Capillary-driven Solute Transport and Precipitation in Porous Media during Dryout

Formation drying and salt precipitation due to gas injection or production can have serious consequences for upstream operations in terms of injectivity and productivity. Recently evidence has been found that the complexity of the pore space and the respective capillary driven solute transport plays a key role for the relationship of permeability and porosity. In this study, we investigate dryout and salt precipitation due to supercritical CO2 injection in single and multi-porosity systems under near well-bore conditions. We image fluid saturation states by means of CT scanning during desaturation. We are able to observe capillary driven transport of the brine phase and the respective transport of solutes on the pore scale. Finally we have access to the precipitated solid-salt phase and its distribution. With these results we verify conceptual models behind permeability porosity relationships – K() – for flow through drying and injectivity modeling. INTRODUCTION Drying processes in porous media are important in many industrial processes, in soil science and also for upstream operations. For gas injection and production in and from saline formations, formation drying will cause precipitation of salt initially dissolved in the brine. Precipitation affects the performance of injection and production wells and can even lead to well clogging, which is a serious risk for such operations. In this paper we consider large-scale geological sequestration of CO2, originating from anthropogenic sources like fossil-fueled power plants or contaminated gas production in order to reduce CO2 emissions. Deep saline aquifers and depleted oil and gas fields are potential subsurface deposits for that purpose (IPCC (2005); Bachu and Gunter (2004)). If dry (or under-saturated), supercritical (SC) CO2 is injected into water-bearing geological formations like saline aquifers, water is removed by viscous displacement of the aqueous phase and by partitioning of water in CO2 – to which we refer in the following as “evaporation” – and the subsequent advection in the injected CO2-rich phase. Both mechanisms act in parallel, but while the rate of advection decreases with SCA2014-Temp Paper A075 2/12 increasing CO2 saturation (diminished mobility), evaporation becomes increasingly important as the aqueous phase becomes immobile. Below residual water saturation, only evaporation takes place and the formation dries out. If water partitions in the CO2 phase, the salts originally present in the water are left behind. The volumes of precipitate depend on brine salinity and the transport of solutes and water in the reservoir. There is no easy way to predict whether, where and how salt precipitates and how precipitation affects injectivity. Due to the lack of well data we rely on numerical simulations and analytical models that are available in literature. In a series of publications, Pruess et al. presented numerical simulations of CO2 injection in saline aquifers investigating the fundamental aspects of formation dryout and salt precipitation (Pruess and Garcıa (2002); Pruess and Muller (2009)). The simulations were performed with a single injection well in idealized 1D and 2D radial geometries. The authors observed that dryout and precipitation occurs only in a narrow zone confined to a few meters around the injection well. The solid salt saturation in this zone has been found to be constant independent of the injection rate. 2D radial simulations were carried out to explore gravity effects. In contrast to the 1D-radial scenario, gravity in combination with capillary-driven flow lead to more heterogeneous precipitation with a maximum observed solid-salt saturation of more than 20% (0.2). Giorgis et al. (Giorgis et al. (2007)) performed field-scale simulation in radial geometry. The authors found that the amount of precipitate depends on brine mobility and can be high if there is a capillary-gradient driven brine flow in direction of the well bore. The authors have further shown that the injection rate is an important factor in controlling precipitation process and in avoiding or allowing complete clogging of the formation. In their simulations, solid salt saturations of more than 60% (0.6) have been reached. However, field scale simulations require experimental input parameter on flow physics. This is crucial for the success of numerical simulations as several studies suggest that a modest change in porosity might lead to a serious reduction in permeability. The literature reports on several mechanisms that lead to porosity variation which range from variations within a lithology/rock type (e.g. Pape et al. (1999)), over mechanical compaction (e.g. Schutjens et al. (2004)) to silica dissolution and precipitation in geothermal systems (Xu et al. (2004)). K() is generally described by power laws, which is surprising because only some of the studies refer to fluid-transport induce porosity changes and it cannot be expected that K(φ) relationships resulting from different mechanisms are comparable and generally applicable. In the present paper we discuss principle flow mechanisms that determine porositypermeability relationships during dryout processes in flow through geometry, i.e. under two-phase flow conditions. We firstly summarize our findings so far and draw a rough picture of the processes that lead to porosity and permeability reduction. We presents results from core-flood experiments that were performed in order to visualize solute transport on a pore scale. For this we inject dry SC CO2 in brine saturated rocks. The SCA2014-Temp Paper A075 3/12 experiments were carried out in a pressure and temperature regime and at flow rates realistic for potential sequestration operations. From past and present observations we draw conclusions on the dependency of K(φ) on the rock’s pore structure. EXPERIEMTAL EVIDENCE AND HYPOTHESIS SO FAR In an earlier study we investigated salt precipitation in Berea sandstone (Ott et al., 2010 and 2012 I). It has been shown that dryout occurs preferentially at the injection side. The resulting capillary pressure gradient might causes brine flow – and hence solute transport – in direction of the injection point if not compensated by advection. As consequence it is possible to precipitate locally (e.g. at the injection point) more salt than originally dissolved in the brine phase in the same volume. This macroscopic capillary-driven solute transport determines the reduction of porosity, however, it does not explain the observed effect on effective permeability, Keff,CO2=K·kr,CO2 – i.e. the K() relationship and the relative permeability saturation function, kr,CO2(SW). Experiments on Berea sandstone revealed an increase of effective CO2 permeability in the course of dryout (Ott et al., 2012 I). Our interpretation so far is that after the drainage process salt precipitates in the remaining brine phase which leaves the initially CO2occupied volume (the CO2 conducting pore system) essentially open. The desaturation by evaporation increases the CO2 relative permeability (kr,CO2) with the overall effect of increasing Keff,CO2. However, the final proof of the microscopic mechanism is pending and subject of the present paper. In a study on carbonates we found the opposite trend; a permeability reduction of one to three orders of magnitude has been found during dryout (Ott et al., 2012 II). In contrast to Berea sandstone, the carbonate (Middle East dolomite field samples) rock type shows a more complicated multi-porosity pore structure. The injected CO2 invades the macropore space, but not the micro porosity due to the high capillary entry barrier. Effectivepermeability reduction would require salt precipitation in the CO2 conducting channels, which would be in contrast to the observations in the sandstone experiments. This needs to assume solute transport from the micro-porous regions to the macro porosity which is on the first view counter intuitive; in literature about porous media drying – e.g. drying of soils in “pass-by” geometry, evaporation causes solute transport in the opposite direction – from macro to micro – driven by capillary pressure gradients (Lehmann and Or, 2009). We explain the flow from micro to macro by considering a dual porosity system characterized by two separate capillary pressure curves being in capillary contact to each other. This situation is illustrated in the left panel of Figure 1. Capillary equilibria between both sub systems, the macro porous (black line) and the micro porous system (red line) are defined as horizontal lines, i.e. as the same capillary pressure in both sub systems. As the entry pressure of the micro porous system has been overcome, by evaporation and not by viscous displacement, a minor further reduction of the water saturation in the macro pores will cause a desaturation of the micro pores by brine flow to the CO2 conducting pore system where water evaporates and salt precipitates. Hence, the micro porosity acts as brine/salt reservoir for the macro porous system. An important SCA2014-Temp Paper A075 4/12 controlling parameter turned out to be the volumetric ratio between macro and micro porosity. However, a direct observation of the solute transport between micro and macro porosity is still pending and is addressed in the following. MICRO-SCALE EXPERIMENTS For verification of the above described model we performed small-scale core flood experiments and use micro computerized tomography (CT) as analytical tool for in-situ pore scale imaging. The samples were small in diameter (4 mm  and 20 mm length) in order to image at a resolution of 2 m (voxel size). The samples were tightly fit in Viton sleeves with an inlet and outlet piece at each front face of the sample for fluid injection and production. This assembly was mount in a carbon fibre based core holder. We used water to apply confinement pressure in the space between core holder and sleeve in order to avoid bypassing of injected fluids. The core holder was vertically placed in a CT scanner for in-situ imaging of the rock fluid system. Fluids were injected and produced via the inlet and outlet piece, respectively. Two displacement pumps were used operating in a constant pressure mode at the inlet and at constant retraction rate at the outlet. Rock types The experiments were performed on two different rock types, Berea sandstone and Estaillades limestone representing a simple-mono modal pore system and a dual porosity system, respectively. The rock types are comparable in the sense of permeability and porosity; the porosity of Berea is in average ~0.22 and ~0.3 in case of Estaillades. Also the permeability is in the same order of magnitude; for Berea ~500 mD and ~200 mD in case of Estaillades. However, the rocks are not ideal single and dual porosity systems, but show complex pore-size distributions as can be seen from the presented MICP curves and CT scans in Figures 1 and 2. In this paper we refer to macro porosity as porosity that can easily be resolved by CT scanning and to micro porosity as porosity that we cannot resolve. Both samples show also porosity with pore sizes in between micro and macro, which we refer to as meso porosity. Experimental procedure The rock samples were scanned in the dry state and subsequently saturated with KI-based brine by flooding the sample with about 100 PV (pore volumes) at ambient pressure and at the experimental temperature of 50°C. We subsequently pressurized the fluid to 100 bar and we continued flooding for another few PV in order to remove dissolved air. The saturation has been checked by CT scanning. There was no indication of a residual gas phase in the pore space. For most of the scans, the CT field of view was smaller than the sample with focus on the central part – the region of interest. Occasionally, the full sample volume has been imaged to rule out local precipitation of salt as discussed above. A single scan took about 45 minutes. We continuously scanned the region of interest during the experiment under flow conditions. SCA2014-Temp Paper A075 5/12 After pre saturation, SC-CO2 has been injected with a flow rate of about 0.12 ml/min (~2 PV/min) comparable to the earlier study (Ott et al., 2010 and 2012 I) representing realistic field flow rates near injection wells. We performed the second CT scan after injection of ~100 PV CO2. At this stage, there is no viscous brine displacement anymore and the cumulative water loss by evaporation is still negligible. In the following, we refer to this state as “after drainage” or “drained”. The scan after drainage is the second reference point for data evaluation. Continuous scanning allows to detect any change of the rock-fluid system. We define the point at which the sample state does not change anymore as point of dryout. The state of dryout or at the end of the experiment has been used as third reference point for evaluation. Experiment on Berea sandstone Figure 3 shows representative CT cross sections recorded during the experiment on Berea sandstone. Image (a) shows the rock sample in the dry initial state recorded with a voxel size of 2 m (as for all images shown in this paper). We observe the typical Berea rock structure with well resolved macro porosity, quartz grains, and meso porous areas in feldspar and clay minerals regions. Image (b) shows the same slice at the end of drainage. The brine phase is highly x-ray absorbing and hence appears as bright areas in the pore system. According to the capillary pressure behavior, CO2 invades the macro pores and the remaining brine phase occupies essentially the meso pore system and part of the macro porous system is still occupied with brine due to the low efficiency of CO2-brine displacement. Image (c) was recorded 10 hours after drainage was started where the sample reached steady state as indicated by CT – the sample is dry and only CO2 and solid salt occupies the pore space. The dryout time is in line with earlier observations (Ott et al., 2010 and 2012 I). The main difference of (c) compared to the initial scan (a) are the bright spots that indicate the location of precipitation. Most salt precipitates in regions where meso porosity is present and where the remaining brine phase was located in the earlier stage. Precipitation in the macro pores is not so obvious. Scans (a) to (c) in Figure 3 can now be used to determine the saturation states in the macro porosity – the CO2-conductive pore system. By segmenting the dry scan we infer the macro-pore system and by segmenting scan (b) the CO2 flooded volume after drainage. To obtain the remaining-brine phase we subtract the binary images (a) from (b), and from the subtraction of the binary images (a) from (c) we infer the solid salt phase in the macro pores. The phases after drainage – CO2 and brine – and after dryout – solid salt – are shown in the images (d) and (e). As indicated earlier, salt precipitates just to a minor extent in the CO2 conducting porosity and furthermore, there is no salt precipitation (green) in the early CO2 percolation channels (orange). A 3D illustration of a sub volume can be found in Figure 4. Image (a) shows the CO2 percolation path system after drainage and (b) the final precipitation pattern at the end of the experiment. Both volumes are superimposed in image (c), which shows that both occupy complementary pore space. SCA2014-Temp Paper A075 6/12 An interesting point is that simultaneously to the observed porosity reduction there are regions in the sample where the porosity seems to increase during dryout. This predominantly occurs in regions where CT indicates clay minerals and feldspar. Figure 3 (f) shows meso porous regions in the initial dry scan. The same region, but after the dryout is shown in (g) – the gray scale change indicates the precipitation of salt in the micro porous region, but new porosity is formed that is resolved in CT. From the current data set we cannot make statements about the underlying mechanism. Mechanisms in consideration are the contraction of clay minerals by increasing salinity and the buildup of crystallization pressure during precipitation that would damage the rock structure. Experiment on Estaillades limestone We performed the same experiment under the same conditions on Estaillades limestone. The major difference to the sandstone experiment is the presence of micro porous grains with pore sizes well separated from that of the macro-porous system as discussed above. Figure 5 shows CT cross sections of the initially dry rock (a), the same cross section after brine saturation (SW=1) (b) and after drainage (c). The images (d) and (e) are obtained during dryout after ~24 hours and after ~7days of CO2 flooding. Due to the high x-ray absorption coefficient of iodine, gray scale changes are determined by salt migration – i.e. solute transport – and other contributions can be neglected. In the following we speak of “depletion” when the gray-scale value decreases and from “enrichment” when the gray-scale value increases corresponding to decreasing and increasing iodine concentration. The early stage of the experiment shows similarities with the sandstone scenario; nonwetting CO2 invades the macro-porous regions. The remaining brine stays in the meso porosity and in the micro porous grains as visible by the inverted gray-scale contrast between solid and micro porous grains compared to the dry scan. From image (c) to (d) we observe two trends: (1) a depletion of the meso-porous regions and (2) an enrichment in the micro-porous grains. Image (j) highlights the changes from image (c) to (d); orange/red color indicate depletion and green/blue indicate enrichment of salt. In a later stage of the experiment the trend is reverse. Between the images (d) and (e) there is a weak depletion of the micro porous area which indicates reverse flow. For visualization we scaled images (c) to (e) to highlight depletion in the micro porous regions. The scaled images are shown in Figure 5 (f) to (h). Image (g) is clearly brighter than image (f) and (h) recorded before and after (g). Depletion of individual spots with larger pore sizes in the grains and enrichment of the surrounding is best visualized in the difference image (j). While micro-porous grains are depleted, we observe enrichment in the surrounding – i.e. in formerly CO2 occupied volume. We expect this trend to continue, but had to terminate the experiment after 7 days of CO2 injection probably without reaching full dryout. SCA2014-Temp Paper A075 7/12 An interesting observation is that the actual time that is needed to dry out a rock sample by gas injection strongly varies from one rock type to another. The reason is likely to be the contact area between the brine phase and the injected drying agent CO2 in the macro, meso and micro porosity, i.e. on the pore structure of the rock type. But also heterogeneity and the associated bypassing of brine saturated regions might play a role. Hence, prediction of the time scale of dryout in carbonates will remain a challenge. ITERPRETATION AND SUMMARY The data presented so far together with a couple of simple arguments allow to construct a model that supports the findings in the frame of earlier work. The respective processes are illustrated Figures 6 (for single porosity sandstone) and 7 (for dual porosity carbonate). For single porosity sandstone we find: o Viscous brine displacement and water evaporation act in parallel, but were found to be dominant at different time scales. While residual brine saturation is usually reached after injection of a few pore volumes, complete dryout was reached after several hundreds to thousands PV of injected CO2. There is essentially no viscous brine displacement during dryout and precipitation. o The structure of the residually trapped brine phase is determined by capillary forces. o After reaching the solubility limit, salt precipitates in the brine phase, and hence in the volume occupied by brine. o The brine saturated volume retracts during evaporation leaving the previously precipitated salt behind. o Water transfer occurs at the CO2-brine interface where the system first exceeds the solubility limit – precipitation will occur predominantly at this interface. o There is water partitioning across the CO2-brine interface but no water/brine flux into the CO2 flow channels – the flow channels serve for water vapor transport, but stay open since salt precipitates in the residual brine phase. o Because of the separation of time scales of viscous displacement and evaporation, the maximum pore volume that can be filled by precipitate is the volume corresponding to the residual brine saturation after viscous displacement, SW,res. We conclude that Ssalt,max=SW,res. In Berea sandstone is SW,res≈0.2, which corresponds to the maximum observed salt accumulation (Ott et al., 2010 and 2012 I). o Since (1) drying decreases water saturation (SW) and respectively increases kr,CO2(SW) in time, and (2) salt precipitates in the brine phase only, the effective permeability (Keff,CO2=K·kr,CO2) can only increase in time, irrespectively of absolute permeability reduction. These mechanisms, in principle, hold also for the more complex pore structure of carbonates. However, micro porosity leads to additional effects substantially influencing where the salt precipitates and eventually K(): o During primary drainage, CO2 invades only the macro porous system characterized by the lowest entry pressure and the micro porous volumes serve as brine reservoirs. SCA2014-Temp Paper A075 8/12 o If a micro porous grain stays in contact with both, the residual brine phase in the macro (meso) pores and the CO2 phase, the volume of water that evaporates gets refilled by the brine phase of the macro (meso) porous system. Evaporation and refill leads to an increase of brine salinity in the micro pores. o This is in analogy to the above discussed dual porosity effect in soils and the pass-by situation (dry CO2 passes brine saturated micro-porous grains). Thus it links to wellknown physics, but does not explain the earlier observation of a generally strong permeability reduction which would require solute transport in the opposite direction – from the micro-porous regions to the CO2 conducting channels. o By further depletion of the macro (meso) porous system, the brine supply to the micro porous grains is not anymore sufficient to prevent dryout – CO2 invades the micro porous grains. This is equivalent to overcoming the entry pressure to the micro porous system, but by evaporation and not by viscous displacement. o After exceeding the capillary-entry pressure of the micro pores, the brine flow is reversed – a small saturation change in the macro pores will led to brine flow from the micro to the macro-porous system to reach capillary equilibrium of both subsystems. o A desaturation of the micro pores by brine flow – and not by water vapor – leads to an effective salt transport to the CO2-flow path ways (the macro porous system) where salt precipitates. This finally leads to the very effective reduction of permeability compared to the rather mild permeability reduction observed in single modal sandstone. ACKNOWLEDGEMENTSThe authors thank Steffen Berg and Saskia Roels for inspiring discussions and reviewingthe manuscript. Alon Arad and Maher Alkarra are acknowledged for support in dataprocessing. This work has been performed with support of the Qatar Carbonates andCarbon Storage Research Centre. REFERENCESBachu, S., Gunter, W. D., 2004. Overview of acid-gas injection operations in western Canada. In:Proceedings of the 7th International Conference on Greenhouse Gas Control Technologies,Vancouver, Canada. Vol. 1.Giorgis, T., Carpita, M., Battistelli, A., 2007. 2D modeling of salt precipitation during the injection of dryCO2 in a depleted gas reservoir. Energy Conversion and Management 48, 1816–1826.IPCC, 2005. IPCC Special Report on Carbon Dioxide Capture and Storage. Cambridge University Press,UK.Lehmann, P., Or, D., 2009. Evaporation and capillary coupling across vertical textural contrasts in porousmedia. Phys. Rev. E 80, 046318.Ott, H., de Kloe, K., Marcelis, F., and Makurat A., 2010. Injection of Super-Critical CO2 in Brine SaturatedSandstone: Pattern Formation during Salt Precipitation. International Conference on GreenhouseGas Technology (GHGT-10), Amsterdam, Netherlands, September 19-23, 2010, Energy Procedia4 (2011) 4425 4432. & Salt precipitation due to supercritical gas injection: I. capillary-drivenflow in unimodal sandstone. Submitted to the International Journal of Greenhouse Gas Control2014. SCA2014-Temp Paper A0759/12 Ott, H., de Kloe, K., van Bakel, M., Vos, F., van Pelt, A., Legerstee, P., Bauer, A., Eide, K., van derLinden, A., Berg, S., and Makurat, A., 2012. Core-flood experiment for transport of reactive fluidsin rocks. Rev. Sci. Instrum. 83, 084501–1–084501–16.Ott, H., Snippe, J., de Kloe, K., Husain, H., and Abri, A., 2012. Salt Precipitation due to SC-Gas Injection:Single versus Multi Porosity Rocks. International Symposium of the Society of Core Analysts,Aberdeen, Scotland, UK, August 27-30, 2012, SCA2012-07. & Salt Precipitation due toSupercritical Gas Injection: II. Single vs. Multi Porosity Rocks. Submitted to the InternationalJournal of Greenhouse Gas Control 2014.Pape, H., Clauser, C., Iffland, J., 1999. Permeability prediction based on fractal pore-space geometry.GEOPHYSICS 64 (5), 1447–1460.Pruess, K., Garćia, J., 2002. Multiphase flow dynamics during CO2 injection into saline aquifers.Environmental Geology 42, 282–295.Pruess, K., Müller, N., 2009. Formation dry-out from CO2 injection into saline aquifers: 1. Effects of solidsprecipitation and their mitigation. Water Resour. Res. 45, W03402.Schutjens, P. M. T. M., Hanssen, T. H., Hettema, M. H. H., Merour, J., de Bree, P., Coremans, J. W. A., G.,H., 2004. Compaction-induced porosity/permeability reduction in sandstone reservoirs: Data andmodel for elasticity-dominated deformation. SPE Reservoir Evaluation & Engineering 7 (3), 202–216.Xu, T., Ontoy, Y., Molling, P., Spycher, N., Parini, M., Pruess, K., 2004. Reactive transport modeling ofinjection well scaling and acidizing at Tiwi field, Philippines. Geothermics 33, 477–491. 0.0 0.2 0.4 0.6 0.8 1.00.010.11101001000 0.01 0.1 1 10 10