A Methodology to Assess Increased Storage Capacity Provided by Fracture Networks at CO2 Storage Sites: Application to in Salah Storage Site Academic research paper on "Earth and related environmental sciences"

Available online at www.sciencedirect.com

SciVerse ScienceDirect

Energy Procedia 37 (2013) 3711 - 3718

GHGT-11

A methodology to assess increased storage capacity provided by fracture networks at CO2 storage sites: Application to In

Salah storage site

James Smith, Sevket Durucan*, Anna Korre, Ji-Quan Shi

Department of Earth Science and Engineering, Royal School of Mines, Imperial College London, London SW7 2BP, UK

Abstract

The presence of fractures in the storage reservoir at CO2 storage sites may increase the reservoir permeability and subsequently cause the CO2 plume extent to increase. Similarly, fractures in the caprock could provide regions of secondary storage if CO2 escapes from the reservoir. An important factor influencing the degree of these effects is whether the fractures form a continuously connected, or percolating, pathway. A methodology assessing the existence of percolating network of fractures, which incorporates the uncertainties in measured fracture properties around wells, was applied to assess secondary storage in the lower caprock at the In Salah Storage Site. It is demonstrated that secondary storage will occur if the fracture line density is equal to or greater than 2 m-1 and further shown what length distributions will provide secondary storage, if line density is less than 2 m-1.

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

Keywords: CO2 storage; fracture networks; caprock fracture connectivity; secondary storage; In Salah

1. Introduction

Movement of geologically stored CO2, within either the primary storage reservoir or within the storage complex, is primarily controlled by the (absolute) permeability of the relevant layers. Flow in the storage reservoir may be due to a combination of matrix and fracture permeability. Given that matrix permeability in the storage reservoir is likely to be significant, individual non-connected fractures may act as flow paths, and greater overall permeability is correlated with a greater degree of fracture connectivity.

* Corresponding author. Tel.: +44-20-7594-7354; fax: +44-20-7594-7444. E-mail address: s.durucan@Imperial.ac.uk

1876-6102 © 2013 The Authors. Published by Elsevier Ltd. Selection and/or peer-review under responsibility of GHGT doi: 10.1016/j.egypro.2013.06.265

In the reservoir, it is desirable to have a large permeability as this allows injected CO2 to spread quickly through the region, thus minimising pressure increase around the injection well.

Secondary storage may also exist in the overburden formations within the storage complex. As the caprock usually has a low matrix permeability, the upward migration of CO2 from the injection horizon to the secondary storage zone would be mainly due to the presence of a network of interconnected fractures if the caprock seal is affected by natural or induced fractures. A fracture network that provides a continuous pathway is from here onwards referred to as percolating. The connectivity of fracture networks is dependent on the geometry of the fractures and can be assessed using the percolation theory.

Building upon research reported previously [1, 2], a methodology was developed to assess whether a percolating fracture network in the caprock exists. A brief description of the methodology developed is presented in Section 2. Its application to the lower caprock of the storage complex at the In Salah storage site is described in Section 3. The assessment of permeability and relative permeability of and multiphase flow through a percolating fracture network has not been considered at this stage of research.

2. Percolation assessment methodology

Percolation theory has been widely applied in petroleum reservoir engineering and in fracture modelling in general. Application to CO2 storage could be quite critical in determining if rock units are a seal or not. Measurements of fractures in the caprock originate from well and seismic data sources. These measurements provide some information on the value of parameters that collectively define the fractures (length distribution, orientation distribution, aperture distribution, spatial distribution etc). There is uncertainty over both these parameter values and the particular way that fractures are arranged in the caprock. The percolation assessment methodology presented here considers both these uncertainties and takes as an input the available measurements and provides information on the uncertainty related to the existence of a percolating network. The methodology can be divided into three main stages.

Fig. 1. Partially cemented fracture in core from well KB-2 at In Salah [4].

The first stage of percolation assessment is to assimilate the available direct fracture data from well measurements such as image logs or core samples. An example from In Salah is shown in Fig. 1. The values of some fracture parameters may be obtained from these well measurements, while the values of some other fracture parameters will be only partially described by measurements. These uncertainties can be accounted for by defining probability distributions for the value of each parameter, which respect the measurement uncertainty. Combining the values from the distributions of different fracture parameters using Monte Carlo sampling, a large array of possible values of all the parameters that describe the fractures can be produced. This is an array where each member contains a value for density, orientation

distribution, length distribution etc. and the parameter values in the array are representative of the measurement uncertainty. In addition to reasonably well known and less known parameters, the values of some fracture parameters will be completely unknown. When a parameter is completely unknown, there is insufficient data for a probabilistic treatment of the uncertainty relating to such parameters. For this purpose, assumptions are made to define distinct scenarios which describe the major parameter uncertainties. The outcome of this stage of the methodology is an array of fracture parameter values, which together describe the relevant uncertainty given some basic assumptions defined by scenarios.

In the second stage of percolation assessment, the array of fracture parameter values derived previously is used as the inputs to the in-house fracture network model developed at Imperial College. These data are considered representative in describing fracture properties across the whole region in the quantitative stage of fracture modelling. Modelling is performed using a Marked Point Process [3] to assign fractures to a region and percolation is tested for between opposing surfaces of the model as shown in Fig. 2. By using a large number of different fracture parameters and modelling a large number of fracture network realisations for each set of fracture parameters, the probability of percolating pathways existing is obtained given the scenario assumptions and uncertain data.

As fracture properties may vary dramatically away from the wellbore, the final stage of percolation assessment considers that direct fracture data gained from wellbores may not necessarily be representative over the whole model area. The fracture modelling will contain unaccounted for uncertainties, so the results of quantitative modelling are combined with qualitative data and the assumptions on which scenarios are based to make a fuller assessment of uncertainty in the occurrence of percolation.

3. Application of methodology to In Salah storage site

3.1. Background

The main CO2 storage aquifer (C10.2) at Krechba is approximately 20-25 metres thick at about 1,880 m below the surface. It is overlain by a tight sandstone and siltstone formation (C10.3) of about 20 m in thickness, which is in turn overlain by a 950 m thick formation of Carboniferous Viséan mudstone interbedded with thin dolomite and siltstone layers. The C10 formation, together with the lower cap rock (C20.1 - C20.3), form the CO2 storage complex at Krechba (Fig. 3).

The In Salah project has injected CO2 into the aquifer leg of the reservoir through three horizontal injection wells (KB-501, KB-502 and KB-503) since August 2004. As the injection site is also a producing gas field, the storage formation is relatively well characterised. It is known that fractures exist in the region and these have been observed at KB-502 and a production well, Kb-14 [4]. The fracture data estimated by Iding and Ringrose are shown in Fig. 3(a), based upon which a schematic diagram of fracturing in the region is drawn in Fig. 3(b). These data indicate similar fracturing to be present in the reservoir and lower caprock section with uncharacterised fracturing in the upper caprock. Additionally, seismic data indicate the presence of a NW fault zone in between KB-502 and a suspended (now fully decommissioned) appraisal well, KB-5 [5]. The observed CO2 breakthrough at KB-5 suggests that this fault zone and/or fractures are contributing to an enhanced pathway of migrating CO2 between KB-502 and KB-5.

Fig. 3. (a) Fracture orientation, density, aperture and length data acquired from KB-502 and KB-14 [4]; (b) schematic diagram of fractures in the reservoir and caprock (not to scale).

Further analysis of surface uplift data and history matching of the bottomhole pressures at KB-502 in recent research [6, 7] has also suggested that injected CO2 is migrating through this non-sealing fault (zone) identified in the main storage reservoir and, moreover, CO2 may have migrated into the lower caprock. A potential source of uncertainty, however, is whether there is secondary storage in the lower caprock (C20), in addition to the reservoir (C10). If there is secondary storage, then CO2 that flows up the fault zone can dissipate into the lower caprock. However, without secondary storage, pressure will build up in the fault zone with the potential for further vertical propagation. Within the lower caprock, where matrix permeability is very low, any secondary storage is likely to be a result of fracture permeability alone. Therefore, an assessment of whether horizontal percolation occurs and whether capillary entry pressures are reached would be very valuable.

The production of trapped hydrocarbons at In Salah is indicative of a sealing caprock; therefore, it is believed that fractures do not form a significant leakage pathway out of the storage complex. This suggests the absence of a vertically percolating pathway through a significant portion of the caprock. However, it is not known whether fracture networks within the storage reservoir are vertically percolating.

3.2. Fracture characteristics

Fracture orientation data [4] suggest a dominant fracture orientation of NW-SE about which there is approximately 30° of uniform variation. The reliability of these data is high as they were acquired from multiple sources. Fracture density data (Fig. 3a) can be considered reliable around the well that it was acquired from. Fracture length data at In Salah, shown in Fig. 3a, were acquired from mud loss events, which are subject to large uncertainties. The length values reported in Fig. 3a should be considered maximum values of fracture length. Image log data have not been used to estimate intersection ratio. The data on fracture apertures, shown in Fig. 3a, were acquired using image log and mud loss data, although there are potentially large uncertainties in these measurements [4]. Additionally, core analysis data show that some fractures were cemented and therefore not conductive, although the number of closed fractures was not reported. Image log data from KB-14 show stratigraphic zonation of fractures [4], which suggests that fractures are stratabound to some extent. The layer truncation percentage (Ltrunc) of a stratabound set is not known.

Seismic surveys show the existence of a number of faults in the region [5], as illustrated in Fig. 4. In particular, there is a fault very close to well KB-502 where the data in Fig. 3a are reported from.

3.3. Fracture modelling

To investigate the potential existence of a secondary storage volume in the lower caprock, a single stratabound fracture layer with a horizontal model domain of 800 m x 800 m x 20 m was considered to represent a section of the lower caprock. A uniform strike distribution of NW-SE ± 15° and a dip value of 86° were used for each realisation. The measured fracture line density was given as 1 - 5 m-1, implying a variation in density along the well path. It was not clear whether the variation was inter- or intra-layer or both. Without this information, the fractures were modelled assuming a random spatial distribution and with different density scenarios defined by a single fracture line density value (1, 2, 3, 4 and 5 m-1). In the absence of data on the proportion of cemented fractures, all fractures were assumed to be open. A major uncertainty was the length distribution, because without intersection data, the length distribution was unknown. Therefore, this was the major parameter investigated with modelling and a power-law length

Fig. 4. Fault locations in the In Salah reservoir region [8].

distribution (n(R)=0R'") was assumed [9]. Here R is the length, n(R) is the number density of fractures of length R, a is a constant describing the power law and /? is the normalisation constant, which describes the fracture density. Rmax was constrained by mud-loss data to between 50 and 200 m, Rmin was unknown and a was assumed to vary between 1.8 and 4.5. Therefore, fracture properties were selected from the ranges shown in Table 1 to define a power-law length distribution for each realisation. An example of a single fracture network realisation, including a magnified section is shown in Fig. 5.

Table 1. Modelled fracture properties.

Orientation Strike uniformly distributed in the range NW-SE ± 15° and dip=86°

Length Power-law: Rmax=50-200m, Rmin=0.1-50m, a =1.8-4.5 (single values

selected for each realisation)

Density 1-5 m-1

Fig. 5. A single fracture network realisation including a zoomed in section.

It was tested whether a percolating network existed, which spanned all sides of the model region. If this condition was achieved, then the modelled realisation was considered to be well connected and therefore provided secondary storage. Considering first realisations with a fracture line density of 1 m-1, percolation was found to occur for all realisations with mean lengths greater than 10m. For realisations with smaller mean lengths, percolation was found not to occur for the ranges of a and Rmin when a <3 and Rmin <2, or a <4 and Rmin <5, or a <4.5 and Rmin <7 respectively. Even for these non-percolating realisations, there still existed a cluster which spanned a distance in the range 200 - 800 m in the direction of the dominant fracture orientation. However, the cluster contained only a small fraction of the fractures in the modelled region and spanned only a small distance in the direction perpendicular to the dominant fracture orientation. Therefore, those realisations would not provide secondary storage. For realisations with a fracture line density of 2 m-1, percolation was found to occur for all realisations. This was also the

case for line densities up to 5 m-1. Fig. 6 displays fracture network realisations with different levels of percolation, ranging from a very small cluster to almost complete connectivity.

Even though the minimum length and power-law exponent remained undefined, it was possible to make some inferences about the existence of secondary storage in the lower caprock. If the measured fracture line density range (1 - 5 m-1) described density variation between layers, then most, if not all layers, were expected to contain a horizontally percolating network and to contribute to secondary storage. However, this was with the qualification that, if mean length was low, there may have been some regions not contributing to the percolating cluster where density was low. It may have been possible to confine fracture length by using image data to find intersections. This would have better characterised the connectivity in the fracture network

Fig. 6. Fracture network realisations with different degrees of connectivity. The red fractures indicate the largest cluster within the realisation.

The results described so far have assumed fracture networks where all fractures were open. However, core data show that some fractures were closed. This would act to reduce connectivity. As shown by the fault map in Fig. 4, the region is heavily faulted, which could cause high spatial variability in fracturing. However, fracture data, which have been derived from both the injection well KB -502 and the production well KB-14, show very similar fracture properties. While these provide only two data sets, and as such cannot be used to make firm conclusions, it would suggest that variability in fracture properties may be

low. Furthermore, the fault map shows that KB-502 is reasonably close to a fault and thus, if there was significant variability, the region around KB-502 may represent higher than average fault densities. These two pieces of evidence would suggest that the fracture connectivity analysis performed here is unlikely to underestimate fracture connectivity and the potential for secondary storage. However, without access to fracture data from other wells in the region or other spatially varying data, it is not possible to rigorously assess this conclusion.

4. Conclusions

Percolation assessment of the In Salah storage site, based upon the published fracture data, carried out in this study suggests that secondary storage could exist in the lower caprock if the fracture line density is equal to or greater than 2 m-1. The length distributions required for secondary storage when fracture line density falls below 2 m-1have also been determined. Since line density in the lower caprock at Krechba is reported to range between 1 and 3 m-1, there is likely to be at least some degree of secondary storage within the lower caprock. However, this result is subject to considerable uncertainty regarding the degree of fracture cementation, density variation away from wells and the uncharacterised fracture set

Previous modelling work by the authors [6] indicated that upwards migration of CO2 from the main storage unit (C10) into the lower caprock has most likely occurred around KB-502. The existence of a secondary storage in the lower caprock thus provides a desirable buffer zone for the migrating CO2.

Acknowledgements

Part of the research reported in this paper was carried out within the framework of European Commission funded project CO2ReMoVe (SES6-2005-CT-518350). The authors wish to thank the EU, the In Salah Gas Joint Industry Project (BP, Statoil, Sonatrach) and CO2ReMoVe project partners for their support and contributions towards their research.

References

[1] Smith J, Durucan S, Korre A, Shi J-Q and Sinayuc C. Assessment of fracture connectivity and potential for CO2 migration through the reservoir and lower caprock at the In Salah storage site. Energy Procedia 2010; 4:5299-5305.

[2] Smith J, Durucan S, Korre A and Shi J-Q. Carbon dioxide storage risk assessment: Analysis of caprock fracture network connectivity. Int. J. Greenhouse Gas Control 2011;5:226 - 240.

[3] Huseby O, Thovert JF and Adler PM. Geometry and topology of fracture systems. Journal of Physics A-Mathematical and General 1997; 30(5):1415-1444.

[4] Iding M and Ringrose P.. Evaluating the impact of fractures on the long-term performance of the In Salah CO2 storage site. Int. J. Greenhouse Gas Control 2010; 4:242-248.

[5] Ringrose P, Atbi M, Manson D, Espinassous M, Myhrer O, Iding M, Mathieson A, Wright I. Plume development around well KB-502 at the In Salah CO2 storage site. First break 2009; 27:49-53.

[6] Shi J-Q, Sinayuc C, Durucan S, Korre A. Assessment of carbon dioxide plume behaviour within the storage reservoir and the lower caprock around the KB-502 injection well at In Salah, Int. J. Greenhouse Gas Control 2012;7:115-26.

[7] Vasco DW, Rucci A, Ferretti A, Novali F, Bissell RC, Ringrose PS, Mathieson AS, Wright IW. Satellite-based measurements of surface deformation reveal fluid flow associated with the geological storage of carbon dioxide, Geophys. Res. Lett. 2010; 37: L03303, doi: 10.1029/ 2009GL041522.

[8] Ringrose PS, Roberts DM, Gibson-Poole CM, Bond C, Wightman R, Taylor M, Raikes S, Iding M and 0stmo S. Characterisation of the Krechba CO2 storage site: Critical elements controlling injection performance 2011. Energy Procedia; 4:4672-4679.

[9] Bonnet E, Bour O, Odling NE, Davy P, Main I, Cowie P and Berkowitz B. Scaling of fracture systems in geological media.

Reviews of Geophysics 2001; 39:347-383.