Beruflich Dokumente
Kultur Dokumente
ABSTRACT
High mole fraction CO2 gases pose a significant risk to hydrocarbon exploration in some areas. The generation
and movement of CO2 are also of scientific interest, particularly because CO2 is an important greenhouse gas.
We have developed a model of CO2 generation, migration, and titration in basins in which a high mole fraction
CO2 gas is generated by the breakdown of siderite (FeCO3) and magnesite (MgCO3) where parts of the basin
are being heated above approximately 330C. The CO2 reacts with Fe-, Mg-, and Ca-silicates as it migrates
upward and away from the generation zone (CO2-kitchen). Near the kitchen, where the Fe-, Mg-, and Ca-sili-
cates have been titrated and destroyed by previous packets of migrating CO2, gas moves upward without lower-
ing its CO2 mole fraction. Further on, where Fe- and Mg-silicates are still present but Ca-silicates are absent in
the sediments, the partial pressure of CO2 is constrained to 0.130 bars and reservoirs contain a few mole per-
cent CO2 as described by Smith & Ehrenberg (1989). Still further from the source, where Ca-silicates have not
been titrated, partial pressure of CO2 in migrating methane gas are orders of magnitude lower. A 2D numerical
model of CO2 generation, migration, and titration quantifies these buffer relations and makes predictions of CO2
risk in the South China Sea that are compatible with exploration experience. Reactive CO2 transport models of
the kind described could prove useful in determining how gases migrate in faulted sedimentary basins.
Corresponding author: Lawrence M. Cathles, Department of Earth and Atmospheric Sciences, Cornell University,
Ithaca, NY 14853-1504, USA.
Email: lmc19@cornell.edu. Tel: 607-255-2844; Fax: 607-254-4780.
Two issues must be clarified to understand how CO2 dioxide gas is: CO2 g ! H HCO 3 H2 O. We take
might be generated and trapped in sedimentary basins: the activity coefficient of H2CO3 (in solution) to be unity.
First we must understand where and how in basins a sepa- We do not consider the impact of solution ionic strength
rate CO2-rich gas phase can be generated. This we do in on CO2 (g) solubility (salting out effect). Pore fluid salinity
the following section by (i) considering the rock-buffered is not important because PCO2 depends on the bicarbonate
concentration of CO2 dissolved in the pore fluid as a func- activity ratio, which is fixed by the rock buffer independent
tion of temperature, and (ii) determining when and where of pH. The mass action equation for the CO2 (g) dissolu-
it could be high enough for a separate CO2-rich gas phase tion reaction is thus:
to be formed. Second, we need to understand what is
log aCO2 log PCO2 log aHCO3 aH log KCO2 g ; 3
required for the CO2 to migrate to shallower reservoirs
without being consumed by carbonate-forming reactions. and we see that the log PCO2 can be computed directly
In the section following the next, we describe our numeri- from the log of the bicarbonate activity ratio and the CO2
cal approach to this issue. dissolution log K.
Equation (4ac) expresses eqn (2) in expanded form
THE SOURCE OF CO 2 GAS for the three buffers we consider here. In these equations
the dissolution log K are evaluated at 200C using the
Above approximately 70C the pore fluids in sedimentary HKF equation of state and the SUPCRT thermodynamic
basins and in igneous geothermal systems are in equilib- database (see Johnson et al. 1992). In evaluating the
rium with the host minerals (Ellis 1970; Elder 1981; Gig- log K for the calcite buffer (4a), we assume that the pore
genbach 1981, 1997; Arnorsson et al. 1983; Henley et al. waters lie on the two-phase curve of water, as is approp-
1984; Cathles 1986, 1993; Hanor 2001; Palandri & Reed tiate in hydrothermal systems, so the pressure is 16 bars
2001). This fundamentally important observation means, at 200C. For the siderite and magnesite buffers we use
among other things, that the chemistry of the pore fluids the generalized basin pressuretemperature profile sug-
can be computed using the mass action equation as a func- gested by Smith & Ehrenberg (1989), and the pressure
tion of temperature, pressure and salinity, if the set of buf- at 200C is 1050 bars. The rows and columns of the
fer minerals is specified. The buffer minerals specify the stoichiometric matrix are labeled to indicate the buffer
activity ratios (to H+) of the basis species such as Ca++/ minerals and the basis species they represent. These anno-
(H+)2, Na+/H+, K+/H+, etc. To obtain the buffered tations are, or course, not part of the equation. The stoi-
chemistry of the solution, pH is usually determined by chiometric matrix describes the dissolution of the buffer
charge balance and the solution chemistry then calculated minerals to the basis species, so, for example, laumontite
by summing the concentrations of all complexes of the dissolves:
basis species. In the present case, however, the partial pres-
sure of carbon dioxide, PCO2, which also equals the activity CaAl2 Si4 O12 4H2 O!2Al3 Ca2 8H2 O8H 4SiO2 :
of CO2 because the activity coefficient of the dissolved
CO2 is 1, can be directly calculated from the bicarbonate with the H+ term incorporated in the activity ratio and the
activity ratio. H2O term dropped because the activity of water is unity,
The mass action equation states that the stoichiometric the result is the second row of the stoichiometric matrix in
matrix describing the dissolution of the buffer minerals, (4a).
Sm multiplied by the log activity of the basis species, 2 3
Lab, equals the log K of the buffer dissolution reactions, log aAl3 a3H
6 7
LK: 6 log aCa2 a 2 7
6 H 7
6 7
0 0 4 log aHCO3 aH 5
Sm L ab Sm Lab SH LaH SH2 O LaH2 O LK : 1
logaSiO2
2 31 2 3
In the second expression (right-hand side of the first Al3 Ca2 HCO3 SiO2
LK
equal sign), we have taken the H+ and H2O column vec- 6 7 6 7
6 calcite 0 1 1 0 7 6 0:61 7
tors out of Sm, defining a pruned stoichiometric matrix, 6 7 6 7
6
6 laumontite 2 1 0 4 7 6
7 6 1:12 7
7
0
Sm . The third term in this expression drops out if the activ- 6 7 6 7
4 kaolinite 2 0 0 1 5 4 4:87 5
ity of water is unity. Multiplying through by the inverse of
0 quartz 0 0 0 1 2:44
Sm , gives: 2 3
0:01
0 0 1 0 1 6 8:63 7
Lab Sm SH L ar Sm LK ; 2 6 7
6 7 4a
where Lar is a vector consisting of the log activity ratios of 4 9:23 5
all the basis species. The dissolution reaction for carbon 2:44
2 31
2 3 Al3 Fe2 HCO
3 SiO2
logaAl3 a 3H 6 7
6 7 6 siderite 0 1 1 0 7
6 logaFe2 a2 7 6 7
6 H 76
daphnite 2 5 0 3 7
6 7 6 7
4 logaHCO3 aH 5 6 7
4 kaolinite 2 0 0 2 5
logaSiO2
quartz 0 0 0 1
2 3
LK 2 3
6 7 0:50
6 2:55 7 6
6 7 6 4:83 7 7
6 7
6 18:23 7 6 7 4b
6 7 4 7:37 5
4 3:61 5
2:30
2:30
2 31
2 3 Al3 Mg2 HCO3 SiO2
logaAl3 a3H 6 7
6 7 6 magnesite 0 1 1 0 7
6 logaMg2 a2 7 6 7 Fig. 2. The basin PT trajectory proposed by Smith & Ehrenberg (1989) in
6 H 76
daphnite 2 5 0 3 7
6 loga 7
aH 5 66 7 which P (bars) 6 (T(C) ) 25) compares well with more realistic PT tra-
4 HCO 7
3
4 kaolinite 2 0 0 2 5 jectories which are shown by dashed lines. In the realistic PT trajectories,
logaSiO2 hydrostatic conditions and a geothermal gradient of 20C km)1 (from a sur-
quartz 0 0 0 1
2 3 face temperature of 20C) pertain to 3 km depth, where a transition over
LK 2 3 0.5 km takes pressure to lithostatic levels (660 bars at 220 bars km)1). The
6 7 0:50
6 0:77 7 6 pressure gradient is again hydrostatic below the seal. The thermal gradient
6 7 6 6:85 7 7 in the half kilometer pressure transition varies from 20 to 100C km)1 as
6 7
6 28:34 7 6 7 4c
6 7 4 7:62 5 indicated. The realistic PT profiles are not significantly different than the
4 3:61 5 Smith and Ehrenberg PT trajectory. This shows the general appropriateness
2:30
2:30 of the Smith and Ehrenberg PT trajectory for PCO2 calculations.
At 200C and 16 bars log KCO2(g) equals )9.23. At ate (magnesite), and the fugacity of CO2 along the S&E
200C and 1050 bars it equals )9.2. Combining the bicar- trend is higher than along the hydrothermal trend where
bonate log activity ratios in the above expressions (third calcium aluminosilicates are present.
value in last column in 4ac) with these log KCO2(g) val- Figure 3 plots the ratio of the sum of the partial pres-
ues, as shown in eqn (3), gives log PCO2 )0.03, 1.83, sures of the gas phases (steam plus CO2) in equilibrium
and 1.58 for the calcite, siderite, and magnesite buffers, with the chemically buffered pore fluid, calculated as
respectively. These log partial pressures are plotted on described above, to the total fluid pressure along the PT
Fig. 1 at 200C. Partial pressures at other temperatures profile [P (bars) 6 (T (C) ) 25)] shown in Fig. 2.
(and pressures) are calculated in a similar fashion and plot- When this ratio equals unity, the gas pressure equals the
ted to produce the buffer curves plotted on that figure. total fluid pressure, and a separate gas phase can and will
Again, in the calcite buffer case, pressure is chosen so that form. Figure 3 shows that this will occur at temperatures
the water lies on the two-phase curve of water. In the sid- of approximately 330C where iron or magnesium carbon-
erate and magnesite buffer cases, the Smith and Ehrenberg ates are present. PCO2/Ptot 1 at 330C for the siderite,
pressuretemperature relationship shown in Fig. 2 is used and at 340C for magnesite buffer. Coudrain-Ribstein
[P (bars) 6 (T (C) ) 25)]. Figure 2 shows that this is a et al. (1998) show that nearly as high partial pressures of
reasonable PT profile for basins that become overpres- CO2 can be produced by a buffer that contains two car-
sured at depth. bonates (e.g. disordered dolomite and calcite), so the CO2
The curves calculated and plotted in Fig. 1 match the source region need not necessarily contain magnesite or
data shown in that figure quite well. The hydrothermal siderite. The essential conclusion at this point is that a sep-
data are matched well by the hydrothermal buffer (4a), arate CO2-rich gas phase can be produced in sedimentary
and the Smith and Ehrenberg non-anomalous basin CO2 basins if the sediments reach temperatures of approximately
(S&E) data are matched will by the siderite and magnesite 330C.
buffers (4b,c). The reason that the S&E CO2 fugacities are
higher than the hydrothermal CO2 fugacities is that Ca-sil-
icate minerals are absent in the S&E case. Fe or Mg silicate
CO 2 MIGRATION AND TITRATION
minerals are present instead. Because silicate calcium is not Because of its buoyancy, a CO2-rich gas phase that is pro-
available, CO2 cannot be precipitated as calcium carbonate, duced in a basin at approximately 330C will have a strong
but only as iron carbonate (siderite) or magnesium carbon- tendency to migrate upward. If this occurs the migrating
A
Not titrated
Titrated of silciate Ca
C
but not silcate Fe, Mg
>330C
Source
estimated for the particular basin being modeled. In the Fig. 7. Simulation of CO2 titration in a basin with two permeable vertical
absence of specific data, the values for a typical shale that channels (white columns) and four permeable strata (arrows and p anno-
are given in Table 1 can be used. We use these values in tation on 1.77-Ma section). Sediments are deposited and the section grows
in thickness with time. Age is indicated at the top of each section, and
the calculation presented in this paper.
strata are indicated by black horizontal lines. The arrows indicate water
From Table 1 it can be seen that the total titratable flow. Their upward increase in size indicates the dominance of compaction
silicate cation content of an average shale is 8.8 in driving pore fluid flow. CO2 is generated where the growing section
10)4 moles cm)3 of sediment (the total reduced oxide cat- moves across the 320C isotherm (roughly the elevation of the top of the
ion content minus the moles tied up in carbonates), and colored (mottled in black and white) band in the 1.77 Ma section). CO2
generated in this source zone is focused by the permeable strata (labeled
the CO2-generating capacity of a typical shale is
p) into the vertical channels, and the channels quickly become titrated of
1.5 10)4 moles CO2 cm)3 of shale. We assume that all their CO2-buffering capacity, as indicated by their white color. Colors show
the CO2 can be produced from the sediment, but that where the CO2 titration capacity (initially green and ending in white) has
about half the overlying sediments would be bypassed and been reduced by reaction with CO2. Reservoirs at the tops of the vertical
thus the overlying sediments will be only about half channels at the present day (t 0 Ma section) would have a high probabil-
ity being filled with gas with a high mole fraction of CO2.
titrated of CO2-neutralizing silicate cations. The CO2 titra-
tion capacity of typical shale is therefore taken to be
4.4 10)4 moles CO2 cm)3 of sediment.
The output of our finite element basin model simula-
tions is (i) the location of the CO2 source regions in a
basin and, (ii) the location of sediments that have been
titrated of their silicate Ca, Fe and Mg content by interac-
tion with CO2. These titrated zones, and their reasonable
geologic extensions, are the areas where the risk is highest
of encountering reservoirs with high mole fractions of
CO2.
critical parameters can provide an estimate of the CO2 risk Boon JA, Hitcheon B (1983) Application of fluid-rock reactions
at the location of a particular well. The critical parameters studies to in situ recovery from oil sand deposits, Alberta,
Canada-II. Mineral transformations during experimental-
include the ratio of CO2 generation to total potential titra-
statistical study of water-bitumen-shale reactions. Geochimica
tion at the basin center and at the margin nearest the well, Cosmochimica Acta, 47, 24957.
and the proximity of structures to the well. The numerical Cathles LM (1986) The geologic solubility of gold from 200
modeling described above thus provides a basis for much 350C, and its implications for gold-base metal ratios in vein
simpler methods of data evaluation and risk management. and stratiform deposits. Canadian Institute of Mining and
Metallurgy, 38 (Special Volume), 187210.
The model presented also has interesting and perhaps
Cathles LM (1993) Oxygen isotope alteration in the Noranda
important scientific implications. The migration routes of Mining District, Abitibi greenstone belt, Quebec, Canada.
natural gases in basins are of both practical and scientific Economic Geology, 88, 1483512.
interest. Practically, knowing the migration pathways might Cathles LM (2003) Processing Geologic Cross Sections to Estimate
usefully direct exploration to areas where traps would be Hydrocarbon Production, Migration and Chemistry, Users manual
for BasinLAB software, http://www.geo.cornell.edu/geology/
less likely to be filled with CO2. Scientifically, we would
easresearch/geomodeling/basinlab/AgeohistFinalReport4.pdf,
like to know how fluids of various kinds move through 167 p.
faulted sediments. This has proven remarkably difficult, Cathles LM, Losh SL (2002) A modeling analysis of the hydrocar-
and there is broad disagreement about even such funda- bon chemistry and gas washing, hydrocarbon fluxes, and reser-
mental issues as whether hydrocarbons move in faults at voir filling, vol. IV. In: Seal Control of Hydrocarbon Migration
and its Physical and Chemical Consequences (ed. Cathles LM ),
all, or only in the fractured zones adjacent to faults. Exam-
63 p. Gas Research Institute, Des Plaines, Report GRI-03/
ining the alteration pattern produced by the passage of a 0065.
reactive gas such as CO2 could provide a way to answer Cathles LM, Schoell M, Simon R (1990) A kinetic model of CO2
some of these questions, particularly since the source of generation and mineral and isotopic alteration during steamfloo-
CO2 can be highly localized, as it is when the CO2 ema- ding. SPE Reservoir Engineering, November, 52430.
Coudrain-Ribstein A, Gouze P, de Marsily G (1998) Tempera-
nates from or is generated by the heat from an intrusion.
ture-carbon dioxide partial pressure trends in confined aquifers.
Chemical Geology, 145, 7389.
Elder J (1981) Geothermal Systems. Academic Press, London, 508 p.
ACKNOWLEDGEMENTS
Ellis AJ (1970) Quantitative interpretation of chemical characteris-
The work on which this paper is based was carried out by tics of hydrothermal systems. Geothermics, 2, 51628.
Giggenbach WF (1981) Geothermal mineral equilibria. Geochimica
the authors while the first author was at Cornell and the
Cosmochimica Acta, 45, 393410.
second author at Chevron. A 1996 report to Chevron Giggenbach WF (1997) Relative importance of thermodynamic
prepared by the first author forms the basis for much of and kinetic processes in governing the chemical and isotopic
the text. We are thankful to Chevron for past support, composition of carbon gases in high-heatflow sedimentary
and for permission to publish the results of the work they basins. Geochimica Cosmochimica Acta, 61, 376385.
Hanor JS (1994) Physical and chemical controls on the composi-
funded. We would like to acknowledge the assistance of
tion of waters in sedimentary basins. Marine and Petroleum
members of the CO2 Chevron study team, including espe- Geology, 11, 3145.
cially Alex Erendi, Steve Patti, and Noelle Schoellkopf, Hanor JS (2001) Reactive transport involving rock-buffered fluids
and the help of Barry Katz in securing the proper permis- of varying salinity. Geochimica Cosmochimica Acta, 65, 3721
sions to publish. We are also grateful for the helpful sug- 32.
Hebner BA, Bird GW, Lngstaff FJ (1986) Fluid-pore mineral
gestions of two reviewers, one anonymous. We thank
transformations during steam injection: implications for reduced
particularly Kurt Bucher and Geofluids editor Bruce Yardly permeability damage. Journal of Canadian Petroleum Technol-
for encouraging a more detailed presentation of the mod- ogy, SeptOct, 6873.
eling methods. Henley RW, Truesdell AH, Barton PB (1984) Fluid-mineral equi-
libria in hydrothermal systems. Reviews in Economic Geology, 1,
Society of Economic Geologists, El Paso, 267 p.
REFERENCES Huang B, Xiao A, Hu Z, Yi P (2005) Geochemistry and episodic
accumulation of natural gases from the Ledong gas field in the
Arnorsson S, Gunnlaugsson E, Svavarsson H (1983) The chemis- Yinggehai Basin, offshore South China Sea. Organic Geochemis-
try of geothermal waters in Iceland. II. Mineral equilibria and try, 36, 1689702.
independent variables controlling water compositions. Geochimi- Hutcheon I, Shevalier M, Abercrombie HJ (1993) pH buffering
ca Cosmochimica Acta, 47, 54766. by metastable mineral-fuid equilibria and evolution of carbon
Ballentine C, Schoell M, Coleman D, Cains BA (2001) 300-Myr- dioxide fugacity during burial diagenesis. Geochimica Cosmochi-
old magmatic CO2 in natural gas reservoirs of the west Texas mica Acta, 57, 101727.
Permian basin. Nature, 409, 32731. Johnson JW, Oelkers EH, Helgeson HC (1992) SUPCRT92: A
Baylis P, Levinson AA (1971) Low temperature hydrothermal syn- software package for calculating standard molar thermodynamic
thesis for dolomite or calcite, quartz and kaolinite. Clay and properties of mineras, gases, aqueous species, and reactions from
Clay Minerals, 19, 10914. 1 to 5000 bars and 0 to 1000C. Computers in Geoscience, 18,
Behar E, Simonet R (1985) A new non-cubic equation of state. 899947.
Fluid Phase Equilibria, 21, 23755.
Kubacki W, Boon J, Bird G, Wiwchar B (1984) Effect of mineral Seewald JS, Benitez-Nelson BC, Whelan JK (1998) Laboratory
transformation on porosity and permeability of dolomite rock and theoretical constraints on the generation and composi-
during in situ recovery of bitumen: a preliminary study. Bulletin tion of natural gas. Geochimica Cosmochimica Acta, 62,
of Canadian Petroleum Geology, 32, 2818. 1599617.
Leong KM (Ed.) (2000) The Petroleum Geology and Resources of Seewald JS (2003) Organicinorganic interactions in petroleum-
Malaysia, Petronas (Petroliam Nasional Berhad), Kuala Lumpur, producing sedimentary basins. Nature, 426, 32733.
665 p. ISBN 983-9738-10-0. Smith JT, Ehrenberg SN (1989) Correlation of carbon dioxide
Levinson AA, Vian RW (1966) The hydrothermal synthesis of abundance with temperature in clastic hydrothermal reservoirs:
montmorillonite group minerals from kaolinite, quartz, and vari- relationship to inorganic chemical equilibrium. Marine and
ous carbonates. American Mineralogist, 31, 4958. Petroleum Geology, 6, 12935.
Muffler PJ, White DE (1968) Origin of CO2 in the Satton Sea Wilson GA (1994) Mineral-fuid equilibria in deeply buried sand-
Geothermal System, Southeastern California, USA. XXIII Inter- stones at McAllen Ranch Field, S. Texas. Geological Society of
national Geological Congress, 17, 18594. America Abstracts with Programs, A-279.
Palandri JL, Reed MH (2001) Reconstruction of in situ composi- Wycherley H, Fleet A, Shaw H (1999) Some observations on
tion of sedimentary formation waters. Geochimica Cosmochimica the origins of large volumes of carbon dioxide accumulations
Acta, 65, 174167. in sedimentary basins. Marine and Petroleum Geology, 16,
Potter PE, Maynard JB, Pryor WA (1980) Sedimentology of Shale. 48994.
Springer-Verlag, New York, 306 p.