RESULTS AND DISCUSSION

RESULTS FROM THE FRIO BRINE PILOT TESTS, TEXAS, USA

3. RESULTS AND DISCUSSION

Chemical analysis of formation water and gas samples obtained from both wells prior to CO₂ injection show that the Frio brine is a Na-Ca-Cl type water, with a salinity of 93,000 ± 3,000 mg/L TDS. The brine also has relatively high concentrations of Mg and Ba, but low values for SO₄, HCO₃, DOC and organic acid anions [19]. The high salinity and the low Br/Cl ratio (0.0013) relative to sea water indicate dissolution of halite from the nearby salt dome (e.g., [21]).

Careful measurements of the volumes of water and evolved gas obtained with downhole samplers show the Frio brine to have 40–45 mM/L dissolved CH₄, which is close to saturation at reservoir conditions (65°C and 150 bar). Gas analysis show that CH₄ comprises 95 ± 3% of total gas, but the dissolved CO₂ content of the gas is low at ~0.3% (Table 1).

During the CO2 injection, October 4–14, 2004, more than 40 water samples were collected from the observation well using the U-tube system, and results from on-site measurements of electrical conductance (EC), pH and alkalinity are discussed in [19]. The EC exhibited only subtle increase from a pre-injection value of ~120 mS/cm (at ~22°C), whereas there were major changes in some chemical parameters as the CO₂ reached the observation well, including a sharp drop in pH (from 6.5 to 5.7) and high increases in alkalinity (from 100 to 3,000 mg/L as bicarbonate). Additionally, laboratory determinations showed

major increases in dissolved Fe (from 30 to 1,100 mg/L) and Mn, and signifi cant increases in the concentration of Ca. The most dramatic changes in chemistry occurred at CO₂ breakthrough 51 hours after injection (Fig. 2), as evidenced also by on-site analysis of gas samples from the U-tube system that showed CO₂ concentrations increasing from 0.3 to 3.6% of total gas [20]. The CO₂ content of gas measured on site and in laboratory then quickly increased, reaching values of up to ~97% of total gas, with CH₄ comprising the bulk of the remaining 3%

(Table 1). Signifi cant shifts were also observed in the isotopic compositions of H₂O, DIC and CH₄ following CO

2 injection.

Results of geochemical modeling, using updated SOLMINEQ [22]

indicate that the Frio formation water in contact with the supercritical CO₂ would have a pH of ~3 at subsurface conditions. This low pH causes the brine to become highly undersaturated with respect to carbonate, aluminosilicate and other minerals in the Frio (Fig. 3 in [19]). Because mineral dissolution rates are generally higher by one or more orders of magnitude at such low pH values [23], the observed increases in concentrations of Ca and equivalent concentration of HCO₃ likely result from the rapid dissolution of calcite, as depicted in reaction (1).

CO_2(g) + H₂O + CaCO_3(s) = Ca²⁺ + 2HCO₃^– (1) TABLE 1. COMPOSITION OF GASES (MOLE %) FROM FRIO “C”

AND “B” SANDSTONES. NOTE THE RELATIVELY HIGH CO₂ IN ³“B”

Gas “C” ¹ “C” ² “B” ³ “B” ⁴

He 0.008 0 0.01 0.011

H₂ 0.040 0.19 0.92 0.012

Ar 0.041 0 0.13 0.010

CO₂ 0.31 96.8 2.86 0.28

N₂ 3.87 0.037 1.51 1.12

CH₄ 93.7 2.94 94.3 98.3

C₂H₆+ 1.95 0.005 0.12 0.11

1 background from injection well, before CO₂ injection.

2 from observation well after CO₂ breakthrough.

3 from observation well ~ 6 mo after injection.

4 from the observation well ~ 15 mo after injection.

The large increases observed in concentrations of Fe and equivalent bicarbonate alkalinity could result from dissolution of siderite, but no siderite was observed in the retrieved core. Hence these increases could be caused by dissolution of the observed iron oxyhydroxides, represented in redox-sensitive reaction (2).

2Fe(OH)_3(s) + 4H₂CO₃^o + H_2(g) = 2Fe²⁺ + 2HCO₃^– + 6H₂O (2) However, some of the increase in Fe and equivalent bicarbonate could also result from corrosion of pipe and well casing that contact low pH brine [24, 25], as indicated by redox-sensitive reaction (3).

Fe_(s) + 2H₂CO₃^o = Fe²⁺ + 2HCO₃^– + H_2(g) (3) Similar reactions may be written for Mn that increased from 3 to 18 mg/L.

There were also increases in the concentration of other metals, including Zn, Pb and Mo, which are generally associated (sorbed and coprecipitated) with iron oxyhydroxides, but could also be present in the low-carbon steel pipe used in petroleum wells [25].

The chemical data coupled with geochemical modeling indicate rapid dissolution of minerals, especially calcite and iron oxyhdroxides and possibly pipe corrosion caused by low pH values of the brine in contact with the injected supercritical CO₂. Such rapid mineral dissolution could have important environmental implications with regard to creating pathways in the rock seals and well pipes and cements that could facilitate leakage of CO₂ and brine.

FIG. 2. Electrical conductance (EC), pH and alkalinity of Frio brine from observation well determined on site during CO2 injection. Note the sharp drop of pH and alkalinity increase with the breakthrough of CO2.

10/4/04 10/5/04 10/6/04 10/7/04 5.5

5.7 5.9 6.1 6.3 6.5 6.7 6.9

pH EC HCO3

pH

0 500 1000 1500 2000 2500 3000 3500

Alkalinity HCO3(mg/L); EC (x10 mS/cm)

Maintaining reservoir integrity that prevents the ultimate escape of CO₂ back to the atmosphere by limiting its leakage to extremely low levels is essential to the success of injection operations [15]. Preventing brine and CO₂ leakage into overlying drinking water supplies is also important, because toxic organic and inorganic components are mobilized by the injected gas, in addition to the chemicals present in the pristine brine [19].

Results of chemical analysis of samples collected ~20 days, 6 and 15 months after CO₂ injection demonstrate decreases in the concentrations of Fe, Mn (Fig. 3), HCO₃ and Ca, and increases in pH. Geochemical modeling indicates that the brine pH increases from dissolution of carbonate and iron oxyhydroxide minerals discussed above, as well as from dissolution of oligoclase and other aluminosilicate minerals present in the Frio. Aluminosilicate mineral dissolution generally is not congruent, but likely follows an incongruent reaction (4), where dawsonite, gibbsite and amorphous silica are precipitated, and/or where kaolinite and amorphous silica are precipitated [1,16].

0.4H⁺ + Ca_.2Na_.8Al_1.2Si_2.8O_8(s) + 0.8CO_2(g) + 1.2H₂O

= 0.2Ca²⁺ + 0.8NaAlCO₃(OH)_2(s) + 0.4Al(OH)_3(s) + 2.8SiO_2(s) (4) As the pH increases from mineral interactions and the mixing of CO₂- saturated and pristine brines, modeling indicates that mineral saturations reverse the trend discussed above, resulting in precipitation of carbonate and other minerals. The overall result is the brine gradually evolves toward its pre- injection composition, but additional modeling is planned to further investigate gas-water-rock interactions in such a system based on results from Frio II test.

FIG. 3. Concentrations of Fe and Mn in Frio-I brine from June, 2004 to January, 2006.

Note the sharp increases in metal content during October 6, 2004, at the time of CO2

breakthrough, and slightly higher Fe and Mn in “B” samples from April, 2005.

3.1. Isotopic composition of water and gases

We observed signifi cant shifts in the isotopic compositions of H₂O and DIC following CO₂ injection, but only subtle changes in the δD and δ¹³C values of CH₄. The δ¹³C values of DIC became profoundly lighter, shifting from –3 to –33 ‰, refl ecting the fact that the injected CO₂ is the dominant C source and is depleted, with v¹³C = –34 to –44‰, depending on the mixing proportions of the two gas sources. The δ¹⁸O values of brine became isotopically lighter with time, shifting from 0.80 to –11.1‰, and there was a corresponding increase in the δ¹⁸O values of CO₂, from 9 to 43‰. Because water and CO₂ rapidly exchange oxygen isotopes even at low temperature, it is possible to use their δ¹⁸O values in mass balance equations to estimate the brine to CO₂ mass and volume ratios in the reservoir. The equation for a closed system and no isotopic exchange with minerals is given [26] by:

δ¹⁸O^f_CO2 – δ¹⁸Oⁱ_CO2 (5) δ¹⁸Oⁱ_H2O – δ¹⁸O^f_H2O X_brine/X

CO2 =

TABLE 2. CALCULATED BRINE/CO₂ VOLUME RATIOS IN THE FRIO FORMATION FOLLOWING CO₂ INJECTION BASED ON THE δ¹⁸O VALUES FOR BRINE AND CO₂.

Date

18O shift ¹⁸O shift Brine/CO2

Brine CO₂ vol. ratio*

10-5-2004 0 0 ∞

10-6-2004 0.37 32 43

10-6-2004 0.69 32 23

10-6-2004 0.77 32 21

10-6-2004 1.22 32 13

10-7-2004 2.24 32 7.1

11-3-2004 1.43 32 11

11-3-2004 1.74 32 9.1

4-4-2005 11.2 22 0.97

5-4-2005 11.7 22 0.93

6-4-2005 11.9 22 0.92

*To convert from mole oxygen basis (eq. 5) to brine/CO2 volume ratio at reservoir conditions, we multiply by 0.495, using a density (gm/cc) of CO₂ = 0.60 and brine= 1.06.

where the superscripts “i” and “f” are the initial and fi nal δ values for brine and CO₂, respectively, and X is the atomic oxygen fraction in the subscripted component.

Results from the observation well (Table 2) show that initially the system is brine dominated, with CO₂ comprising ~10% of the fl uid at reservoir conditions from one day after the CO₂ breakthrough on October 7, 2004 through November 3, 2004. However, samples collected from the injection well on April 4–6, 2005 yield a value of ~50% for the volume of CO₂ at reservoir conditions. The initial brine dominated system could indicate that the injected CO₂ acts like a piston pushing the pore water out with minimal mixing and isotopic exchange. Contact and isotopic equilibration with a larger volume of injected CO₂ is indicated from data of April, 2005. These results are comparable to ‘residual’ CO

2-saturation values obtained with the reservoir saturation (RST) and other geophysical tools [8], indicating the usefulness of this isotopic approach.

3.2. Subsurface monitoring

Monitoring at and in close proximity to the surface for CO₂leakage signal in soil gas was not effective primarily because of the induced perturbation as a result of injection operations. Signifi cant amounts of CO₂were released to the atmosphere during injection operations, and venting of CO₂with PFT and other tracer gases during the purge cycle of the U-tube sampling system, released tracer to the atmosphere and the soils around the wells. Monitoring results obtained from the four shallow groundwater wells showed rapid chemical changes during the monitoring period, with a preinjection region of high salinity water migrating down-gradient across the monitoring array.

Groundwater monitoring is continuing, but the observed chemical changes are tentatively attributed to the extraction of large amounts of groundwater for drilling and other fi eld operations and to the construction of a large fresh- water mud disposal pit [8].

Because of the anticipated diffi culty of near surface monitoring, we planned a rigorous program for monitoring immediately above the injection zone. Results of brine and gas analyses from the “B” sandstone, fi rst perforated and sampled six months after CO₂ injection, showed slightly elevated concentrations of bicarbonate, Fe, and Mn and signifi cantly depleted δ¹³C values (–5.9 to –17.5 vs. ~ –4‰) of DIC relative to preinjection “C” composition. A more defi nitive proof of the migration of injected CO₂ into the “B” sandstone is obtained from the presence of two of the four (PMCH and PTCH) PFT tracers added to the injected CO₂ [27]. Additional proof of the migration of injected CO₂ into the “B” sandstone is obtained from the high concentration

(2.9 vs. ~0.3%) of CO₂ in dissolved gas obtained from one of the two downhole Kuster samples.

Results obtained from samples collected in January 23–27, 2006 indicate brine and gas compositions that are approximately similar to those obtained from the “C” sandstone before CO₂ injection. These results indicate the absence of signifi cant amounts of injected CO₂ in the “B” fl uids sampled. However, a contrary conclusion is indicated based on the fact that PMCH and PTCH were measured in the six samples also analyzed for PFT tracers [27]. It is possible that the measured PMCH and PTCH concentrations represent desorbed PFT tracers that were introduced into “B” earlier and do not represent migration of additional injected CO₂ into the “B” sandstone.

Results from the “B” sandstone show signifi cant CO₂ migration from the

“C” sandstone. We can not rule out migration through the intervening beds of shale, muddy sandstone and siltstone, but a short-term leakage through the failed squeeze on perforations in the “C” or remedial cement around the casing of a 50-year old well is a more likely explanation. These results highlight the importance of investigating the integrity of cement seals, especially in nearby abandoned wells, prior to the injection of large quantities of reactive and buoyant CO₂.