Results And Discussion

Approximately 25 samples were collected during the M 1 (March 2001), M 5 (July 2002), M 7 (March 2003), and M 9 (September 2003 partially analysed, not discussed) surveys for 87Sr/86Sr analyses. 87Sr/86Sr ratios found in the produced fluids from the Mississippian Midale beds fall between 0.7077 and 0.7082, consistent with published values for Mississippian fluids and carbonate minerals (approximately from 0.7076 - corresponding to 330 Ma ago - to 0.7082 corresponding to 360 Ma ago - Bruckschen et al., 1995; see Fig.1, Table 1).

Most of the sedimentary formation waters migrated over long distances prior to entrapment in the reservoir rocks from which the waters are sampled. Nearly two-thirds of the formation brines have higher 87Sr/86Sr ratios than marine Phanerozoic seawaters, whose values range between 0.7068 and 0.7062 (Veizer and Compston, 1974; Notsu et al., 1988).

Table 1. Sr isotope data for the M1, M5, M7 Weyburn oil-waters.

Monitor 1 Monitor 5 Monitor 7

Table 1. Sr isotope data for the M1, M5, M7 Weyburn oil-waters.

Monitor 1 Monitor 5 Monitor 7

87Sr/86Sr

Min

0.70775

0.70800

0.70797

Max

0.70825

0.70818

0.70818

Mean

0.70798

0.70804

0.70807

S87Sr

Min

-0.63

-0.28

-0.32

Max

0.08

-0.03

-0.03

Mean

-0.31

-0.23

-0.19

Manville

Mississipian

1st survey

3 survey

87Sr/86Sr 0.7073 0.7082 0.70798 0.70807

System

Hydrostratigrapliy

Quaternary

Upper Aquifer System (1.4 <TDS < 46 gfl)

9 %

Tcrtiaxy

e

Crciaccous Aquiurd System

?

S

Viking Aquifer (< 36 g/l)

Jouli Fou Aquitard

*% ;

0.7873

140 M a

McxxviHeAqiigfr System f*8i

♦ ■ \î

0.7072

Jurassic 200 Ma

Jurassic Aquifer <4 - 150 £>'l}

X »

Triassic

Misstssippiiin Jurassic Aquitard System

! ;

Permian

Pennsylvania«

/

1 350 Ma _

/

0.7076

-Aijjsirsjwi« Aquifer System f<330 tfl: Weyiwa: 40-130

Uakkcn Aquitard

Devonian

Devonian Aquifer System (< 300 g/1)

Prairie Aquiclud«

Winmpcgosis Aquifer <40 - 330 g/1}

Silurian

Silurian Devonian Aquitard

Ofdovician

Basal Aquifer Sysiem

Cambrian

(TDS - 335 gyl)

Prccambnan

Basemenl

Figure 1. Williston Basin hydro-stratigraphy and nomenclature (modified from Bachu and Hicheon, 1996). Arrow (I) indicates that water from the Mannville Aquifer System is used for water-flooding. The symbols (?) are added to leave in question the use of upper aquifers for water-flooding. Arrow (II) indicates that water produced with oil is continuously re-injected.

The Pennsylvanian seawater (over the Mississippian ones) has 87Sr/86Sr values around 0.7083 (normalised to NBS 987 = 0.710255; Popp et al., 1986; Denison et al., 1994; Bernaby et al., 2004).

The Mannville aquifer, used since 1959 for water flooding of the Weyburn oil-field, is a Cretaceous arenite, with 87Sr/86Sr values between 0.7072 and 0.7073 which corresponds to an age of 140 Ma BP (Jones et al., 1994). A small component of the recycled fluid is derived from the Mannville aquifer and is re-injected into the Midale beds via a water-alternating-gas (WAG) EOR technique, as mentioned. Strontium isotopic compositions of the produced water prior to water flooding (i.e., prior to 1959) are not available. Other case histories describe the use of Sr isotopes to discriminate water-flooding injection components from the formation water (i.e., Smalley et al., 1988), mostly when the injected water has a Sr content and 87Sr/86Sr ratio which is different from the in-situ formation water. The lowest values of the Sr isotopic ratio found in the oil waters could represent a higher degree of mixing between the two aquifers (Fig. 2). We started to analyse the oil water Sr isotopes in 2001, around 40 years after the beginning of water flooding. Despite this fact we found Sr values of the oil water in the range of the Mississippian host rock (Midale reservoir), although a certain percentage of the Sr isotopic ratio could be referable to the Mannville component. Average Sr ratios and mass balance calculations suggest that as much as 25% of the produced fluid in 2001 was derived from the Mannville aquifer, decreasing to 15% in 2003.

The progressively more positive (i.e. higher) trend in the Sr isotopes from 2001 to 2003 may be due to: i) a smaller Mannville aquifer component in the water flood over the last three years and/or ii) the enhanced dissolution of Mississippian host rock during progressive CO2 injection.

If the leaching of Mississippian host rocks is increased (as a consequence of CO2 injection) it could shift the values towards the pre-water-flooding Sr isotopic "baseline" values, affected by 40 years of water-flooding which added a 25% Mannville component. In general terms, the lowest Sr isotopic ratio values found thus far may represent a higher "contamination" of Mississippian Midale fluids by re-injected Mannville flooding water. It may involve progressively lower Sr isotopic values which may coincide with the highest injection volumes of Cretaceous waters, as there is no natural communication or mixing between the two aquifers (no fluid pathways have yet been found in the highly impermeable Jurassic-Triassic Watrous Formation). Considering this general process, we observed instead a progressive decrease of the Mannville "contamination" from 2001 to 2003. Therefore, the more probable scenario is that the 87Sr/86Sr values are progressively approaching the Mississippian host-rock values and may point towards zones of carbonate dissolution resulting from continued CO2

injection. Carbonate minerals within the Midale beds, which were precipitated from waters during the Mississippian, would have Sr ratios consistent with what would be considered a "baseline" isotopic composition. The leaching of strontium during dissolution of these minerals would drive the Sr composition of the produced waters to heavier values, thereby masking the Mannville aquifer contribution.

In summary, the sectors (Fig. 2) characterised by the lowest Sr isotopic ratio values represent "contamination zones" of Mississippian Midale fluids by re-injected Mannville flood water. These zones exhibit lower Sr isotopic values that may coincide with the highest injection volumes of Cretaceous water. Concurrently, the average field-wide 87Sr/86Sr values are approaching Mississippian host-rock values and may point towards sectors of the field (Fig. 2) characterised by higher carbonate dissolution, as a result of continued CO2 injection.

This hypothesis is strengthened by the 813C data available up to September 2004, as well as by other chemical data of previous surveys (Hutcheon et al., 2003 and papers in http://www.uregina.ca/ghgt7/). Using 813C values of produced bicarbonates and CO2, the University of Calgary has outlined both injected CO2 and carbonate mineral dissolution in the reservoir.

The available Sr-C isotopic ratios in the produced fluids (up to September 2004) corroborate the hypothesis of chemical and isotopic input from dissolved carbonates, although further surveys could modify this statement relative to early stages of the water-rock interactions in the oil reservoir. Initially, the injected CO2 had a distinctive 813C signature of -35%o., however since 2002 CO2 recycling and re-injection has changed this signature to between -20 to -25%. Therefore the second strontium isotope scenario outlined above fits well with the 813C data. The Sr isotopic ratio in produced fluids tends to increase with time, suggesting input from dissolved strontium bearing carbonates.

Sr isotopic ratios of minerals (anhydrite, dolomite, calcite of the Midale and Vuggy Formations) from the Weyburn reservoir Phase A1 area are available from Queen's University. These data are consistent with the produced Mississippian brine because they range from 0.70567 to 0.708995, with an average value of 0.708134 ± 0.0001563. It is possible to assume that the dissolution of these minerals would push the Sr isotopic ratio of the oil waters from values closer to 0.7073 (Cretaceous Mannville aquifer) towards the reservoir values of 0.7076-0.7082 (Mississippian Midale-Vuggy), again strengthening the second Sr isotope theory mentioned above.

Figure 2. Contoured Sr isotopic ratio maps (a1s b1s c1) from 2001, to 2002, up to 2003.

If a number of sources exist, evolving from varied mixed compositions, it could be represented by binary diagrams as 1/Sr versus where coexistent linear trends are often found (Chaudhuri and Clauer, 1993). An isotopic trend with a very high positive slope could be explained in terms of the mixing of Sr derived from two sources which are markedly different in both their isotopic composition and Sr contents. On the other hand a low positive slope in the isotopic trend could be explained by either mixing between two sources of Sr or dilution of a single source of Sr. In the Weyburn case, considering only the first two analysed surveys (Sr available only for M1-2001 and M5-2002; see Fig. 3 a), it is clear that a unique "gross" population effectively changed in terms of mixing from 2001 (lower 87Sr/86Sr ratios and higher Sr concentrations closer to the Mannville component) to 2002 (higher 87Sr/86Sr ratios and lower Sr concentrations closer to the Mississippian component). A small group forms an anomaly of relatively low Sr values (WEY 35, 42 and 45 - wells corresponding to the vertical EnCana wells 12-25 (6-14), 14-23 (6-14), and 12-19 (6-13), respectively), which may have been caused by dilution, considering also the slightly lower salinities as a whole. Alternatively a local differential immobilization by sulphates-carbonates could be possible ("short term" geochemical modelling considering kinetic parameters is in progress, while "long term" modelling has been partially completed; Perkins et al., 2004). The WEY 5 well (corresponding to the vertical EnCana well 1-11 (6-14)) seems to be pertinent to slightly deeper strata.

87 86 87 86

Figure 3. Binary diagrams of (a) Sr/ Sr versus versus Na/Cl of the

87 86 87 86

Figure 3. Binary diagrams of (a) Sr/ Sr versus versus Na/Cl of the

Albitization of K-feldspar has been used (Chaudhuri and Clauer, 1993) to explain high-slope points in the graph, in other words direct replacement of K-feldspar or replacement of K-feldspar by some other minerals (calcite, anhydrite, etc...) which in turn are replaced by albite. This effect of albitization would result in an elevation of the Sr isotopic ratio in the waters without any significant change in the Sr content of the water. In the Weyburn case the "end member" component (as defined by Chaudhuri and Clauer, 1993) common to all the analysed waters may be identified as one whose 87Sr/86Sr ratio is closer to the Mannville component used for water flooding, progressively changed during the CO2 injection and consequent rock dissolution. This is also suggested by the 87Sr/86Sr versus Na/Cl binary diagram (Fig 3 b), where the Na/Cl ratio is higher as a whole for the M1-2001 waters than for the M5-2002 waters and vice versa for the 87Sr/86Sr ratios. As a whole, the Weyburn data do not display a trend of higher 87Sr/86Sr ratios if the Na/Cl molar ratio approaches 1:1, as found in the literature (i.e., solutes are increasingly dominated by halite dissolution, as could also occur for the Weyburn oil brines; Burke et al., 2004). Moreover the past water-flooding modified the original 87Sr/86Sr and Na/Cl binary relationship, now affected by the host rock dissolution linked to the new industrial CO2 injection.

Expected Normal Vali

Figure 4. Normal probability plots (after Sinclair; 1991) for the Weyburn "gross composition" of dissolved gases (atm. P.); (a) dissolved CO2; (b) dissolved CH4; (c) dissolved H2S; (d) dissolved H2. All data are reported as cc/L (STP), except H2S which is reported as composition % in the head-space (not as total dissolved concentration).

Expected Normal Vali

Figure 4. Normal probability plots (after Sinclair; 1991) for the Weyburn "gross composition" of dissolved gases (atm. P.); (a) dissolved CO2; (b) dissolved CH4; (c) dissolved H2S; (d) dissolved H2. All data are reported as cc/L (STP), except H2S which is reported as composition % in the head-space (not as total dissolved concentration).

Also the available "gross composition" of the dissolved gases (see the limitations in the methods paragraph detailed in Quattrocchi et al., 2003 and in Riding & Rochelle, 2005) are helpful in jointly interpreting the Sr isotopic data. Available dissolved gas data strengthens the second hypothesis, i.e. enhanced Mississippian rock dissolution. In oil brines CO2 solubility is dependent on T, P, fugacity coefficient, salinity (eq. NaCl), and detailed chemical composition. Theoretically, from chemical equilibria calculations (Czernichowski-Lauriol et al., 2001; Perkins et al., 2004), more than 1 mole of CO2 can dissolve into 1 kg of water for typical Weyburn reservoir brines (50°C, 14 MPa, salinity range from 35 to 110 g/l).

The steep salinity gradient in the Weyburn area will greatly affect CO2 solubility, migration and reactivity as demonstrated by the regional 2D modelling of natural fluids in the Mississippian aquifer (BRGM deliverables within the EC Weyburn Project, Riding & Rochelle, 2005). Normal probability plots (Sinclair, 1991) of the 2001-2004 dissolved gases in the Weyburn oil-waters (Fig. 4) exhibit, as mentioned, the "gross composition" evolution, before the "degassing correction". We found a general increase of the main dissolved gases (CO2, CH4, H2S) for the different surveys (20012004), parallel to the increase of the Sr isotopic ratio increases. In particular progressive increases of dissolved CO2, CH4 and H2S were observed, while H2 initially increased then it decreased after 2002. Dissolved He, after an abrupt early increase, decreased after 2003; the initial increasing stage was very probably due to the release of accumulated crustal He from the rockmatrix during CO2-driven dissolution (Torgersen & Clarke, 1987).

Our dissolved CO2 data are essentially consistent with the Tian et al. (2004) calculations, performed following the Material Balance Equation method (MBE). These authors stated that approximately 86% of injected CO2 can be dissolved into reservoir oil, 7% into water and 7% will occur as free gas. This occurs mainly in the early stages of CO2 injection (Gunter et al., 2000). Moreover with the increase of CO2 injection time, the percentage of CO2 dissolved in oil decreased by a small degree and the percentage of CO2 dissolved in reservoir water and as free CO2 increased slightly. This is because progressive oil recovery will result in a decrease in the average oil saturation and increase in the average water saturation. As a result, for the authors, the injected CO2 had more chance to contact (dissolve in) the water phase than the oil phase, even though under reservoir conditions the solubility of CO2 in oil is larger than that in water (b factor).

The increase of both dissolved gases and selected trace metals (Quattrocchi et al., 2003, reported in Riding & Rochelle, 2005) during the Phase 1 A is consistent with the second Sr isotope scenario outlined above, which highlights host-rock dissolution as the main process during the B0-M1-M9 (i.e. injection) period. During the first year of CO2 injection preliminary trace element data in the Weyburn oil waters showed an initial increase of Al, Be, Ba, Fe, Cr and a decrease of Ni, Cu, Zn. This different behaviour among trace metals depends mainly on redox/acidic geochemical barriers (Perel'man, 1986) developing within the reservoir and mainly around the injection wells during CO2 injection.

Our available trace metals data for the Weyburn oil-field could be interpreted within the framework of the Basinal Brine Theory (Kharaka et al., 1987), such that high dissolved metal concentrations in the presence of aqueous sulphide species require acidic solutions with pH values lower than those generally obtained in basinal brines. Very acidic solutions in some sectors of the reservoir are assured by CO2 injection. From the mixture it will lead to non-equilibrium in the reaction: H2S (gas) > H2S (aqu) and will result in the transfer of gas to the water phase (Kharaka et al., 1987) in agreement to our dissolved H2S data. This process will continue until essentially all the metals are precipitated as sulphides, as suggested by early PHREEQC chemical equilibria calculations (in progress and outlined in Quattrocchi et al., 2003). We suggest, in the frame of future Weyburn field monitoring, to merge more strictly the periodically measured parameters pertaining to different phases, i.e., dissolved H2S together with the H2S "free gas phase" (U of C data), as well as with the trace metal contents, to also study the sulphide precipitation/dissolution equilibria. Both chemical dissolution of the Mississippian Midale-(Marly)-Vuggy oil host-rocks and CO2 dissolution as part of "solubility trapping" (Gunter et al., 1993; 1997, 2000) are clearly outlined by merging the Sr isotopic ratio evolution with other isotopic and chemical data collected for Weyburn oil waters (Hutcheon et al., 2003; Perkins et al., 2004; Shevelier et al., 2004).

Was this article helpful?

0 0

Post a comment