Geophysical Monitoring of CO2 Injection at Citronelle Field, Alabama

Carbon dioxide (CO2) injection at the Citronelle oil field in Alabama has been deployed to determine the feasibility of carbon storage and enhanced oil recovery (EOR) in the depleted oil field. Citronelle is a small size city right above the oil field, hence, to detect geohazard risks, geophysical testing method using wireless sensor, and passive seismic technique is used: the non-intrusive measurements were made at well sites along two linear arrays. The outcomes of the geophysical monitoring at the Citronelle oil field are shear-wave velocity profiles that are correlated to the static stress distribution at different injection stages. Injection history interpretation using the stress wave monitoring indicates that CO2 injection resulted in the stressing of the strata.


Introduction
Carbon dioxide (CO 2 ) is a greenhouse gas, and the relationship between global warming and greenhouse gases has become more and more of a concern to the scientific community [1][2][3]. Because there is continual rise in what is already a high concentration of CO 2 in the atmosphere, it is imperative that a viable solution be implemented. Carbon capture and geologic storage is a promising method for reducing the concentration of CO 2 in the Earth's atmosphere [4][5][6][7]. The technology involves collecting CO 2 from an emission heavy source, compressing and transporting the CO 2 to a qualified site and injecting the now supercritical CO 2 at highpressure into the reservoir. The qualified storage reservoir must satisfy stringent geological storage criteria, which may include anticline, porosity, permeability, void volume, pressure limits, seepage characteristics, and cap rock characteristics. The reservoir should be sufficiently distanced from any potable groundwater acquifers to avoid contamination issues. Using underground depleted oil and gas reservoirs as CO 2 storage sites may have secondary advantages of enhanced oil recovery (EOR) or enhanced gas recovery (EGR) [8][9][10]. The additional fossil resources recovery can provide the economic incentives for CO 2 storage. Currently, there are several hundreds of oil field injection sites worldwide and in the US [11,12].
While no CO 2 leakage has been reported at any injection sites, it is important to instate careful monitoring programs for such practices to avoid potential geohazards. Recent studies of remote sensing data indicate that some of the injection sites may have experienced surface deformation [13][14][15]. Effective monitoring programs should be adopted at the injection sites in order to detect changes in geological formation (geomorphology) and the presence and migration of CO 2 within the storage reservoir [16][17][18][19].
Since 2010, CO 2 has been injected for possible CO 2 -EOR at a highly heterogeneous and discontinuous sandstone reservoir of the Citronelle oil field, Alabama. Citronelle oil field is a matured oil reservoir and an ideal site for CO 2 -EOR and sequestration, from both reservoir engineering and geological perspectives [20,21]. The Citronelle oil field is located about 50 km north of Mobile on the crest of the Citronelle Dome, which is a giant salt-cored anticline in the eastern Mississippi Interior Salt Basin (Figure 1). The field covers an area of 16,400 acres directly below the city of Citronelle. In 1955, oil was first discovered by the Gulf Oil Company in the Zack Brooks Drilling Company No. 1 Donovan well. Since then, over 500 wells have been drilled and cumulative oil production has exceeded 169 million barrels at Citronelle Field.
Citronelle is a small city with a population of 3900 (2010 consensus). With most of the oil wells integrated into the cityscape, possible geohazards such as CO 2 leaks can be detrimental to the local citizens, live stocks, and the environment. Some of the wells exist within residents' backyards and farmlands (Figure 2). To detect risks of geohazards and monitor the injection process, geophysical testing has been performed at the site. This paper reports the outcomes of the field tests due to CO 2 injection into the Unit B-19-10 #2 well (Permit No. 3232). The goal of the geophysical testing is to establish possible relationships between shear-wave velocity profiles and the static stress distribution before, during and after the injection. Such relationships are helpful in understanding the site condition changes due to the injection activity.

Geophysical monitoring strategies
The intent of the carbon injection is to explore the feasibility of EOR in the tertiary production of an existing oil reservoir, as well as the potential of subsequent CO 2 sequestration. The geophysical monitoring strategy is used to assess the site geostability and potential geohazards. Site geostability for mineral extractions can be associated either with the site geological conditions pertaining to sustained production or with the large geological deformations such as land subsidence or landslides. In the context of the current study, the association is with the latter definition. The key purpose of a geostability analysis is to determine the possibilities of significant geohazards due to formation instabilities, which may result from the CO 2 -oil replacement in the Rodessa formation.
Geo-instability of an oil-producing stratum can result from the collapse of voids during the oil extraction process. The repercussions may include global subsidence, localized straining and possible microtremors to earthquakes. Figure 3 shows a schematic of the geomaterial straining at an oil extraction well. The hypothesis is that as oil is being depleted, the surrounding geomedium may experience straining due to interfacial shear stresses resulting from the settlement and collapse of the strata. Geostability in a narrow oil field within a deep stratum, such as the Citronelle oil field, is typically not a major concern since relatively small settlement is anticipated. The geo-instability concerns in such cases can be generalized as compressibility potential assessment as well as localized stability projection.
The compression or settlement issue may involve both local elastic settlement (non-permanent deformation) and long-term creep (long-term deformation due to sustained loading). Elastic settlement is instantaneous and is a function of the weight of overburden above the layer of interest. For an oil-producing layer, elastic settlement is also a function of the system pressurization, where pressure is kept to ensure the injection fluids remain in the oil layer. Creep is difficult to assess since it is a function of time. Current geostability analysis does not include considerations of thermoelastic effects or micro-poroelastic effects for the pressurized system, for example, the Biot's equations. The rationale is that the difficulties in establishing geomaterial properties based on current geophysical measurements and assumed values only allow a grossly simplified bulk material analysis. As a result, the extensive works by Biot on poroelasticity has been greatly simplified.
Stresses in geomaterials are derived essentially from the self-weight of the overburden materials (predominantly the geo-matrix), the liquid within the voids (pore water pressure), and the externally induced pressures. Hence, the total stress within the geomaterial system is equal to the summation of stress within geo-matrix and pore water pressure. The geostability study considers the effective stress, σ', which is defined as the stress carried by geomaterial skeletons and not pore water that causes elastic deformation of the oil-producing layer: where σ is the vertical total stress derived from unit weight of material and μ is the pore pressure. Neglecting thermal effects, the effective stress equation is further modified to include injection pressure σ injection : The effective stress pressure is then used to compute the producing layer elastic deformation (non-permanent settlement) using simplified computation of rock bulk modulus (P-wave modulus), M [22]: where ρ is the rock density, V p is the P-wave velocity derived from geophysical testing conducted at Citronelle.

Geophysical testing at Citronelle field
Micro-seismicity tests have been successfully applied to address specific issues in the oil and gas industry [23,24]. The basic principle of passive seismic monitoring is to detect small movements (regarded as microseismic events) from unknown seismic sources that can be recorded on geophones placed on site. Contrast to active geophysical testing, the passive seismic monitoring is a testing method that does not rely on a source of ground excitation. The main advantage of the passive monitoring is that it can be carried out at any time and does not require regulated field access. The disadvantage of passive sensing is the uncertainty introduced due to the lack of controlled input energy, which can result in both poor data sensitivity and poor detection accuracy.
A modified passive sensing Refractive Microtremor (ReMi) technique, Derivative of ReMi (DoReMi), as discussed below, is used at the Citronelle oil field, Alabama [25]. To improve mobility and avoid the cumbersome wiring, wireless triaxial micro-electro-mechanical system (MEMS) accelerometers have been used for the field testing. The MEMS sensors are encased in hard metal boxes and buried into the ground at sufficient depth to ensure good coupling between the sensor and the surrounding soil (at least 1 ft. (0.3 m) deep with fully compacted soil on top). The wireless sensor unit with the three directional acquisition channels can record seismic energy in three Cartesian directions (vertical and two horizontal directions). The vibration signals obtained by the wireless accelerometer are acceleration time histories, which are processed in spectral domain using p-τ transformation, or slant-stack analysis [26].
Since passive sensing assumes the signals are random in nature and the analysis is done in the spectral domain, time sequence of the sampled data is not considered. Only the vertical direction has been used in the wave motion analysis for this study.
To monitor the responses of the reservoir throughout the CO 2 injection process, two linear test arrays were conducted at the Citronelle oil field. Each test array consists of 24 measurement points, which are all located near the oil wells. The site test layout is shown in Background measurement was deployed prior to the start of CO 2 injection in the field. It should be noted that in order to restore the pressure in the well to the level suitable for production, water injection at the well has been conducted since 2007. CO 2 injection in well No. B-19-10 #2 started in December 2009 and at the rate of 46.5 tons/day. The CO 2 injection was stopped from December 30, 2009 to January 26, 2010, due to the triplex pump not being able to maintain the injection pressure. After a thorough problem detection process, the pumping was resumed and, as a result, the average injection rate of CO 2 was stabilized at 31.5 tons/day. The CO 2 injection history in short tons until late September 2010 is presented in Figure 5. Final amount of CO 2 injected in the pilot well is about 8036 short tons. The record of well head pressure at Well B-19-10 #2 from the beginning of CO 2 injection to the end of the injection is shown in Figure 6. The pressure has been normalized in order to compare it with the normalized stresses at the oil-bearing layer quantified based on the geophysical testing results to be presented later.
In the first month of CO 2 injection, the well head pressure changed from 2400 psig (16,547.4 kPA) to 3800 psig (26,200.1 kPA). After CO 2 injection was resumed on January 27, 2010, the range of well head is between 3800 psig (26,200.1 kPA) and 4200 psig (28,957.9 kPA). Passive tests were conducted at the Citronelle oil field in December 2009 when the start of significant CO 2 injection,

Injection history analysis
Since the monitoring process involved the three injection stages, namely, water injection (pressure building), CO 2 injection, and post-injection, it is of interest to interpret the results according to the stages. For each stage, at least three monitoring tests were performed. Hence, there are three test group data. To compare the field pressure responses at different injection stages, statistical parameters have been adopted including average shear-wave velocities and coefficient of variations.
Statistical analysis is performed first by determining the averaged shear-wave velocities at different strata for each test group along each of the test lines. The average wave velocities are defined as [27]: where θ represents the average wave speed, x i represents the wave speed data at the corresponding depth for each test group, and N represents the number of tests in each test group. After calculating the average shear-wave velocities, the standard deviations, α, of the corresponding data are determined as: The average and standard deviation values are then used to compute the coefficient of variation (COV), C v : The coefficient of variation illustrates how far a set of numbers deviates from the average value-an indication of the consistency of the layer responses as well as the repeatability of the measurements. The CO 2 injection process instigated a continuous stress building within the oil-bearing layer at the Citronelle oil field. Due to the presence of the anhydrite layer, the CO 2 remains within the oil-bearing rock and will slowly flow into the oil-bearing rock resulting in the stress built-up dissipating throughout the oil-bearing layer. As long as the anhydrite retains its integrity and that there are no break-throughs within the rock medium, the pressure at the oil-bearing layer should be consistently higher than in the strata above. Thus, the stress wave velocity at the oil-bearing layer should be higher than in the strata above. The COV and average values of the wave speed profile will be used to determine the stress state in the strata system. Table 2 lists the C v values from both Line 1 and Line 2 tests. Table 2 shows that the C v values for each layer are reduced during the injection history, indicating a stressing of the strata.
The C v values of the wave speed at the oil-bearing layer is an indication of the stabilization of the strata pressurization process: as the oil-bearing layer pressure is building up, a larger C v value is expected, which dropped later indicating stable pressure in the oil-bearing rock.  Table 2 shows the C v values for each layer and each survey line. The table shows that in all cases, C v values are less than 0.2 indicating that the strata responses are slow, and consistent and that there are no drastic events occurring during the whole injection process. The C v values also are consistently dropping during the three stages indicating that the pressurization is gradually stabilized during the process.
Careful evaluation of Table 2 indicates that there is a difference between the results from both survey lines: for the after CO 2 injection stage, it is seen where the C v value is shown to be 0.01 for Line 1 and the value is 0.001 for Line 2 (Layer 14). This is an order of magnitude different. For the before CO 2 injection (initial water pumping) stage, where the C v value is 0.07 for Line 1 and is 0.03 for Line 2. This observation may be detrimental considering the experimental resolution of the geophysical testing method, which is discussed below.

Shear-wave speed determination
As mentioned earlier, each test line has 24 measurements points representing a total of 24 channels in data processing. To process the shear-wave velocity data, SeisOpt ReMi software was used [26]. The procedure of stress wave signal processing involves first a wave field data transformation (ReMi Vspect module was used), which converts the time domain data acquired in the field to frequency domain. An interactive Rayleigh-wave dispersion modeling was then conducted with the outcomes is 1-D shear-wave velocity models. At the end, the dispersion curves were generated [28]. Figure 7 shows the typical averaged shear-wave velocity profiles as a function of depth (measured from Line 1 and Line 2). The shear-wave velocity curve was obtained based on the averaging of the test data sets during each test stage and is shown to have a total of 14 strata. The 14th strata correspond to the measurements of shear-wave velocity to depths at around 12,500 ft. (3810 m), which is about the oil-bearing Donovan sand. As described earlier, most of the injection pressures were retained within the oil layer at around 12,500 ft. (3810 m). Hence, the test results confirmed about the pressurization of the Donovan sand and that the anhydrite layer has maintained its leak prevention integrity.
In order to compare the changes of the shear-wave velocity obtained from the geophysical tests, the data were divided into three groups: before CO 2 injection, during CO 2 injection, and after CO 2 injection. Figures 8-10 show the shear-wave velocity curves from both Line 1 and Line 2 tests for the three stages: Figure 8 shows the results of average shear-wave velocity versus depth curve for test 1, test 2, and test 3 (before CO 2 injection) for Line 1 and Line 2, respectively. Error bars are used to indicate the deviation of shear-wave velocity in the measurements of each group.
Both Line 1 and Line 2 show that different strata experienced different stress histories: for Line 1, the top seven layers (approximately at 6000 ft. (1829 m) depth) are shown to experience initial increase in wave speed (during CO 2 injection) and then decreasing wave speed (after CO 2 injection); the trend reversed after 6000 ft. (1829 m) depth showing increasing wave speeds for both during and after CO 2 injections; and finally, the oil-bearing layer [around 12,000 ft. (3658 m) depth] showed slowly decreasing wave speeds. Line 2, on the other hand, showed an increase decrease trend up to 3000 ft. (914 m) depth; followed by an increasing pattern above the oil-bearing layer; and decreasing wave speeds at the oil-bearing layer. The explanation of the response history is that the oil-bearing layer experienced strata expansion due to the injection pressure and the inability of oil to escape quick enough from the oil sand; hence, the pressure is transferred to the strata above the oil-bearing layer (mostly salient saturated material), which experienced stressing (increasing wave speed). This trend reversed for the upper layer above the salient layers, which is dependent upon the balancing act of the weight of the overburden and the upward lifting of the injection pressure. Figure 9 shows the results of average shear-wave velocity versus depth curve for test 4, test 5, test 6, and test 7 (during CO 2 injection) for Line 1 and Line 2, respectively. Wave speeds results of the last four layers shown in Figure 9 are higher than the corresponding results shown in Figure 8. The increase in shear-wave velocity is associated with CO 2 injection, which caused an increase in the effective stresses in layers above the injection zone (pressurization). Figure 10 shows the results of average shear-wave velocity versus depth curve for test 8, test 9, and test 10 (after CO 2 injection) for Line 1 and Line 2, respectively. The deviations on the graphs shown in Figure 10 are significantly smaller when compared to Figures 8 and 9 indicating that the strata pressurization has stabilized. It is important to point out that the test line selection is constrained by available monitoring sites and is selected in order to help determine possible directional effects of the CO 2 migration at the oil field. Hence, the first test line would determine the likely CO 2 migration in the north-east direction, and the second test line would determine the flow in the northeast and southwest directions. When comparing the average velocity at the oil-bearing layer, the results from Line 1 indicates that the wave speed has reached 12,392 ft./s (3,777 m/s) during water injection, 11,365 ft./s (3,464 m/s) during CO 2 injection, and has dropped slightly to 11,109 ft./s (3,386 m/s) after the CO 2 injection. This indicates that there is a possibility that the supercritical CO 2 may be migrating slowly in the north-east direction.  It is noticed that the strata pressure above the oil-bearing layer is slow in building up as it takes time for the pressure to dissipate into the upper strata. To study this effect, the wave speed responses above the oil-bearing layers are studied: it is shown for Line 1, the wave speed above the oil-bearing layer has increased from 8144.6 ft./s (2,482.5 m/s) initially, to 9512.9 ft./s (2,899.5 m/s) during CO 2 injection, and finally, increased to 9963.7 ft./s (3,036.9 m/s) post-injection. This indicates a slow building up of pressure. For Line 2, the wave speed immediately above the oil-bearing layer has increased from 8207.6 ft./s (2,501.7 m/s) before injection to 9664.0 ft./s (2,945.6 m/s) during CO 2 injection, and finally, to 9935.8 ft./s (3,028.4 m/s) postinjection. The interpretation of this observation is that the oil pressure is pushing against the strata above the Donovan sand and has resulted in the strata pressurization. It is concluded that the pressure build-ups are almost identical in both directions indicating uniform buildup of pressures at all directions at the Citronelle oil field.

Discussion on geophysical sensing for CO 2 injection studies
Geophysical testing has been applied to projects similar to the Citronelle field study for the purposes of determining production induced stress changes in the oil-bearing strata and site anisotropy changes. In most high-resolution seismic detections, the tests are performed with controlled excitations such as the use of explosions, seismobile vibrations, or gun shots. The results have sensitivities that can indicate possible migration of injected fluids. However, the interpretation of strata stress changes based on wave speed changes is inherently challenging, as a result of the constrained temporal and spatial resolutions. As a result, the velocity change ratio function (∆ v / v) has been suggested as a means to establish the detection of geomechanical condition changes due to oil production or fluid injection [29] and has been successfully implemented in a study to synchronized field measurements to localized microtremors [30].
To determine the stress wave speed changes, the velocity change functions are computed for before, during and after CO 2 injection: (9) Figure 11 shows (∆ v / v) for different stages of the injection process at Citronelle field indicating different strata stress plays: for both Line 1 and Line 2, it is shown that the stress waves have reduced in the injection layer (Layer 14) after CO 2 injection indicating that the CO 2 gas may have migrated at this stage. The velocity change functions for Layers 8-10 (corresponding to salient layer) are negative, most likely indicating a reduction in effective stress. Following Eq.
(1), this may be interpreted as an increase in pore water pressure in the salient formation.
The last figure in Figure 11 shows a comparison between Line 1 and Line 2 using Eq. (9). This figure shows that the two trends are in consistent in general and that both show the same trend of velocity increase right above the oil production strata (which shows negative velocity change functions). This further enhances the interpretation that the stress within the injection may be reduced due to migration of CO 2 plum.
A reduction of the strata pressure (shear-wave velocity) could mean a likely leak occurs within the system, which has not been identified at the Citronelle field. The shear-wave velocities at the Donovan oil-bearing layer are normalized by their average value and are plotted against the normalized well head pressure in Figure 6. Assuming the shear-wave velocity is a good representation of the stress level within the oil-bearing stratum, the wellfitting of the two sets of data represents that the geophysical testing method has accurately quantified the stresses within the reservoir.

Conclusions
Carbon sequestration through injection into a depleted oil field is an effective method to reduce atmospheric CO 2 . However, proper monitoring of the CO 2 injection process is essential in order to ensure the geomechanical stability of the storage reservoir and to minimize risks of potential geohazard to the terrestrial and sub-terrestrial environments. This chapter reports the use of a passive microseismic sensing technique to monitor the CO 2 injection process at the Citronelle oil field, Alabama. The ability of the passive DoReMi technique to monitor the CO 2 sequestration process in the heterogeneous oil reservoir is demonstrated through analysis of the wave speed profiles indicating that there are strata stress build-ups during and after the injection of CO 2 , which resulted in the pressurization of the Rodessa oil-bearing layers. Clear demarcation of the shear-wave velocity profile is shown for before, during, and after the CO 2 injection in the field. The detection of geomechanical deformation within the overburden of the reservoir is important for monitoring the long-term CO 2 storage-continued monitoring may provide information on possible reservoir breakthroughs and possible pathways for CO 2 leakage.
The COV value associated with the shear-wave velocity changes is suggested as a measure of the conditions at the oil field and is observed to drop in value during the CO 2 injection process, indicating that the stress state in the oil-bearing layer has reached a stable state. Thus, the COV values can be used as an indication of oil field stability during the CO 2 injection operation and have the potential for long-term monitoring of the strata stress change throughout the field operations. Further studies are needed to develop the COV value into risk index that can be used to indicate geohazard. The strata stressing is especially important to the City of Citronelle, where the oil wells (potential CO 2 leak sites) are in close vicinity to humans and livestock. Continued geophysical monitoring of the strata stress changes can help mitigate potential geohazards due to the CO 2 injection operation.