Available online at www.sciencedirect.com

SciVerse ScienceDirect

Energy Procedia 37 (2013) 4420 - 4427

GHGT-11

Development and field application of model for leakage of

CO2 along a fault

Qing Taoa*, David Alexanderb, and Steven L. Bryanta

aCenter for Petroleum and Geosystems Engineering, The University of Texas at Austin, Austin, TX 78712, USA _bThe University of Trinidad and Tobago, Trinidad_

Abstract

One of the key issues when considering geologic storage of CO2 is the risk of leakage from the storage area. Faults and the damage zones surrounding them may provide such a conduit. Quantifying CO2 flux along a fault requires methods to estimate fault properties. We describe a simple leakage model that estimates the worst-case CO2 flux along a fault in the Mahogany Field, Trinidad using geologic data from the fault. Considering the boundary condition of the permeable layers that intersect the fault and the ratio of vertical over horizontal fault zone permeability, we estimate the attenuation rate, i.e. the rate at which CO2 could flow from the fault into these layers. Results show that with layers of sufficient permeability the CO2 flux at the uppermost end of the fault could be attenuated to zero. However the attenuation is temporary if layers are sealed at other end. The flow rate at top of fault increases asymptotically after the attenuation starts decreasing.

© 2013 The Authors. Published by Elsevier Ltd. Selection and/or peer-review under responsibility of GHGT

Keywords: CO2 leakage; fault permeability; attenuation; shale gouge ratio; permeability multiplier; pressure elevation

1. Introduction

Captured CO2 from coal-fired power plants can be stored in depleted oil and gas reservoirs or saline aquifers that are not currently earmarked for other uses. However under typical storage conditions the CO2 is buoyant relative to brine, and thus the injected CO2 could leak out of these reservoirs along conductive faults and fractures present in the earth. In oil and gas provinces, one view of faults is that they were the main conduits from a hydrocarbon source, with reservoirs forming as hydrocarbon migrates from the fault

* Corresponding author. Tel.: +1-512-471-3250; fax: +1-512-471-9605. E-mail address: qing.tao@utexas.edu.

1876-6102 © 2013 The Authors. Published by Elsevier Ltd. Selection and/or peer-review under responsibility of GHGT doi: 10. 1016/j .egypro .2013.06.347

into sands intersected by the fault. Another model of reservoir charge invokes general area-wide migration of hydrocarbon upward into the lowermost sands, independently of whether faults are conductive. Indeed faults are commonly invoked as seals, with hydrocarbon collecting within a sand until exceeding the entry pressure at some location along the faults. Given this range of potential behavior, to quantify the CO2 leakage rate along a fault it is important to accurately estimate the conductivity of the fault. Knowledge of the petrophysical properties of fault rocks improves flow predictions within structurally complex petroleum reservoirs [1][2]. Fault-seal analysis has been widely used to estimate the permeability of fault rocks based on the analysis of small-displacement faults collected from core samples or well log data [3][4].

The properties of the Mahogany field have been well characterized. This study uses the geology of this hydrocarbon-bearing structure to represent the geology of analogous structures that do not contain hydrocarbon, i.e. deep saline aquifers. Such structures are attractive targets for sequestration since both infrastructure and knowledge of local geology are usually already available. For brevity, we refer to "injecting C02 in the Mahogany field." In reality this would correspond to a combined EOR/sequestration project. For the purpose of this study it simply means "injecting C02 into a brine-filled structure analogous to the Mahogany field."

The study will focus upon the possible leakage of CO2 along and across faults since the migration pathways are of major concern. It will also include leakage of CO2 from the migration pathway into permeable formations intersected by the pathway. Field data are used to constrain the pathway permeability and thus to calculate the mass flux of CO2. The insight gained here is that although the CO2 may leak from the reservoir, it may not reach to the surface. This information using actual field data helps to further understand the risk associated with CO2 storage in geologic reservoirs. In this study we will compute the flux as well as the attenuation, for the arrangement of sand and shale in fault blocks from the Mahogany field.

2. Model Development

Following extensive field work on evaluating fault seals for hydrocarbon reservoirs, we estimate the fault permeability and capillary entry pressure from a shale gouge ratio (SGR) correlation [5]. The TrapTester was used to build a 1-D model to quantify the SGR using well log data and well trajectories for different wells using Mahogany field data. The model will present different fault zone permeabilities (FZP) on estimated shale gouge at different positions along the fault. Shale gouge is a first-pass estimate for whether a fault will be sealing. Here we can estimate variation in permeability along the fault by correlating permeability with entry pressure, i.e. large SGR causes large entry pressure and small permeability. We use the SGR correlation of Manzocchi et al. [4] given by

log kf= -4SGR-1 logCD)(1 - SGR)5 ^

where kf is the fault permeability across fault in units of md, D is the displacement or throw of fault in units of meters. With the TrapTester method, there is uncertainty in the model (e.g. how accurate and precise are the interpretations), how reliable is the estimate of Vshale in the well (e.g. is the Vshale derived from the gamma-log only or does it take into account different clay materials?). The amount of uncertainly in the interpretation increases as the size of the fault throw decreases [6].

The vertical (or along) fault permeability is considered the main factor for controlling CO2 flow along the fault. The correlation from SGR provides the permeability across fault. To estimate the CO2 flux along fault, we need an estimate of the vertical fault permeability. One way of relating permeability along and across fault is to apply a permeability multiplier m, given by

kv =kfoXm

where kv is the permeability along fault and kh is the permeability across fault. The permeability multiplier m is usually equal to or greater than 1, since the permeability along fault tends to exceed that across fault. On the other hand, we assume a close correlation between along and across fault permeability, and hence m does not go above 10 (the other limiting case).

Once the CO2 is in the fault, then flow would be along the 'path of least resistance' which is most likely to be either within the fault itself or the damage zone adjacent to the fault, depending upon the geometry and structure of the fault zone. One limiting case for the rate of CO2 flux is that the along-fault permeability is smaller than all other conductivities along the pathway. Another limiting case is that horizontal flow into the fault from the storage formation is slower than any other step along the pathway. We assume that fault seal analysis only applies to the entry of CO2 into the fault. Hence once the seal is breached, then the analysis will require the results from fluid flow simulation and permeability estimation to get a better understanding.

We incorporate a single phase flow model along the fault to establish a worst-case (upper bound) estimate of CO2 flux [7][8], as illustrated in Fig. 1. We consider a single phase steady flow of CO2 along the fault, and we assume that CO2 has established steady saturation along the fault. The fault is treated as a one-dimensional conduit with spatially varying permeability. The leakage rate along fault can be estimated once fault zone permeability is obtained.

Fig. 1. CO2 plume leaks from a storage formation along a fault, driven by buoyancy and possibly by pressure elevation due to injection. The rate of CO2 leakage at the top of the fault, qtop, can be attenuated by flow of CO2 from the fault into layers intersecting the fault. The layer flow occurs only if the capillary pressure exceeds the capillary entry pressure of the permeable layer. The attenuation rates in permeable layers are functions of pressure gradient along the layers, relative permeability and compressibility. The boundary condition of permeable layers far from the fault is treated as two limiting

cases: constant pressure or no flow.

It is instructive to see how much CO2 flux at the top of the fault could be attenuated by permeable layers as it migrates towards the surface (Fig. 1). We use a semi-analytical 1D two-phase flow model to estimate the CO2 flow rate into the layers, treating the intersection of the layer with the fault as a constant pressure source of CO2 [8]. The two-phase model for permeable layers represents a process of CO2 displacing water. The far-field boundary conditions for permeable layer are considered in two limiting

Qtop- worst case flux

constant no flow

cases: constant pressure and no flow, which represent the best case and the worst case for attenuation of CO2 leakage in permeable layers, respectively.

3. Application on Mahogany Field, Trinidad

The Mahogany field is located offshore from the southeast coast of Trinidad in approximately 285 feet of water [9]. The field is a series of stacked gas sands in seven separate major fault blocks, situated in the Columbus Basin within the Eastern Venezuelan Basin. The field was discovered in 1968 but was not slated for development until the Atlantic LNG project was initiated in 1994 [10].

Our main focus is placed on fault blocks X (FBX) and Y (FBY) because this area contains the largest storage capacity relative to other fault blocks for CO2 storage (Fig. 2). The fault separating these two fault blocks is assumed to be the conduit through which CO2 could flow, if a plume reached the fault. Fig. 2 shows three major hydrocarbons bearing sands, i.e. sand 19, 20 and 21 which are considered target horizons. These target horizons can be categorized into two different groups, i.e. shallow high permeability (>700 md) horizons and deep lower permeability (<50 md) horizons [9]. The (FBY) was developed in the first phase because of the large gas cap and larger reserves based in the sand 21 [10]. Further, this field is dominated by water drive mechanism except for sand 20 [11]. These two pieces of data were used to develop the model for estimating fault permeability and leakage attenuation model designs as described earlier.

Fig. 2. Cross-section of the Mahogany field. Adapted from Maharaj et al [9].

We use the SGR correlation to estimate the cross-fault permeability. The SGR data are interpreted from well logging close to the fault by converting gamma ray to shale content logs. Together with well trajectories and fault displacement, the fault permeability is estimated from Eq. (1) and plotted against the depth in Fig. 3. The permeability is in the range of microdarcies to millidarcies with a mean of about 50

3.1. Fault Permeability

We use the SGR correlation to estimate the cross-fault permeability. The SGR data are interpreted from well logging close to the fault by converting gamma ray to shale content logs. Together with well trajectories and fault displacement, the fault permeability is estimated from Eq. (1) and plotted against the

depth in Fig. 4. The permeability is in the range of microdarcies to millidarcies with a mean of about 50

Fig. 3. The cross-fault permeability is generated from the SGR correlation (Eq. 1) using TrapTester, with a mean of 50 //d.

Another independent assessment of the fault zone permeability is to model the pressure response to production of adjacent fault blocks (Fig. 4). Historical pressure trends for FBX and FBY indicate pressure communication across the fault separating the blocks [12]. An analytical solution has been derived to relate the fault transmissivity to the pressure response in the compartments [5]. Uncertainty exists in the analytic model developed to estimate transmissivity (Tf) due to the uncertainty in the productivity index for FBY (PI2). We use a Monte Carlo approach to estimate a plausible range of Tf based on a reasonable range of PI2 (Fig. 5). The results show that the transmissivity for the model lies within 17.83 to 43.40 md-ft/psi-day with a mean value of 30 mdft/ psi-day. This corresponds to a range of FZP to be 0.03 to 0.07 md. The fault permeability estimated from the cross-flow model is thus consistent with the range obtained from the SGR correlation.

Fig. 4. Curves generated by the fault transmissibility model.

Name Cell Graph Min Mean Max 5% 95% Errors

Category: PI2

PI2 / MONTE CARLO SIMULATION 20 m * J \ 1 £0 b *

TO ESTIMATE C13 21.73159 35.99993 49.79071 30.07512 41.91845 0

FAULT PERMEABILI TY

Category: T f

T f / MONTE GARLO SIMULATION 15 /\ 45 w

TO ESTIMATE Cll 17.83184 30.00009 43.40521 25.06373 34.93281 0

FAULT PERMEABILI TY

Fig. 5. Monte Carlo Simulation showing the possible range of transmissivity for a given range of bottomhole flowing pressure that can allow the analytical model to match the field pressure profile for FBX and FBY.

3.2. CO2 Leakage Rate

Using the cross-fault permeability estimated from SGR correlation, we calculate the CO2 leakage flux under buoyancy-driven flow along the fault in Mahogany Field in the case that none of the flux is attenuated by intersecting permeable layers. The uncertainty in the permeability multiplier is considered in two limiting cases (m=1 and m=10). The predicted CO2 fluxes at top of the fault range from tenths to several mg/m2/s. For comparison Allis [13] reported that CO2 background flux at the earth's surface is up to 0.2 mg/m2/s. An example of a large surface flux would be Crystal Geyser (Utah, USA), where seepage from a thermogenic accumulation of CO2 reaches 8 mg/m2/s at localized high flux surface. The CO2 fluxes along the fault range from slightly above background to comparable to high fluxes in nature. The elevated pressure due to injection adds additional driving force for CO2 leakage. In the case of a 500 psi pressure elevation at the bottom of pathway, the CO2 flux is increased by about 20%. We then calculate the CO2 leakage rate corresponding to the cross-sectional area of the fault. In the worst case with 500 psi pressure elevation and largest along-fault permeability (m=10), the leakage rate is 5000 kg/d.

Next we compute the attenuation rate of each permeable layer. CO2 enters a layer only if the capillary entry pressure is exceeded. The boundary condition at the other end of the attenuation layers are considered in the two limiting cases described earlier: constant pressure and no flow. The summation of these rates is referred to as the total attenuation rate. We compare with the worst-case leakage rate from formation (5000 kg/d) to determine whether the leakage is fully attenuated by the layers (if so, the leakage rate from top of fault is essentially zero).

Multiple CO2 attenuation scenarios are presented by considering the pressure elevation at leakage depth and boundary condition of attenuation layers (Fig. 6). Cases (a) and (b) have a large pressure elevation of 500 psi at the base of the fault with constant pressure and no flow boundary respectively. In cases (c) and (d) a moderate pressure elevation of 200 psi above hydrostatic is applied on both cases.

The total attenuation rate is compared to the worst-case CO2 leakage rate of 5000 kg/d computed earlier in this section. Note that the leakage rate assumes a steady state flow and that the corresponding pressure profile along the fault does not change with time. For Case (a), the total attenuation rate is greater than the leakage rate from the formation, and thus the CO2 leakage rate at top of the fault is zero. This corresponds to a situation that all CO2 leaving the storage formation enters the encountered permeable layers. For Case (b), the CO2 leakage rate at top of fault is zero initially. However once the total attenuation rate starts to decline, more leaking CO2 migrates towards the surface and thus the

leakage rate at top of fault increases asymptotically to its worst-case rate. For Case (c), the total attenuation rate increases to its asymptotic value and the leakage rate at top of fault is zero after 1500 days. For Case (d), the total attenuation rate is less than the worst-case CO2 leakage rate, and thus the attenuation is not enough to appreciably reduce the impact if CO2 leaks at its worst-case rate.

Fig. 6. Total CO2 attenuation rate (cumulative for five attenuation layers intersecting the fault in Fig. 2): (a) pressure elevation 500 psi, constant pressure boundary (legend "CP"); (b) pressure elevation 500 psi, no flow boundary (legend "NF"); (c) pressure elevation 200 psi, constant pressure boundary; (d) pressure elevation 200 psi, no flow boundary. The total attenuation rate is compared with the worst-case CO2 leakage rate along the fault (dashed line). Thus attenuation is comparable to the worst-case flux in all cases except one, meaning that leakage from the top of the fault would be much less

than from the storage formation during the time period.

4. Conclusions

We estimate a plausible range of fault properties from field data using a shale gouge ratio correlation. The corresponding vertical fault permeability is estimated using a permeability multiplier. Using these values and other fault properties measured in the field, we estimate the worst-case leakage flux if a CO2 plume encounters the bottom of the fault. The predicted fluxes along the fault are between 0.1 and 10 mg/m2/s, which correspond to a range from slightly above background to comparable to high fluxes in nature. We use a simple two-phase model to estimate the flow of CO2 from the fault into permeable layers that intersect the fault. The results show that CO2 flux at the top of the fault can be attenuated to zero by the flows into permeable layer above the storage formation. In this case the overall security of the storage scheme would rely heavily on the secondary containment provided by other permeable layers. A sealing fault at other end of the layers yields a temporary attenuation, however, in which case the lack of leakage early in the storage project would give a false sense of security.

One major contribution of this work is that the models are tested using actual field data. This will provide a greater insight for the application of these models and techniques in similar type reservoirs.

Also a greater understanding of the possible rates of leakage will enable policy makers to perform better risk assessment when selecting storage sites for injecting CO2.

Acknowledgements

This work was supported by Center for Frontiers of Subsurface Energy Security (an Energy Frontier Research Center funded by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences under Award Number DE-SC0001114), by the Geologic CO2 Storage Joint Industry Project at UT-Austin, and by the Natural Gas Institute of the Americas at the University of Trinidad and Tobago.

References

[1] Jolley, S.J., Dijk, H., Lamens, J.H., Fisher, Q.J., Manzocchi, T., Eikmans, H., Huang, Y., 2007. Faulting and fault sealing in production simulation models: Brent province, northern North Sea. Petroleum Geoscience, 13, 321-340.

[2] Zijlstra, E.B., Reemst, P.H.M., Fisher, Q.J., 2007. Incorporation of fault properties into production simulation models of Permian reservoirs from the southern North Sea. In: Jolley, S.J., Barr, D., Walsh, J.J., Knipe, R.J. (Eds.), Structurally Complex Reservoirs. Geological Society, London, Special Publications, vol. 292, pp. 295-308.

[3] Hull, J., 1988. Thickness - displacement relationships for deformation zones. Journal of Structural Geology, 10, 431^35.

[4] Manzocchi, T., Walsh, J. J., Nell, P., and Yielding, G., 1999. Fault transmissibility multipliers for flow simulation models.

[5] Alexander, D., 2012. The Effect of Faults on the Plume Dynamics for Carbon Dioxide Sequestration in Trinidad and Tobago. Ph.D. Thesis. University of Trinidad and Tobago, Trinidad.

[6] Bretan, P., & Yielding, G., 2005. Using Buoyancy Pressure Profiles to Assess Uncertainty in Fault Seal Calibration, in Evaluating Fault and Cap Rock Seals: AAPG Hedberg Series, No. 2, 151-162.

[7] Tao, Q., Checkai, D.A., Huerta, N.J. and Bryant, S.L., 2010. Model to Predict CO2 Leakage Rates Along a Wellbore. SPE Annual Technical Conference and Exhibition, Florence, Italy, 20-22 September.

[8] Tao, Q., 2012. Modeling CO2 Leakage from Geological Storage Formation and Reducing the Associated Risk. Ph.D. Dissertation, The University of Texas at Austin.

[9] Maharaj, S., Jones, J. R., Mackow, H. M., Lachance, D. P., 2000. Design and Interpretation of Long-Term Tests in the Mahogany Field, Offshore Trinidad. SPE/CERI Gas Technology Symposium, Calgary, Alberta Canada, 3-5 April.

[10] Mackow, H.M., Bally, Y.K., Perry, J.T. and Lachance, D.P., 2003. Learning Along the Way: Mahogany Field in Trinidad After 5 Years, 700 Bscf, 22 MMbo and No Lost Time Incidents. SPE Annual Technical Conference and Exhibition, Denver, Colorado, USA, 5-8 October.

[11] Ali-Nandalal, J. and Gunter, G., 2003. Characterizing Reservoir Performance for the Mahogany 20 Gas Sand Based on Petrophysical and Rock Typing Methods. SPE Latin American and Caribbean Petroleum Engineering Conference, Port-of-Spain, Trinidad and Tobago, 27-30 April.

[12] Alexander, D and Byrant, S.L., 2011. Pressure Transient Effects of CO2 Sequestration in Faulted Reservoirs: Saline Aquifers. Energy Procedia, Volume 4, 2011, Pages 4331-4338.

[13] Allis, R., Bergfeld, D., Moore, J., McClure, K., Morgan, C., Chidsey, T., Heath, T. and McPherson, B., 2005. Implications of results from CO2 flux surveys over known CO2 systems for long-term monitoring. Proc. 4th Annual Conf. on Carbon Capture and Sequestration DOE/NETL. 2-5 May.