This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.
The experimental studies described in Chapter 5 demonstrated that salt precipitation can cause severe impairments in injectivity. The porosity and permeability data of the altered samples was used to calibrate a Verma-Pruess "tube-in-series" model (Verma and Pruess, 1988) employed in numerical simulations, which are presented in this Chapter. Similarly to the 1-D simulations presented in Chapter 3, these are performed with the general-purpose reservoir simulation code TOUGH2, coupled with the fluid property module ECO2N which can provide accurate fluid properties for the system CO2/water/NaCl for pressures up to 60 MPa, temperatures in the range 10 - 110 oC, and salinity up to full NaCl saturation (Pruess, 2005). TOUGH2/ECO2N is used to model multiphase flow subject to viscous, capillary and gravity forces; partitioning of H2O and CO2 between aqueous and CO2-rich phases; precipitation and dissolution of halite (NaCl). As solubility of water in the CO2-rich phase is small, typically a fraction of percent, effects of dissolved H2O on the density and viscosity of the CO2-rich phase are neglected.
This work focuses exclusively on the precipitation of salt contained in the formation water during CO2 injection-induced formation dry-out. Chemical reactions between CO2, fluids and formation minerals can also be responsible for precipitation and dissolution effects. However, as explained in Section 5.4.2, such interactions appears to have a minor impact in the wellbore area compared to salt deposition induced by water vaporisation and are not considered in this work. Non-isothermal effects are also not considered. As Giorgis et al. (2007) indicated in their study, simulations of supercritical CO2 injection under non-isothermal conditions may not produce different results than when the CO2 is injected at the same temperature of the undisturbed reservoir. These, however, could be important in the cases where CO2 is injected at a temperature significantly different from that of the rock formation.
Two different sets of simulations are presented in this chapter. In the first, a two-dimensional radial-vertical geometry (2-D R-Z) is used to study the impact of the interplay of gravity and capillary effects, and of important parameters such as injection rate, rock porosity, irreducible water saturation and water salinity on salt deposition. In the second set of simulations, the effect of salt precipitation on the sealing unit is tested using a similar 2-D R-Z geometry.
An idealised 2-D radial model is used to represent CO2 injection in a saline aquifer. The geological formation is assumed to be homogeneous and isotropic and it is modelled as a horizontal disc with thickness of 10 m, discretised into 10 grid layers of constant height of 1 m. The numerical grid is extended to the distance of 100,000 m in order to make the system infinite acting for the time period simulated (10 years). Figure 6.1 shows a schematic representation of the geometric model used.
The wellbore is located on the axis of symmetry of the disc and has a radius of 0.2 m. The total injection rate is uniformly distributed among the 10 reservoir layers. The horizontal discretisation was designed in such a way as to obtain great spatial resolution near the injection well, i.e. the region where highest gradients occur. From the well sandface, a fine radial grid has been built starting with an element of 0.01 m width and then increasing the element widths following a logarithmic progression. In total, 75 grid blocks have been used for each layer. The upper and lower boundaries of the formation are close, simulating impermeable layers at the top and at the bottom of the reservoir.
geometric model_Chapter 6.jpg
Figure 6.1: Schematic representation of the geometric model used.
Initial pressure and temperature were respectively chosen as 12 MPa and 45 oC, representative of the thermodynamic conditions which might occur in a saline formation at a depth of approximately 800-1000 m. The brine salinity value of 25-wt% dissolved NaCl has been chosen to be close to the salt concentration used for the vaporisation tests at reservoir conditions. Porosity was set at 20 % and permeability at 100 mD. The aqueous phase irreducible saturation was assumed to be 0.30. The relative permeability and capillary pressure has been calculated with the van Genuchten models choosing parameters similar to the those used for the 1-D numerical simulations presented in Chapter 3. As shown by Pruess and Müller (2009), aqueous diffusion has negligible effects on solids precipitation and was not included in this study.
CO2 injection was simulated for 10 years at a constant rate of 1 kg/s. Before performing the injection, a numerical simulation was ran in order equilibrate the pressure inside the formation according to the hydrostatic gradient, which gives a pressure of around 11.95 MPa at the top of the formation and a slightly higher pressure of 12.05 MPa at the bottom. The specifications chosen for this model are summarised in Table 6.1.
Salt precipitation induced by the CO2 injection process is monitored and quantified in the different regions of the formation analysing the changes of solid saturation, that is the fraction of pore volume occupied by salt. From the values of solid saturation, the resulting porosity is computed. Corresponding permeability is also estimated according to the Verma-Pruess pore network model (See Appendix C).
The Verma-Pruess model has been recently used in numerical modelling studies to represent the effect of salt precipitation (Pruess and Müller, 2009). However, in Pruess and Müller (2009) the model was not calibrated with experimental data derived from salt deposition but instead using similar parameters (Î“=0.80; Ð¤c=0.90) to a previous study of permeability changes due to precipitation of amorphous silica at a geothermal injection well in the Philippines (Xu et al., 2004). This parameterisation gave very severe decrease in permeability for a limited porosity reduction.
Müller et al. (2009) also used a Verma Pruess model to represent the relationship between changes in petrophysical properties induced by the formation dry out effect. In this work, a vaporisation test was specifically executed to calibrate the permeability model used in the simulation. The observed experimental maximum permeability reduction (âˆ¼60%) was matched with the maximum salt precipitation (âˆ¼16%) obtained from numerical modelling. The parameters Î“ and Ð¤c were assumed both equal to 0.567. In the author's view, this approach presents two limitations, namely the model was calibrated with a single result of permeability alteration, and the values of porosity and permeability reduction used are not related to one another.
In this work, as illustrated in Chapter 5, vaporisation tests have been conducted at different pressure/temperature conditions to accurately represent the effect of salt precipitation. Figure 6.2 shows the different Verma-Pruess models obtained. The continuous blue line (Lab-TUD) is the model calibrated from the experiments conducted at TU Delft at reservoir conditions. The parameters Î“ and Ð¤c were assumed respectively equal to 0.70 and 0.59. The continuous red line is instead the model calibrated with the data from the experiments executed at Imperial College at atmospheric pressure and temperature. In this case, the parameters Î“ and Ð¤c were respectively 0.95 and 0.85. For comparison, the permeability model used in Pruess et al. (2009) and Müller et al. (2009) are also displayed.
Figure 6.1: Verma-Pruess salt precipitation models.
It can be observed that the two models obtained in this study are significantly different. Data from experiments at atmospheric pressure and temperature yielded a model which indicates great permeability changes for small porosity modifications, similar to Pruess et al. (2009). Whereas, the model calibrated with the results obtained at reservoir conditions is surprisingly close to the one used in Müller et al. (2009).
For all the numerical simulations presented in the next sections, the Lab-TUD permeability reduction model has been used.
In order to simplify modelling, the simulations reported in this Chapter were performed without implementing the modified permeability values in the flow calculation. On the other hand, previous studies showed that saturation profiles for gas and solid precipitate do not depend on reduction of permeability, as long as the decrement does not induce extreme pressurization, which in turn would modify the fluid properties (Pruess and Müller, 2009).
Table 6.1: Physical properties of the aquifer
100 x 10-15 mÂ²
4.5 x 10-10 Pa-1
Rock grain density
2,600 kg m-3
1 kg s-1
Relative permeability model and capillary pressure models
Liquid (van Genuchten, 1980)
Residual liquid saturation
Gas (van Genuchten, 1980)
Residual gas saturation
Sgr = 0.05
Capillary pressure (van Genuchten, 1980)
Residual liquid saturation
Slr = 0.00
P0 = 19.6 kPa
Assessment of Salt Precipitation within the Storage Formation
Initially, the injection of dry supercritical CO2 displaces the brine around the wellbore. A small part of CO2 dissolves in the brine, whereas a fraction of the saline water is vaporised by the flowing CO2. As water evaporates the salt content contained into the brine is enriched. Halite deposition occurs where the solubility limit of approximately 26.5% by weight is outreached. As the injection of CO2 proceeds, saline water continues to vaporise around the wellbore area until it completely disappears.
As Figure 6.3 shows, one year after commencing the injection, the plume of CO2 extended more across a region localised near the top of the formation at a distance of up to around 200 m from the wellbore. The presence of higher values of gas saturation at the top of the permeable interval can be attributed to the effects of buoyancy.
Figure 6.3: Gas saturation distribution after CO2 injection for 1 year.
According to the simulation (Figure 6.4), salt deposition is confined to a small area up to 4 m inside the formation. Within this area, the amount of the precipitated NaCl appears to be fairly uniform, with an average solid salt deposition of approximately 10%. A slight lower solid salt deposition can be measured at the top, whereas near the bottom a localised region with larger solid deposition (up to 13%) can be detected. These higher solid deposition values identified near the lower portion of the dry-out front are due to the interplay between capillary and gravity forces. In these areas, gravity has the effect of reducing the horizontal component of the flow vector, while the capillary forces remain unchanged. Therefore, the potential of backflow of aqueous phase increases towards the injection well providing a greater supply of precipitable salt. This behaviour had been previously reported by Giorgis et al. (2007).
As Figure 6.5 shows, on average, the absolute permeability is reduced to 40% of the initial value, as indicated by the laboratory experiments. Permeability reduction peaks to 48% in the region where the highest amount of NaCl is deposited.
SS_1yr.jpgFigure 6.4: Solid salt deposition distribution (NaCl) after CO2 injection for 1 year.
Figure 6.5: Permeability reduction (k/k0) due to halite scaling after CO2 injection for 1 year.
As the injection continues, the CO2 plume and the dry-out front progressively advance inside the formation. After 10 years, CO2 would travel as far as 650 m from the injection well (Figure 6.6), i.e. 3.25 times the advancement after one year of injection. Solid deposition profiles, shown in Figure 6.7, indicate that the halite scaling front also advanced 3.25 times from the first year, up to 13 m into the geological formation. It is possible to notice a slighter more pronounced vertical solid deposition trend, with a peak of 14% reduction of the pore volume. Absolute permeability reduces down to 51 % of the original value (Figure 6.8).
Figure 6.6: Gas saturation distribution after CO2 injection for 10 years.
Figure 6.7: Solid salt (NaCl) deposition distribution after CO2 injection for 10 years.
Figure 6.8: Permeability reduction (k/k0) due to halite scaling after CO2 injection for 10 years.
Starting from the 2-D injection model described in Section 6.3, which can be referred as Case 0, a sensitivity study was conducted to analyse the effects of different formation parameters on salt precipitation. Problem variations include using, one at a time, a different injection rate, porosity, residual water saturation, salt content and reservoir thickness. Table 6.2 resumes the different cases explored in this study.
Table 6.2: Sensitivity of salt precipitation to variations in formation parameters.
Problem variation from Case 0
Injection rate increased to 2.0 kg/s.
Injection rate decreased to 0.5 kg/s.
Porosity increased to 30 %.
Irreducible water saturation increase to 40 %.
Irreducible water saturation decreased to 20 %.
NaCl concentration decreased to 15 % by weight.
Aquifer thickness increased to 100 m and flow rate increased to 10 kg/s.
Injection distributed among only the last 5 m at the bottom.
Results for two cases with different injection rates, 0.5 kg/s and 2.0 kg/s, are displayed in Figures 6.9 - 6.12. The graphs are referred to different injection time periods, respectively 4 years and 1 year, in order to achieve the same amount of CO2 injected for both the cases.
For the larger injection rate (Case 1), the effect of gravity override is less important, and therefore vertical gradients appear less prominent. This can be seen observing the gas saturation map (Figure 6.9) which indicate a smaller tendency of CO2 to accumulate near the top of the formation. Close to the wellbore, the profile of gas saturation is nearly uniform, insomuch as the flow can be approximately considered as 1-D. As shown in Figure 6.10, salt precipitation also seems to be uniform, with average solid deposition of 9% and a peak of 10 % in close proximity to the injection well, which induces a maximum reduction of the absolute permeability to as little as 37 %.
Figure 6.9: Gas saturation distribution after CO2 injection for 1 year with a rate of 2.0 kg/s (Case 1).
Figure 6.10: Solid salt (NaCl) deposition distribution after CO2 injection for 1 year with a rate of 2.0 kg/s (Case 1).
For the smaller injection rate (Case 2), the horizontal components of injection-induced pressurization are reduced, whereas buoyancy forces are not changed and, therefore, the effects of gravity override on gas saturation distribution become more pronounced (Figure 6.11). According to this, more CO2 accumulates at the top of the formation. Rather differently from Case 1, for low injection rates, salt deposition varies significantly in the different regions of the formation, with marked vertical solid saturation trend (Figure 6.12). Average solid deposition is larger than in the previous case (approximately 13 %). Similarly to the first case studied (Case 0: injection rate equals to 1 kg/s), it is possible to identify an area near the lower portion of the dry-out front where larger amounts of precipitated salt can be detected. In this case, the maximum solid saturation is 24 %, which induces a permeability reduction as high as 79 % of the original value.
Figure 6.11: Gas saturation distribution after CO2 injection for 4 years with a rate of 0.5 kg/s (Case 2).
Figure 6.12: Solid salt (NaCl) deposition distribution after CO2 injection for 4 years with a rate of 0.5 kg/s (Case 2).
Case 3 investigates the influence of initial porosity. By increasing porosity from 20 % to 30 %, the absolute amount of salt precipitating is also increased; however, the solid deposition distribution predicted by numerical modelling appears to be unchanged. Due to the increment of global pore space in the formation, the CO2 plume reached a slightly smaller distance than in Case 0.
Case 4 and Case 5 examine the effects of a different value of irreducible liquid saturation. This parameter modifies the amount of brine that can be moved through the immiscible displacement mechanism, and consequently the fraction of liquid which is removed by evaporation into the CO2 stream. As expected, reducing irreducible liquid saturation down to 0.20 resulted in less precipitation of salt and reduction in permeability. Increasing liquid saturation to 0.40 had the opposite effect.
Case 6 investigates the effects of reduced brine salinity. As NaCl concentration was reduced by 40 %, from 25 %-wt to 15 %-wt, the salt deposition substantially decreased. After 10 years of CO2 injection, average solid deposition within the area affected by salt precipitation became 6 %, i.e. 40 % less than the average solid saturation for Case 0. Thus, average solid deposition proportionally decreased with the reduction in brine salinity. However, the peak solid deposition in Case 6 resulted in only 7 %, i.e. 50 % less than that determined for Case 0. Comparing to Case 0, also the gas saturation data presents some differences. Results indicate that the CO2 plume extends to a smaller distance into the aquifer. This is because less deposition of salt allows more CO2 to be stored in the pore space.
An analogous sensitivity study had been previously conducted by Pruess and Müller (2009) for a simpler 1-D case, which gave similar indications regarding the effects of variation in porosity, salinity and residual water saturation.
Case 7 investigates the effects of an increase in aquifer thickness. This simulation has been conducted increasing the vertical length of the formation by 10 times, from 10 m to 100 m. The injection rate was increased by 10 times as well, in order to maintain the number of pore volumes of CO2 injected over the 10 years injection period. Figure 6.13, illustrates the gas saturation distribution at the end of 10 years of CO2 injection. The results indicate that by increasing the aquifer thickness, the effects of buoyancy become stronger and the portion of CO2 that accumulates near the top of the reservoir increases. Figure 6.14 shows the presence of a local region near the bottom of the formation where salt accumulated, reaching a solid saturation of 18 %. Such porosity variation corresponds a permeability reduction of 64 %, i.e. 13 percentage points more than the maximum permeability reduction determined for Case 0.
Figure 6.13: Gas saturation distribution after 10 years of CO2 injection at 10 kg/s in a formation with a thickness of 100 m (Case 7).
Figure 6.14: Solid salt (NaCl) deposition distribution after 10 years of CO2 injection at 10 kg/s in a formation with a thickness of 100 m (Case 7).
Case 8 examines the influence of a different injection strategy. In this simulation, injection has not been distributed among all the reservoir layers but only among the lower 5 m. The total injection rate is unchanged. The gas saturation distribution did not present substantial differences from Case 0; however, the salt precipitation pattern was considerably modified. Figure 6.15 shows the solid salt deposition distribution after 10 years of CO2 injection. As previously observed, also for this case, it is possible to notice the presence of a region near the lower part of the dry-out front where salt abundantly precipitates. However, differently from the previous cases, Figure 6.15 shows also another area where solid saturation is particularly high (around 16 %). This is in close proximity to the wellbore, 3-4 m below the top of the formation, i.e. right above the part of the reservoir where CO2 injection is conducted. In this area, vaporisation is strong. The vertical components of injection induced-pressurisation is instead weak, since the flow near the well is approximately 1-D. Therefore, capillary pressure forces are able to recharge the vaporisation front with a continuous supply of new brine from the top of the formation provoking a severe accumulation of salt.
In the top regions of the aquifer, halite scaling is much lower compared to Case 0. This is because CO2, moving from the bottom due to buoyancy only, arrives to the top of the formation already saturated to an extend with water and can induce only limited vaporization of brine in the surrounding areas and resulting in limited salt deposition.
Figure 6.15: Solid salt (NaCl) deposition distribution after 10 years of CO2 injection distributed into only the lowest 5 m of the injection interval (Case 8).
Assessment of Salt Precipitation on the Caprock
The results presented in Section 6.3 indicate that capillary driven backflow can be responsible of the formation of local regions where salt accumulates resulting in great impairments in the reservoir petrophysical properties. This phenomenon is much more pronounced when the advancement of the drying front is somehow thwarted. As an example, it has been observed that due to the effect of gravity override, the horizontal progression of the dry-out front is delayed near the bottom of the formation and larger amounts of salt are deposited. In the ideal case of no advancement of the dry-out front in one direction, the localised reduction in permeability that can be obtained would be extremely high. This is because brine would constantly supply new brine on the same evaporation interface resulting in a continuous accumulation of salt, which would end only with the complete plugging of the pore space. A situation similar to the ideal case described could occur on the interface between the aquifer and the impermeable sealing unit. The vertical progression of the drying front is indeed abruptly interrupted by the caprock, under which CO2 accumulates. CO2 cannot permeate the caprock layer, however, can vaporise the water contained in it resulting in precipitation of salt. The importance of this phenomenon steams on the possibility that the associated permeability reduction can additionally improve the sealing capacity. Evidences suggesting this behaviour can be found in the afore mentioned study from Peysson et al. (2010).
A numerical study has been conducted in this research to investigate the effects of halite scaling on the sealing unit. The numerical model used is similar to the one illustrated in section 6.2. The main difference is that an additional impermeable 1 m layer is included at the top of the formation representing the caprock. The following formation parameters have been chosen for this layer: porosity equals 10 %, permeability equals 1 mD and the capillary pressure was computed using the van Genuchten model and a value of P0 equal to 5,000 kPa, which prevents any Darcy-driven flux. Similarly to the previous simulations presented in the Chapter, rock fluid interactions are not considered.
For the first simulation, referred to as Case 9, the same parameters employed for Case 0 in section 6.2 have been used for the aquifer formation. Unfortunately, the simulation had to be interrupted after only 10 days because salt saturation in the top first grid block representing the aquifer reached 1. In fact, when solid saturation reaches value 1 in one of the grid blocks the software is not able to continue the simulation. Figure 6.16 shows the solid saturation distribution following 10 days of CO2 injection. It is possible to observe the formation of a layer of very high solid saturation and zero permeability just underneath the caprock, around the wellbore. After 10 days, this layer extends only for a limited distance into the reservoir, within a radius of 0.10 m. However, during the overall time of a CO2 injection project, a NaCl layer with a radius of several metres can be expected to form. This would add extra sealing capacity in the area close to the wellbore, which is perhaps the most sensitive region for the risk of leakage.
Figure 6.16: Solid salt (NaCl) deposition distribution map after 10 days of CO2 injection (Case 9).
It is important to observe, that the eventual formation of such a salt barrier can be considered as a relevant mechanism in the context of CO2 storage because it could further improve the sealing effectiveness of a caprock in good conditions. However, it would not prevent leakages from one that is faulty at the first place. In fact, as Figure 6.17 illustrates, after 10 days of injection the CO2 plume entered almost 30 m into the reservoir, i.e. much more widely than the formed salt layer. Therefore, CO2 can easily permeate into the caprock if this is fractured or has an inadequate capillary entry pressure and permeability. As shown in Figure 6.18, the pressure disturbance involves even a larger area, reaching out to 3,000 m away from the wellbore.
Figure 6.17: Gas saturation map after 10 days of CO2 injection (Case 9).
Figure 6.18: Pressure field (Pa) after 10 days of CO2 injection (Case 9).
By reducing the NaCl salinity to 15% (Case 10), the formation of the salt layer is slowed down. Solid salt deposition reaches value 1 in the aquifer top left grid block after 18 days of injection.
Salinity was further reduced in order to allow the simulation to run for a longer period of time (Case 11). Formation parameters are the same used in Case 9 but NaCl concentration was set as low as 0.5 % by weight. Similarly to the previous cases, the total injection rate of 1 kg/s is distributed uniformly among the aquifer 10 layers. Solid salt deposition and permeability reduction data for an injection time of 1 year are respectively shown in Figure 6.19 and 6.20.
Following 1 year of CO2 injection, average solid deposition into the formation is almost null. This is due to the very low salinity chosen for the simulation. However, the two aquifer top layers present significant higher levels of salt deposition, reaching a maximum solid saturation of 47 %. Despite the very low brine salinity, according to the permeability reduction model used, underneath the caprock a completely impermeable salt barrier can be formed also in this case (Figure 6.20).
Figure 6.19: Solid salt deposition distribution after 1 year of CO2 injection (Case 11).
Figure 6.20: Permeability reduction due to halite scaling after 1 year of CO2 injection (Case 11).
An additional sensitivity study was conducted in order to analyse the effects of the different aquifer parameters on the formation of the salt barrier underneath the sealing unit. From Case 11, a number of problem variations were explored including different injection rates, porosity, residual water saturation and injection strategy. Table 6.3 presents the details of the different cases investigated.
Table 6.3: Case characteristics used to investigate the sensitivity of salt precipitation to variations in formation parameters.
Problem variation from Case 11
Injection rate increased to 2.0 kg/s.
Injection rate decreased to 0.5 kg/s.
Porosity increased to 30 %.
Irreducible water saturation increase to 40 %.
Irreducible water saturation decreased to 20 %.
Injection only in the bottom 5 layers
Cases 12 and 13 investigate the effect of changes in the injection rate. As seen before, when increasing the flow rate, solid salt deposition slightly decreases in most parts of the aquifer for the same amount of pore volume of CO2 injected. However, it seems that the effect on the formation of the salt barrier below the caprock is opposite. In fact, injection of 2 kg/s for 6 months (same total amount of pore volumes injected as in Case 11) increased the amount of salt deposited below the caprock. Solid saturation reached the maximum value of 53 % (Case 12). Similarly, decreasing the flow rate down to 0.5 kg/s (Case 13) increases salt precipitation in all the aquifer regions underneath the caprock.
Case 14 analysed the influence of initial porosity on the formation of the salt barrier below the caprock. Increasing porosity to 30 % slows the formation of the salt layer. In fact, using the hypothesis that the vaporization rate is independent to porosity as assumed in these simulations, the amount of salt deposited after a certain time will be the same than in Case 11. However, since in this case more pore volume is available, the relative reduction in void space that occurs is slower.
Cases 15 and 16 examine the importance of irreducible liquid saturation. Results presented in the previous Section demonstrated that this parameter controls the portion of brine that is not removed by immiscible displacement and, therefore, also controls the amount of salt deposited into the pores. However, the effect on the formation of the salt barrier is minimum. In fact, the large solid saturation reached underneath the caprock is only marginal due to the local salt contained in the brine but it mostly comes from the salt dissolved in the brine continuously coming towards the vaporisation front.
Finally, Case 17 investigates the effects of a different injection strategy. Similarly to the sensitivity study discussed in the previous section, the total injection rate of 1 kg/s was distributed only among the bottom 5 layers. Results show that no salt barrier underneath the caprock is formed by injecting only from the bottom. This is because CO2 that rising by buoyancy and coming to contact with the caprock is already saturated with water and cannot cause significant vaporisation.
In this Section, the possibility of a salt layer formation adjacent to the caprock was numerically explored together with the influence of the main petrophysical formation parameters onto the process. However, the actual occurrence of such phenomenon requires validation from field and laboratory experiments. It is also important to highlight that the rate of the salt layer formation obtained in this study was strictly dependent to the vertical discretisation used and it cannot be considered representative of field conditions. In these simulations, the reservoir layer under the sealing unit, in which the salt barrier forms, had a thickness of 1 m. However, previous studies demonstrated that salt deposition can accumulate in a much thinner layer, of the order of few millimetres of thickness, adjacent to the evaporating surface (Peysson et al., 2010). If, as indicated by the authors, halite deposits in a significantly narrower region, high values of solid salt deposition could be expected within a much shorter period of time.
The numerical study showed that halite scaling is likely to concentrate underneath the caprock, however, in the field some salt deposition could occur also inside the sealing unit. In fact, assuming very high vaporisation rates, the caprock might not be able to provide sufficient capillary-driven source of brine to the dry-out front and, therefore, the evaporation front could enter into the caprock promoting halite scaling on the inside.
To summarise, the following were established by the numerical simulations presented in the first part of this Chapter:
Salt deposition concentrates in an area of few meters around the wellbore inside the dry-out region.
Gravity and capillary effects slow the advancement of the CO2 vaporisation front and induce backflow of brine producing local areas near the bottom of the aquifer where deposition of salt occurs to a larger measure. For an equal amount of CO2 injected into the reservoir, these effects result accentuated as the injection rate is decreased.
Gravity override effects become more significant in aquifers with larger aquifer thickness, where the formation of local region near the bottom of the formation with very high values of solid NaCl deposition can be expected.
The relative amount of solid salt precipitated increases with brine salinity and irreducible water saturation while list it results no significantly dependent to porosity.
The second set of numerical simulations showed that at the interface between aquifer and caprock a stationary dry-out front can be formed, inducing the formation of a salt layer which acts as an extra protection against leakages around the wellbore. The formation of the salt layer cannot prevent leakages from fractured on unsuitable caprocks, but it can be important in improving the sealing capacity of caprocks in satisfactory conditions. The numerical simulations have also established the following:
Increments of brine salinity accelerates the formation of the salt layer between aquifer and caprock.
For an equal amount of CO2 injected into the aquifer, the formation rate of the salt barrier appears to rise as injection rate is increased.
In aquifer with larger values of porosity, high values of solid saturation are reached in larger periods of time.
Irreducible water saturation seems to have not significant influence on the process.
This process seems to occur only when part of the CO2 is injected in the close proximity of the caprock. In fact, in a simulation where CO2 was injected only in the lower part of the formation, there is no sign of such phenomenon.
The occurring of the salt layer has not been previously reported in the literature and it requires verification from field and laboratory experiments.
Laboratory experiments appear indispensable to verify the occurrence of these mechanisms. Here, an experimental approach is proposed in order to study the presence of the salt barrier phenomenon. For this experiment, a caprock sample saturated with saline water is placed in a flow cell in vertical position. The sample is put in contact with CO2 on one free side at the bottom and with a brine reservoir on the top side. The CO2 and brine reservoir have to be at the same pressure in order to prevent Darcy flow movement of one phase into the other. This pressure can be representative of reservoir conditions. Anyway, it is important that the pressure is kept below the capillary entry pressure of the caprock. On the lower face of the caprock, CO2 and brine will be in contact and will slowly start to dissolve in each other. As the brine vaporises, more saline water from the brine reservoir will move towards the evaporating surface. Diffusion of water saturated CO2 into the gas reservoir assures that CO2 will be able to take up more water guaranteeing continuous vaporisation on the core face. As the evaporation process continues, salt concentration progressively increases and halite scaling eventually occurs. The process slowly carries on until all the CO2 contained in the gas reservoir becomes fully saturated with water. However, this experiment might take very long time because the mechanism is governed by diffusion. An alternative faster option consists in placing the flow cell in an oven, with one side free and the other connected to the brine reservoir. The rate of vaporisation in this case can be regulated adjusting the temperature in the oven.