:  Prepare a summary report of efficacy of gathered data for refining risk estimates in the form of PDFs.

 

Task 6:  The outcome of Task 6 will be the primary outcome of the entire project: the final ARAF will take the form of a documented protocol for USDW risk assessment published in a summary report. We also will develop a set of companion Web pages to provide practical access to the ARAF. At the end of the project, we will deliver this report and these Web pages to EPA, but we plan to continue updating the framework over the long term.

 

Task 7:  Prepare annual reports of progress, followed by the final summary project report.

 

Task 8:  (1) Develop a website targeting K-12 students for education regarding geologic CCS, USDWs, potential risks to USDWs, and plans to mitigate those risks; (2) make presentations at annual project review meetings; and (3) make presentations at national and regional conferences to describe the outcomes of this research to colleagues and other stakeholders.

 

 

 

Determine autotrophic growth under CO₂ flux:  In this laboratory-scale experiment, we will use a pure culture of an autotrophic organism. The purpose is to simulate the growth of the autotrophic bacteria under the flux of CO₂ in case CO₂ leakage occurs to the upper fresh water strata. This experiment will determine the growth kinetics of this autotrophic organism under the flux of CO₂ in which case, the organism will be grown in the presence of inorganic carbon source (primarily bicarbonate) and in the absence of the inorganic carbon source but under a CO₂ flux. The data obtained from this experiment will be available for modeling purposes. This task will be jointly conducted by the graduate students and a post doc under the direct supervision of Dr. Goel.

 

Analyze water samples from Gordon Greek site for biogeochemistry:  During actual CO₂ injection tests, we will obtain integrated water samples from the point of injection and will analyze those for common water chemistry parameters and bacteria. During well drilling for CO₂ sequestration purposes, we also will try to obtain water samples and aquifer material at the subsurface level where fresh water aquifer exists. Water samples will be used to evaluate the biogeochemical parameters. The aquifer material will be used to simulate the actual in-situ conditions that exist in the fresh water aquifer and will be subjected to CO₂ fluxes to record the changes in the biogeochemistry of water. Graduate students will be responsible for conducting all the tests related to the geochemistry and the water quality, whereas all experiments related to the bacterial identification and isolation will be conducted by the post doc.

 

Educational outreach:  A significant component of the project involves educational outreach and results dissemination. Dr. Ramesh Goel will coordinate with all other PIs to achieve educational outreach for the project to K-12 and undergraduate students. He will organize a summer workshop for K-12 students and teachers to update them about the importance of CO₂ sequestration and about the issues related to groundwater contamination. During regular semesters (Fall 2011 and Spring 2012), Dr. Goel will work with the College of Engineering’'s outreach office (Ms. Deidre Schoenfeld) to go to minority serving schools to present a seminar about CO₂ sequestration science and ground water contamination. Dr. Goel also will help Dr. McPherson on maintaining an active Web page for the project. The post doctoral fellow will help Dr. Goel in this task.

 

 

1. Finalize development of laboratory methods:  Although much progress was made during year 1 with regards to developing a laboratory method for analyzing helium in quartz, several important aspects need additional development. Dr. Solomon and a graduate student will evaluate the linearity and variability of the helium-quartz isotherm by performing partitioning tests on various quartz separates. This may make it possible to simplify our field testing if we find that a single isotherm can apply to all quartz mineral. We also will compare quartz-derived pore water helium values with those obtained from the method developed by Osenbrück, et al., 1998.

 

2. Obtain core samples from sites that have published values for helium pore water concentrations:  Dr. Solomon has arranged to obtain core samples from Dr. Jim Hendry at the University of Saskatchewan collected near the King research site. A detailed profile of helium pore water samples has been collected at this site. We will perform mineral separations and measure helium in quartz on these core samples. We then will evaluate the difference between quartz and direct pore-fluid methods.

 

3. Obtain core samples from Gordon Creek:  Dr. Solomon and a graduate student will obtain core samples from the Gordon Creek site as part of the drilling operation this summer. We will perform mineral separations and measure helium in quartz. This will allow us to calculate the pore water helium concentration, which then will be used to estimate long-term fluid flux through caprock formations at this site.

 

4. Develop a generic simulation of the helium transport:  We will develop a numerical model of helium transport through a generic caprock. This model will be used to test our primary hypothesis that the helium pore-water concentration is a function of long-term fluid flow and hence the permeability of the formation.

 

 

1. Use the methodology developed to calculate the PDFs to determine risk of leakage into USDWs:   Our ultimate objective is to determine the risk of leakage into USDWs. Dr. Deo will do this by computing PDFs created for an appropriate domain consisting of a target reservoir, seals and aquifers. We will populate the domain with fluid and material properties and then perform screening and initial experimental design to extract the most important parameters. From this we will set up a matrix of simulations and create a response surface. Finally, we will calculate the PDFs of outcomes (for example, CO₂ flux into aquifers) given the uncertainty in input parameters.

 

 

1. Develop and conduct reservoir simulations for natural analog sites:  Dr. McPherson and postdocs/students will continue to develop and conduct reservoir simulations to support analysis of fundamental CO₂ migration, trapping mechanisms and leakage for natural analog sites (Crystal Geyser and Farnham Dome).

 

2. Develop and conduct reservoir simulations for the Gordon Creek site:  Dr. McPherson and postdocs/students will continue to develop and conduct reservoir simulations to support analysis of fundamental CO₂ migration, trapping mechanisms and leakage for the Gordon Creek injection site. Dr. McPherson also will provide geologic, geophysical and other field data, including rock core from wells planned for drilling in the summer of 2012. The combination of these model simulations and these data will support other tasks in the project (Tasks 3, 4, 5), especially for development of PDFs to determine risk of leakage into USDWs.

 

3. Understand overpressure sensitivity at Gordon Creek site:  An activity of particular emphasis will be continued refinement of the overpressure sensitivity study began for the Gordon Creek site, as well as focused collaboration with Professor Deo to develop PDFs representing risks of induced seismicity via fault slip. Also, Dr. McPherson will continue development of the general and detailed risk registries for the project.

 

Allis, R., T. C. Chidsey, W. Gwynn, C. Morgan, S. White, M. Adams, and J. Moore (2001), Natural CO₂ reservoirs on the Colorado plateau and southern Rocky Mountains: candidates for CO₂ sequestration.

 

Altevogt, A. S., and M. A. Celia (2004), Numerical modeling of carbon dioxide in unsaturaetd soils due to deep subsurface leakage, 40.

 

Annunziatellis, A., S. E. Beaubien, S. Bigi, G. Ciotoli, M. Coltella, and S. Lombardi (2008), Gas migration along fault systems and through the vadose zone in the Latera caldera (central Italy): Implications for CO₂ geological storage, International Journal of Greenhouse Gas Control, 2, 353-372.

 

Assayag, N., M. Bickle, N. Kampman, and J. Becker (2009), Carbon isotope constraints on CO₂ degassing in cold-water geyser, Green river, Utah, Energy Procedia, 1, 2361-2366.

 

Azzalini, A., and A. W. Bowman (1990), A look at some data on the Old Faithful Geyser, Applied Statistics, 39(3), 357-365.

 

Bachu, S., and D. B. Bennion (2009), Experimental assessment of brine and/or CO₂ leakage through well cements at reservoir conditions, International Journal of Greenhouse Gas Control, 3, 494-501.

 

Barlet-Gouedard, V., G. Rimmele, O. Porcherie, and J. Desroches (2009), A solution against well cement degradation under CO₂ geological storage environment, International Journal of Greenhouse Gas Control, 3, 206-216.

 

Carroll, J. J., J. D. Slupsky, and A. E. Mather (1991), The solubility of carbon dioxide in water at low pressure, Journal of Physical and Chemical Reference Data, 20(6), 1201-1209.

 

Dockrill, B., and Z. K. Shipton (2010), Structural controls on leakage from a natural CO₂ geologic storage site: Central Utah, U.S.A., Journal of Structural Geology.

 

Duan, Z., and R. Sun (2003), An improved model calculating CO₂ solubility in pure water and aqueous NaClsolutions from 273 to 533 K and from 0 to 2000 bar, Chemical Geology, 193, 257-271.

 

Ellis, A. J. (1959), The solubility of carbon dioxide in water at high temperature, American Journal of Science, 257, 217-234.

 

Ellis, A. J., and R. M. Golding (1963), The solubility of carbon dioxide above 100°C in water and in sodium chloride solutions, American Journal of Science, 261, 47-60.

 

Emberley, S., I. Hutcheon, M. Shevalier, K. Durocher, W. D. Gunter, and E. H. Perkins (2004), Geochemical monitoring of fluid-rock interaction and CO₂ storage at the Weyburn CO₂-injection enhanced oil recovery site, Saskatchewan, Canada, Energy, 29, 1393-1401.

 

Freethey, G. W.; Cordy, G. E. U.S. Geologic Survey 1991, 1411, 118-119.

Gasda, S. E., S. Bachu, and M. A. Celia (2004), Spatial characterization of the location of potentially leaky wells penetrating a deep saline aquifer in a mature sedimentary basin, Environmental Geology, 46, 707-720.

 

Glennon, J. A., and R. M. Pfaff (2005), The operation and geography of carbon dioxide- driven, cold-water “geysers,” The GOSA Transactions, IX, 184-192.

 

Gouveia, F. J., and S. J. Friedmann (2006), Timing and prediction of CO₂ eruptions from crystal geyser, UTRep., Lawrence Livermore National Laboratory.

 

Han, W. S., G. A. Stillman, M. Lu, C. Lu, B. J. McPherson, and E. Park (2010), Evaluation of potential nonisothermal processes and heat transport during CO₂ sequestration, Journal of Geophysical Research, 115(B07209).

 

Heath, J. E., T. E. Lachmar, J. P. Evans, P. T. Kolesar, and A. P. Williams (2009), Hydrogeochemical characterization of leaking, carbondioxide-charged fault zones in east-central Utah, with implications for geologic carbon storage, in Carbon Sequestration and its Role in the Global Carbon Cycle, edited by B. J. McPherson and E. T. Sundquist, pp. 147-158, American Geophysical Union, Washington, D.C.

 

Hendry, J. M., Kotzer, T. G., and Solomon, D. K., 2005, Sources of radiogenic helium in a clay till aquitard and its use to evaluate the timing of geologic events, Geochimica et Cosmochimica Acta, Vol. 69, No. 2, pp. 475-483.

 

Ingebritsen, S. E., and S. A. Rojstaczer (1993), Controls on geyser periodicity, Science, 262, 889-892.

 

Ingebritsen, S. E., and S. A. Rojstaczer (1996), Geyser periodicity and the response of geysers to small strains in the Earth, Journal of Geophysical Research, 101, 21891-21907.

 

K. Osenbrück, J. Lippmann, and C. Sonntag, Dating very old pore waters in impermeable rocks by noble gas isotopes, Geochim. Cosmochim. Acta 62 (1998) 3041–3045.

 

Katz, D. L., and R. L. Lee (1990), Natural gas engineering, McGraw-Hill New York.

 

Klusman, R. (2006), Detailed compositional analysis of gas seepage at the National Carbon Storage Test Site, Teapot Dome, Wyoming, USA 21(9), 1498-1521.

 

Koschel, D., J.-Y. Coxam, L. Rodier, and V. Majer (2006), Enthalpy and solubility data of CO₂ in water and NaCl(aq) at conditions of interest for geological sequestration, Fluid Phase Equilibria, 247, 107-120.

 

Lu, X., A. Watson, A. V. Gorin, and J. Deans (2005), Measurments in a low temperature CO₂-driven geysering well, viewed in relation to natural geysers, Geothermics, 34, 389-410.

 

Mayo, A. L., D. B. Shrum, and T. C. Chidsey (1991), Factors contributing to exsolving carbon dioxide in ground water systems in the Colorado plateau, Utah in Geology of East- Central Utah, edited by T. C. Chidsey, Utah Geological Survey, Salt Lake City.

 

Nordbotten, J. M., M. A. Celia, and S. Bachu (2004), Analytical solutions for leakage rates through abandoned wells, Water Resources Research, 40.

 

Nordbotten, J. M., M. A. Celia, S. Bachu, and H. K. Dahle (2005), Semianalytical solution for CO₂ leakage through an abandoned well, Environmental Science and Technology, 39, 602-611.

 

Oldenburg, C. M. (2007), Joule-Thomson cooling due to CO₂ injection into natural gas reservoirs Energy Conversion and Management, 48, 1808-1815.

 

Oldenburg, C. M., and J. L. Lewicki (2006), On leakage and seepage of CO₂ from geologic storage sites into surface water, 50, 691-705.

 

Pasala, 2010, CO₂ Displacement Mechanisms, Phase Behavior Effects and Carbon Dioxide Sequestration Studies, Ph.D. Dissertation, University of Utah, December 2010.

 

Pruess, K. (2008), On CO₂ fluid flow and heat transfer behavior in the subsurface, following leakage from a geologic storage reservoir, Environmental Geology, 54, 1677-1686.

 

Rinehart, J. S. (1972), Fluctuations in geyser activity caused by variations in earth tidal forces, barometric pressure, and tectonic stresses, Journal of Geophysical Research, 77, 342-350.

 

Rinehart, J. S. (1980), Geysers and Geothermal Energy, Springer-Verlag, New York.

 

Rojstaczer, S. A., D. L. Galloway, S. E. Ingebritsen, and D. M. Rubin (2003), Variability in geyser eruptive timing and its causes: Yellowstone National Park, Geophysical Research Letters, 30(18).

 

Sheldon, A, L. and Solomon, D. K., 2003, Radiogenic helium in shallow groundwater within a clay till, southwestern Ontario, Water Resources Research, Vol. 39, No. 12, 1331, doi:10.1029/2002WR001797.

 

Shipton, Z. K., J. P. Evans, B. Dockrill, J. Heath, A. Williams, D. Kirchner, and P. T. Kolesar (2005), Natural leaking CO₂-charged system as analogs for failed geologic storage reservoirs, in Carbon Dioxide Capture for Storage in Deep Geologic Formations, edited by D. C. Thomas and S. M. Benson, pp. 699-712, Elsevier, New York.

 

Shipton, Z. K., J. P. Evans, D. Kirschner, P. T. Kolesar, A. P. Williams, and J. Heath (2004), Analysis of CO₂ leakage through ‘low-permeability’ faults from natural reservoirs in the Colorado Plateau, east-central Utah, in Geological Storage of Carbon Dioxide, edited by S. J. Baines and R. H. Worden, pp. 43-58, Geological Society, London.

 

Waltham, T. (2001), Crystal geyser - Utah’s cold one, Geology Today, 17(1), 22-24.

 

Watson, T. L., and S. Bachu (2007), Evaluation of the potential for gas and CO₂ leakages along wellbores, SPE.

 

White, D. E. (1967), Some principles of geyser activity, mainly from Steamboat springs, Nevada, American Journal of Science, 265, 641-684.