Transp Porous Med (2013) 100:259–278 DOI 10.1007/s11242-013-0215-1

Using Compressed Air Injection to Control Seawater Intrusion in a Confined Coastal Aquifer Dong-mei Sun · Stephan Semprich

Received: 8 November 2012 / Accepted: 1 August 2013 / Published online: 28 August 2013 © Springer Science+Business Media Dordrecht 2013

Abstract Seawater intrusion into groundwater is an important problem in many coastal regions. Freshwater injection has been widely used to avoid seawater contaminating freshwater systems, but is not an attractive solution where freshwater is limited. We investigated the effects of injecting compressed air on seawater movement in a confined coastal aquifer using a numerical simulation. We used TOUGH2/EOS7 software to analyze the effects of injecting compressed air in preventing seawater intruding into a hypothetical confined coastal aquifer (Henry’s problem), simulating steady-state initial conditions and comparing the results with the literature sources. We then performed a transient-state numerical simulation to quantify the seawater intrusion control efficiency achieved by injecting air. The results showed that injecting compressed air can mitigate seawater intrusion: Saltwater was ejected from the aquifer and the seawater circulation disappeared. The injected air flowed upward and spread laterally near the top of the aquifer because of the groundwater and air densities. Injecting air significantly increased the groundwater and gas pressures near the air injection zone and at the top of the aquifer. The air injection rate increased rapidly, then increased gradually. Freshwater injection was also simulated using settings similar to those used for air injection, and this showed that seawater intrusion is prevented more efficiently by freshwater injection than air injection. However, freshwater resources are valuable, whereas air is readily available, so injecting air to mitigate seawater intrusion has great potential. The modeling approach that we used will be used as a foundation for future work. Keywords Groundwater · Seawater intrusion · Compressed air injection · Numerical simulation

D. Sun (B) State Key Laboratory of Hydraulic Engineering Simulation and Safety, Tianjin University, Tianjin 300072, China e-mail: [email protected] S. Semprich Institute of Soil Mechanics and Foundation Engineering, Graz University of Technology, 8010 Graz, Austria

123

260

D. Sun, S. Semprich

1 Introduction Groundwater is the main source of freshwater in many coastal areas, but seawater intrusion can degrade groundwater quality. Seawater intrusion increases the chloride concentration in groundwater, making it unsuitable for use without purification. Therefore, the efficient control of seawater intrusion is very important in protecting groundwater resources. A number of approaches have been used to control seawater intrusion, including artificial recharging (recharging with freshwater), constructing hydraulic or physical barriers (e.g., building underground dams), and abstracting saltwater (Todd 1974; Newport 1977; Kiuchi et al. 1996; Moridis et al. 1999; Sherif and Hamza 2001; Dhar and Datta 2009; Jorreto et al. 2009; Kacimov et al. 2009; Abd-Elhamid 2010; Javadi et al. 2012). These methods involve huge investments of money and time for the construction work involved, to insure that freshwater is protected, and in the use of available freshwater resources for injection (Bear et al. 1999). It is also difficult to procure enough freshwater resources to inject in areas with limited freshwater sources. New methods for controlling seawater intrusion into coastal aquifers are, therefore, desirable. Injecting air or gas mixtures into the subsurface has been successfully used in a wide range of geo-environmental situations. For example, gas injection has been used to enhance oil recovery or revitalize mature oil reservoirs by displacing oil with injected gases (Rao 2001). Air has also been injected into aquifers to remove, and to enhance the aerobic biodegradation of, volatile compounds (Brooks et al. 1999; Reddy and Adams 2001). Compressed air is used in compressed air tunneling to dewater and stabilize the ground by applying air pressure to the tunnel to balance the groundwater and stabilize the tunnel face (Kramer and Semprich 1989). A water recovery approach using air injection has been proposed by Moridis and Reddell (1991), with the aim of recovering water stored in the unsaturated zone of a depleted aquifer, recovering water that would be unavailable using conventional techniques. Air is compressed in compressed air storage systems during periods of low electric power demand and stored in an underground reservoir. The compressed air is then withdrawn from the reservoir to drive electric generators in periods of high electric power demand (Meiri 1981). Injecting air could also be used to solve the problem of seawater intrusion. Dror et al. (2004) showed the physical and chemical effects of air injection into sand-filled laboratory cells in their preliminary assessment of the use of air injection barriers for reducing groundwater flow. These tests offer preliminary proof of the concept that air injection barriers might effectively inhibit the undesired subsurface flow of saltwater or contaminated groundwater. In this study, we used numerical simulations to quantify the effects of injecting compressed air on the movement of intruded seawater in confined coastal aquifers. Simulations of air–water two-phase flow and solute transport induced by compressed air injection were conducted using TOUGH2/EOS7 software. EOS7 is a module for the TOUGH2 simulator for the two-phase flow of saline water and air (Pruess 1991; Pruess et al. 1999).

2 Numerical Model Construction 2.1 Governing Equations The governing equations in the TOUGH2/EOS7 software have been described by Pruess (1991). Three mass components (water, brine, and air) are considered, but heat transport is not considered, and all processes are assumed to be approximately isothermal. The following mass balance equation is solved by TOUGH2.

123

Air Injection to Control Seawater Intrusion

261

∂ Mκ κ κ = −div(Fadv + Fdiff ) + qκ , ∂t

(1)

κ and F κ are where M κ is the local mass density of component κ (water, brine, or air). Fadv diff κ the advective and diffusive mass flux rates of component κ, and q is a sink/source term for each component. The general form of the local mass density in Eq. 1 can be written as  X βκ φ Sβ ρβ . (2) Mκ = β

The total local mass density of component κ is obtained by summing over the fluid phase β (liquid or gas phase), the liquid phase being a mixture of (pure) water and brine. φ is the porosity, Sβ is the degree of saturation of the β phase, ρβ is the density of the β phase, and X βκ is the mass fraction of component κ in the β phase. The diffusive mass flux of component κ is given by  κ = −φτ0 τβ (Sβ )ρβ dβκ ∇ X βκ , (3) Fdiff β

dβκ

where is the molecular diffusion coefficient of component κ in phase β. τ0 τβ (Sβ ) is the tortuosity, which is split into a soil-dependent factor τ0 and a saturation-dependent factor τβ (Sβ ). Using effective diffusion coefficients, Eq. 3 can be rewritten as  κ κ κ =− ρβ Deff, (4) Fdif β ∇(X β ) β

with the effective diffusion coefficients κ = φτ0 τβ (Sβ )dβκ . Deff,β

(5)

The diffusion coefficients of the gases are functions of pressure ( p) and temperature (T ):   p0 T + 273.15 ◦ C θ (6) dβκ ( p, T ) = dβκ ( p0 , T0 ) p 273.15 ◦ C with θ = 1.8. TOUGH2 offers three tortuosity models: τ0 τβ = τ0 krβ (Sβ ) (the relative permeability model), τ0 τβ =

10/3 φ 1/3 Sβ

(7)

(the Millington–Quirk Model), and

(8)

τ0 τβ = τ0 Sβ (constant diffusivity),

(9)

to which constant diffusivity is applied using τ0 = 1.0. The advective flux of a component in mass balance Eq. 1 is defined by the flux Fβ in the appropriate phases:  κ = X βκ Fβ . (10) Fadv β

The mass flux Fβ in Eq. 10 is determined using a multiphase version of Darcy’s law: Fβ = −k

ρβ krβ (Sw ) (∇ pβ − ρβ g), μβ

(11)

123

262

D. Sun, S. Semprich

where k is the absolute permeability, krβ is the relative permeability to phase β, μβ is the viscosity, pβ is the fluid pressure, and g is the gravity acceleration vector. The fluid pressure pβ is the sum of the gas pressure pg and the (negative) capillary pressure pc : pβ = p g + p c .

(12)

There are several options for defining the capillary pressure function pc (Sw ) in TOUGH2. The simulations presented here use the function proposed by Leverett (1941),   pc = p0 γ 1.417(1 − Se ) − 2.120(1 − Se )2 + 1.263(1 − Se )3 , (13) where p0 is the air entry pressure, γ is the water–air surface tension, and Se is the effective liquid phase saturation, Se = (Sl − Slr )/(Sls − Slr )

(14)

with liquid saturation Sl , residual liquid saturation Slr , and maximum liquid saturation Sls . For Eq. 11, TOUGH2 offers several relative permeability functions. In the simulations presented here, the relative permeability–saturation relationship is modeled using the function according to Fatt and Klikoff (1959): krg = (1 − Se )3

(15)

krl = Se3

(16)

for the gas phase and

for the liquid phase. 2.2 Domain Description and Numerical Method In this study, we used a simplified and hypothetical confined coastal aquifer to investigate seawater intrusion into a confined coastal aquifer, a problem also known as Henry’s problem (Henry 1964a,b). This case involves seawater intrusion into a confined coastal aquifer, which was studied in a slice of an aquifer region using a vertical cross section perpendicular to the coast and parallel to the steady-state aquifer flow path. The aquifer under consideration was homogenous and isotropic, and bounded above and below by impermeable strata. The aquifer was exposed to hydrostatic seawater at the right boundary of the domain and a constant freshwater flux (or the equivalent hydrostatic head) at the left boundary (Fig. 1). Henry’s problem has been studied by Pinder and Cooper (1970), Lee and Cheng (1974), Segol et al. (1975), Frind (1982), Huyakorn et al. (1987), Voss and Souza (1987), Oldenburg and Pruess (1995), Cheng et al. (1998), Liu et al. (2001), Rastogi et al. (2004), Karasaki et al. (2006), and Abd-Elhamid and Javadi (2011). The domain was 200 m long and 100 m thick (Fig. 1). The parameters used are given in Table 1. The dimensions and parameters were selected to allow the results of our simulation to be compared with simulation models that have previously been published (Frind 1982; Huyakorn et al. 1987; Cheng et al. 1998; Liu et al. 2001; Abd-Elhamid and Javadi 2011). Seawater intrusion into this simulated confined coastal aquifer was treated as an idealized 2D problem. Assuming air was injected into the confined aquifer through a row of closely spaced wells perpendicular to the vertical cross section, the model was used to simulate air injection into a 2D section perpendicular to the coast and to the row of injection wells. Further work needs to be performed to develop a 3D numerical model for simulating air injection using separate boreholes.

123

Air Injection to Control Seawater Intrusion

263

Fig. 1 Vertical section of the simulated aquifer and boundary conditions Table 1 The parameters used in the numerical simulation of Henry’s problem (Liu et al. 2001) Parameter

Value

Water and brine molecular diffusion coefficient (dlκ )

18.8571 × 10−2 m2 /day

Porosity (φ)

0.35

Water dynamic viscosity (μ)

1.009 × 10−3 Pa s

Hydraulic conductivity (k)

1.0 m/day

Intrinsic permeability (K )

1.18 × 10−12 m2

Seawater density (ρs )

1,025 kg/m3

Fresh water density (ρw )

1,000 kg/m3

Fresh water inflow velocity (Vin )

6.6 × 10−3 m/day

The first step was to run the numerical model to obtain steady-state initial conditions, i.e., before compressed air was injected. In the second step, a transient-state numerical simulation was performed to quantify the efficiency of the compressed air injection method in controlling seawater intrusion.

3 Analysis and Discussion of the Simulation Results 3.1 Steady-State Initial Conditions Before Air Injection The aquifer region was divided into 800 quadrilateral elements, each 5 m long and 5 m thick, so there were 800 nodal points, each located at the center of an element. The freshwater recharge per unit width, Q in = Vin dρw (d is the thickness of the domain (100 m) and ρw is the fresh water mass density), was 7.639×10−3 kg/s, and this was assigned evenly to all of the elements at the left boundary. The whole domain was liquid saturated under the initial steadystate conditions, and the primary variables are pl (the liquid phase pressure), X b (the mass fraction of brine in the liquid phase), X la (the mass fraction of air in the liquid phase), and T (the temperature, in ◦ C). The Dirichlet boundary condition, pl = patm + ρs (100 − z), where

123

264

D. Sun, S. Semprich

X b = 1.0 and X la = 1.0 × 10−10 , was applied at the right-hand (seaward) side boundary. Here, patm is the atmospheric pressure, 1.013 × 105 Pa. No flow across the boundary was considered at the top or bottom. The initial conditions in the aquifer were set as pl = patm + ρw (100 − Z ), X b = 0.0, and X la = 1.0 × 10−10 . We assumed that the aquifer system was isothermal (T = 20 ◦ C). There were no capillary forces under these conditions, and the relative permeability was 1.0. The steady-state initial conditions were obtained by running a transient problem for a total simulation time of t = 6, 000 day, and steady state was reached at t = 5, 164 day. The results for the mass fraction of seawater in the liquid phase, X b = 0.5 isochlor (Fig. 2), matched previously published solutions (Frind 1982; Huyakorn et al. 1987; Cheng et al. 1998; Liu et al. 2001; Abd-Elhamid and Javadi 2011) well. The seawater mass fraction distribution in the liquid phase (X b ) at the steady-state initial conditions is shown in Fig. 3. The groundwater flow velocity distribution at the initial steadystate conditions and the location of zero horizontal velocity are shown in Fig. 4. The outflow region at the seaward side boundary was 48 m ≤ Z ≤ 100 m. The outflow velocity at the top right-hand corner element was 1.59 × 10−6 m/s. Note that the lengths of the arrows in the leftmost and rightmost elements are about half the lengths in the adjacent inner elements, which is caused by an error in the post-processing program.

Fig. 2 The 0.5 isochlor distribution at the steady-state initial conditions

Fig. 3 The seawater mass fraction distribution at the steady-state initial conditions

123

Air Injection to Control Seawater Intrusion

265

Fig. 4 The groundwater flow velocity distribution at the steady-state initial conditions (the solid line is the location of zero horizontal velocity)

Fig. 5 The hydraulic head (in meters of H2 O) distribution at the steady-state initial conditions

Figure 5 shows the hydraulic head distribution in meters of H2 O (H = Z + pl /(ρw g), where pl is the liquid phase (groundwater) pressure and ρw is the freshwater mass density, 1,000 kg/m3 . The flow field and the hydraulic head distribution both show the circulation of seawater, which flows from the sea floor into the aquifer and flushes the aquifer out at the top right-hand boundary. The equipotential lines were not perpendicular to the no-flow boundary because of variations in the liquid density (Fig. 5). Figure 6 shows the groundwater pressure distribution, which will be useful for determining the air pressure applied during air injection. 3.2 Applying Compressed Air Injection The intersection of the X b = 0.01 isochlor with the base of the aquifer was located at about X = 80 m, and the intruded distance from the seaward side boundary was about 120 m, which was assumed to be the maximum distance that the seawater intruded into the aquifer (Fig. 3) and where the row of air injection wells should be placed to mitigate seawater intrusion. Compressed air was introduced into the aquifer through each of the boreholes in the row of air injection wells. A thin steel pipe was installed inside each borehole, the lower part of the pipe being perforated and the top of the pipe being connected to an air pressure line (to control the injected air pressure) supplied by an air compressor (Kramer and Semprich 1989). The maximum groundwater pressure in the air injection section (X = 80.5 m, 15 m ≤ Z ≤

123

266

D. Sun, S. Semprich

Fig. 6 Groundwater pressure (kPa) distribution at the steady-state initial conditions

Fig. 7 The discretization of the aquifer, the air injection section, and the observation points (1–5)

30 m; Fig. 7) was about 830 kPa (Fig. 6). The magnitude of the air pressure applied at the air injection section should be equal to or greater than the maximum groundwater pressure in the aquifer where the air is injected, so that the air can penetrate the aquifer. In our simulation, the injection air pressure was set at 900 kPa. Figure 7 shows the discretization of the aquifer and the elements where air injection was applied. The elements above the injection interval had the same properties as the aquifer. The connection between the element above the injection interval and the upper injection element was removed from the connection list in the TOUGH2 input file to avoid the upward flow of air from the injection grid block without stopping horizontal flow, because the borehole would be a negligible obstacle to groundwater and air flow under realistic 3D conditions. The air injection period was 365 days. The boundary conditions at the right-hand top and bottom boundaries were the same as for the steady-state initial conditions. At the left boundary, pl = ρw g(102.5 − Z ) (equivalent to the sink terms under the steady-state initial conditions), X b = 0.0, and X la = 1.0 × 10−10 . The elements were in liquid–gas two-phase states at constant air pressure, and the primary variables, pg = patm + 9.0 × 105 , X b = 0.0 (the brine mass fraction in the gas phase), Sg + 10 = 10.999, and T = 20 ◦ C, remained constant at the injection grid block elements. The initial aquifer condition was the result of the steady-state initial condition calculations.

123

Air Injection to Control Seawater Intrusion

267

Fig. 8 The movement of the 0.5 isochlor when compressed air was injected, from t = 0 day to t = 365 day

The residual water saturation value Slr was 0.15 for the relative permeability–saturation relationship (Eq. 13) and the capillary pressure–saturation relationship (Eqs. 15, 16). The water–air surface tension, γ , at T = 20 ◦ C, was supplied internally by the TOUGH2 program, and the air entry pressure, p0 , was 25 kPa. Five observation points (labeled 1–5) were selected to assess the pressure changes (Fig. 7). Observation points 1, 2, and 3 had the same X-coordinate and points 3, 4, and 5 had the same Z-coordinate. 3.2.1 Changes in Salt Concentration The temporal displacement process of seawater intrusion can be observed as the transient distribution of the 0.5 isochlor from t = 0 day to t = 365 day for the application of compressed air injection (Fig. 8). The 0.25, 0.5, and 0.75 isochlors migrated with time along the base of the aquifer with the application of compressed air injection (Fig. 9). The 0.25, 0.5, and 0.75 isochlors moved 30.02, 34.73, and 34.96 m, respectively, to the seaward side. The brine component in TOUGH2/EOS7 is defined as concentrated NaCl brine, a more convenient choice than solid NaCl and water components because the brine and water volumes are approximately linearly additive. In the present model, we defined seawater, with a density of 1,025 kg/m3 , as the brine component. TOUGH2 prints the mass balances of all of the mass components within the discretization grid each time the results of the simulation are printed. The mass of intruded seawater (the mass accumulation of brine component) changed with time (Fig. 10), decreasing from 382.88 × 103 to 198.85 × 103 kg over the simulation period. All of the results indicated that the salt concentration decreased during the application of compressed air injection (Figs. 8, 9, 10). 3.2.2 Changes in Fluid Pressures and the Flow Field The gas phase pressure ( pg /1,000) distribution and air flow after 365 days of injecting air are shown in Fig. 11a. The air phase pressure increased significantly near the air injection zone

123

268

D. Sun, S. Semprich

Fig. 9 The 0.25, 0.5, and 0.75 isochlor intersections with the base of the aquifer over time while compressed air was injected

Fig. 10 The mass of intruded seawater in the aquifer versus time while compressed air was injected

and at the top of the aquifer. The injected air tended to flow upward and spread laterally near the top of the aquifer because of the different densities of the air and groundwater. The groundwater pressure ( pl /1,000) distribution (which was similar to the gas phase pressure distribution) and the groundwater flow after 365 days of air injection are shown in Fig. 11b. The initial seawater circulation had disappeared at that time, and the groundwater had been pushed toward the left and right boundaries of the aquifer from the injection zone, and the maximum groundwater velocity was about 0.4 × 10−5 m/s near the air injection zone and 0.1 × 10−5 m/s near the boundaries.

123

Air Injection to Control Seawater Intrusion

269

Fig. 11 The a groundwater pressure (kPa) distribution and flow and b gas phase pressure (kPa) distribution and flow when compressed air had been injected for 365 days

The air pressure increased with time in similar ways at points 1–5 (Fig. 12), increasing rapidly to a maximum, then decreasing rapidly from the maximum before decreasing slowly until a quasi-steady-state pressure was reached. The peak and quasi-steady-state pressures decreased with increasing distance from the air injection zone. The peak pressure was reached because of the effect of the relative air permeability. When the injected air entered the confined aquifer, the initial air saturation and the relative gas phase permeability (which is related to the initial air saturation) were very low, and overcoming the low relative gas phase permeability would require a high pressure gradient. Higher air saturation and relative gas phase permeability were achieved as the displacement of water continued, so a lower pressure gradient would be required. 3.2.3 Changes in the Air Injection Rate The mass flux of air, G (kg/s), injected into the aquifer through the injection section was calculated using the model, and the air density at atmospheric pressure was calculated using the ideal gas law, allowing the volume flux of air q (m3 /s) at atmospheric pressure to be calculated. The air injection rate through the air injection section increased from q = 0.004 m3 /s at the beginning of the simulation to q = 0.092 m3 /s at t = 365 day (Fig. 13). The air injection rate first increased rapidly and then increased more slowly because the injected air

123

270

D. Sun, S. Semprich

Fig. 12 Changes in air pressure with time at observation points 1–5 in the aquifer while air was injected

Fig. 13 Air injection rate during the air injection simulation

plume reached the left and right boundaries, allowing air to flow out of the modeled section and a steady state to be approached. The fate of the lateral spreading plume is rather important because it can affect underground structures in the area, such as other wells (in use or abandoned). Water is usually extracted from the bottom of an aquifer when people use the aquifer as a water source. If underground structures are present, the new air-injection-based equilibrium in the aquifer will be disturbed and the air injection rate will increase. Wells for water extraction should, therefore, be located outside of the range of the spreading injected air.

123

Air Injection to Control Seawater Intrusion

271

3.2.4 Comparison with the Freshwater Injection Method The freshwater injection method was also simulated using similar settings to those used for the air injection simulation. The discretization of the aquifer, initial conditions, and boundary conditions was the same for the freshwater injection model as for the air injection model. However, the elements of the injection grid block (the same as was used for the air injection model, shown in Fig. 7) contained a single-phase state (i.e., liquid only), and the primary variables pl = patm + 9.0 × 105 Pa, X b = 0.0, X la = 1.0 × 10−10 (the mass fraction of air in the liquid phase), and T = 20 ◦ C remained constant, representing freshwater injection at the same injecting pressure as used for air in the air injection simulation (900 kPa). The freshwater injection period was 365 days. The temporal displacement of intruding seawater was observed as the transient distribution of the 0.5 isochlor from t = 0 day to t = 365 day while freshwater was being injected (Fig. 14). The 0.5 isochlor migrated, with time, along the base of the aquifer as freshwater was injected, and this migration is shown for both freshwater and compressed air injection in Fig. 15. Air injection and the freshwater injection caused the 0.5 isochlor to move 34.73 and 63.91 m toward the sea, respectively, by t = 365 day. The mass of intruded seawater changed with the time freshwater was injected, and this, and the same effect from compressed air injection, is shown in Fig. 16. The mass of intruded seawater decreased from 382.88×103 to 57.30 × 103 kg over the time freshwater was injected. All of the results indicate that the freshwater injection method is more efficient than the air injection method in preventing seawater intrusion (Figs. 14, 15, 16). However, freshwater resources are very valuable, especially in areas with limited freshwater resources, whereas air is abundant, readily available and free of charge, so injecting air to mitigate seawater intrusion does have great potential benefits. The groundwater pressure ( pl /1,000) distribution and groundwater flow after 365 days of freshwater injection are shown in Fig. 17. The groundwater pressure increased significantly near the freshwater injection zone and increased more slowly further away from the freshwater injection zone. Groundwater was pushed from the injection zone toward the left and right boundaries of the aquifer, and the maximum groundwater velocity was about 6.5 × 10−5 m/s

Fig. 14 The position of the 0.5 isochlor when freshwater was injected, from t = 0 day to t = 365 days

123

272

D. Sun, S. Semprich

Fig. 15 Movement of the 0.5 isochlor with time along the base of the aquifer when air was injected and when freshwater was injected

Fig. 16 The mass of intruded seawater in the aquifer over time when air was injected and when freshwater was injected

near the air injection zone and 0.34 × 10−5 m/s near the boundaries. The induced laterally diverted groundwater flow was faster using freshwater injection than using air injection (Fig. 11b). The freshwater injection rate through the injection section was almost constant, at 0.3 kg/s, throughout the freshwater injection period (Fig. 18), and comparing this with the air injection rate (Fig. 13) the injection-based equilibrium was approached more quickly using freshwater injection than using air injection.

123

Air Injection to Control Seawater Intrusion

273

Fig. 17 The groundwater pressure (kPa) distribution and groundwater flow when freshwater had been injected for 365 days

Fig. 18 The freshwater injection rate over the freshwater injection period

4 Air Injection Method Sensitivity Analysis 4.1 Sensitivity to the Applied Air Pressure Figure 19 shows how the 0.5 isochlor moved along the base of the aquifer with time at different applied air pressures. Applied air pressures of 850, 900, and 950 kPa caused the 0.5 isochlor to be displaced toward the sea by 21.69, 34.73, and 42.50 m, respectively, after 365 days. The air pressures at observation point 3 over the simulation period, at the different applied air pressures, are shown in Fig. 20. The air injection rates during the air injection period, at the different applied air pressures, are shown in Fig. 21. The applied air pressure was directly related to the amount of air injected into the aquifer. The pressure difference between the air injection section and the aquifer increased when the

123

274

D. Sun, S. Semprich

Fig. 19 The position at which the 0.5 isochlor intersected the base of the aquifer at different applied air pressures

Fig. 20 Air pressure versus time at observation point 3, at different applied air pressures

applied air pressure increased, leading to more compressed air being injected into the aquifer and resulting in an increase in the gas phase pressure in the aquifer. It was, therefore, more efficient in preventing seawater intrusion if higher air pressure was applied, although the cost of the air injection method would be increased correspondingly because of the higher air injection rate. 4.2 Sensitivity to the Air Entry Pressure Figure 22 shows how the 0.5 isochlor moved along the base of the aquifer with time at different air entry pressures. The seaward displacement of the 0.5 isochlor at the air entry pressures of

123

Air Injection to Control Seawater Intrusion

275

Fig. 21 The air injection rate versus time at different applied air pressures

Fig. 22 The position at which the 0.5 isochlor intersected the base of the aquifer at different air entry pressures

20, 25, and 30 kPa was almost the same. The air pressures at observation point 3 over time, at the different air entry pressures, are shown in Fig. 23. There were small differences in the quasi-steady-state air pressures at observation point 3 for the different air entry pressures, a lower air entry pressure giving a slightly higher air pressure at observation point 3. The air injection rates during the air injection period, at the different air entry pressures, are shown in Fig. 24. There were clear differences between the air injection rates for the different air entry pressures, more injected air being required at lower air entry pressures. Air pressure higher than the hydrostatic pressure is required for the injected air to penetrate the aquifer. The excess air pressure required is commonly known as the formation air entry pressure, and it is the minimum capillary pressure needed to cause the air to flow into the

123

276

D. Sun, S. Semprich

Fig. 23 Air pressure versus time at observation point 3 at different air entry pressures

Fig. 24 The air injection rate versus time at different air entry pressures

saturated porous medium. A lower air entry pressure means that it is easier to inject air into the aquifer. 5 Summary and Conclusions The aim of this study was to assess the efficiency at which injecting compressed air could control seawater intrusion into a confined coastal aquifer, with a view to exploring whether this might be a relatively simple and effective way of combatting seawater intrusion into fresh groundwater systems in areas with limited freshwater resources. A simplified hypothetical confined coastal aquifer (Henry’s problem) was used, and idealized 2D settings were chosen

123

Air Injection to Control Seawater Intrusion

277

to represent the air injection system, the air being injected through a row of boreholes situated an appropriate distance from each other and parallel to the coast. The results showed that air injection could effectively mitigate seawater intrusion. We compared the air injection results with a freshwater injection simulation using similar settings and found that freshwater injection was more efficient than air injection in preventing seawater intrusion. However, air injection may be an attractive method for controlling seawater intrusion in areas that have limited freshwater resources, or it may allow the efficiency of seawater intrusion mitigation to be improved when combined with freshwater injection. The proposed air injection method has some potential drawbacks and limitations. Injecting air into an unconfined aquifer would give completely different (and less effective) results to those presented here because the injected air entering the saturated aquifer will rise because of its buoyancy relative to the groundwater, then flow through the saturated zone toward the unsaturated zone in the unconfined aquifer. This process will result in large air losses and a lower pressure gradient for combatting seawater intrusion. Detailed characteristics of the air injection method need to be simulated using a 3D model, including injection strategies for multiple wells (e.g., the number and placement of injection wells and the vertical air injection intervals), the fate of injected air migrating to the aquifer top, the effect of lower permeability heterogeneities in sub-horizontal layers (below which the uprising air would accumulate), the effects of a sloping “cap-rock,” the effects of the relative permeability and capillary pressure curves, the overall modification of groundwater flow (the groundwater flow would be diverted laterally because of pressure building up, caused by the air injection), the effects on wells used for groundwater extraction upstream of the air injection wells, contaminant risks (because air injection may cause dissolved/adsorbed volatile contaminants to accumulate in the air cap beneath the top confining layer in a polluted aquifer), the stability of the top confining layer beneath which air pressure builds up, and the possibility of air leakage from existing boreholes. Despite these potential problems, and considering that freshwater resources are very valuable, whereas air is abundant, readily available, and free of charge, the air injection method for mitigating seawater intrusion has the potential to be useful for preventing seawater intrusion in regions with limited freshwater resources. Acknowledgments Support for this research from the National Nature Science Foundation of China (Grant Nos. 51179118 and 50809044) and the Science Fund for Creative Research Groups of the National Natural Science Foundation of China (Grant No. 51021004) is greatly appreciated.

References Abd-Elhamid, H.F.: A simulation–optimization model to study the control of seawater intrusion in coastal aquifers. Ph.D. thesis, University of Exeter, UK (2010) Abd-Elhamid, H.F., Javadi, A.A.: A density-dependant finite element model for analysis of saltwater intrusion in coastal aquifers. J. Hydrol. 401, 259–271 (2011) Bear, J., Cheng, A.H.-D., Sorek, S., Ouazar, D., Herrera, I.: Seawater Intrusion in Coastal Aquifers—Concepts, Methods and Practices. Kluwer, Dordrecht (1999) Brooks, M.C., Wise, W.R., Annable, M.D.: Fundamental changes in in situ air sparging how patterns. Ground Water Monit. Rem. 19, 105–113 (1999) Cheng, J.R., Strobl, R.O., Yeh, G.T., Lin, H.C., Choi, W.H.: Modeling of 2D density-dependent flow and transport in the subsurface. J. Hydrol. Eng. 3, 248–257 (1998) Dhar, A., Datta, B.: Saltwater intrusion management of coastal aquifers. I: linked simulation-optimization. J. Hydrol. Eng. 14, 1263–1272 (2009) Dror, I., Berkowitz, B., Gorelick, S.M.: Effects of air injection on flow through porous media: observations and analyses of laboratory-scale processes. Water Resour. Res. 40, W09203 (2004). doi:10.1029/ 2003WR002960

123

278

D. Sun, S. Semprich

Fatt, I., Klikoff, W.A.: Effect of fractional wettability on multiphase flow through porous media. Trans. AIME 216, 426–432 (1959) Frind, E.O.: Simulation of long-term transient density-dependent transport in groundwater. Adv. Water Resour. 5, 73–88 (1982) Henry, H.R.: Effects of dispersion on salt encroachment in coastal aquifers. In: Cooper Jr, H.H., Kohout, F.A., Henry, H.R., Glover, R.E. (eds.) Sea Water in Coastal Aquifers, Geological Survey Water-Supply Paper 1613-C, pp. C71–C84. U.S. Geological Survey, Washington, DC (1964a) Henry, H.R.: Interfaces between salt water and fresh water in coastal aquifers. In: Cooper Jr, H.H., Henry, H.R., Glover, R.E. (eds.) Sea Water in Coastal Aquifers, Geological Survey Water-Supply Paper 1613-C, pp. C35–C70. U.S. Geological Survey, Washington, DC (1964b) Huyakorn, P.S., Andersen, P.F., Mercer, J.W., White Jr, H.O.: Saltwater intrusion in aquifers: development and testing of a three-dimensional finite element model. Water Resour. Res. 23, 293–312 (1987) Javadi, A.A., Abd-Elhamid, H.F., Farmani, R.: A simulation–optimization model to control seawater intrusion in coastal aquifers using abstraction/recharge wells. Int. J. Numer. Anal. Methods 36, 1757–1779 (2012) Jorreto, S., Pulido-Bosch, A., Gisbert, J., Sanchez-Martos, F., Frances, I.: The fresh water-seawater contact in coastal aquifers supporting intensive pumped seawater extractions: a case study. Comptes Rendus Geosci. 341, 993–1002 (2009) Kacimov, A.R., Sherif, M.M., Perret, J.S., Al-Mushikhi, A.: Control of sea-water intrusion by salt-water pumping: coast of oman. Hydrogeol. J. 17, 541–558 (2009) Karasaki, K., Ito., K, Maekawa, K.: Simulation of salt water intrusion. In: Proceedings of the TOUGH Symposium 2006, Lawrence Berkeley National Laboratory, Berkeley, California, 15–17 May 2006 Kiuchi, M., Kaspar, T.C., Horton, R.: Managing soil-water and chemical transport with subsurface flow barriers. Soil Sci. Soc. Am. J. 60, 880–887 (1996) Kramer, J., Semprich, S.: Erfahrungen über Druckluftverbrauch bei der Spritzbetonbauweise. In: Taschenbuch für den Tunnelbau, Deutsche Gesellschaft für Erd- und Grundbau ( in German), pp. 91–153. Essen, Germany, (1989) Lee, C.H., Cheng, RT-Sh: On seawater encroachment in coastal aquifers. Water Resour. Res. 10(5), 1039–1043 (1974) Leverett, M.C.: Capillary behavior in porous solids. Trans. AIME 142, 152–169 (1941) Liu, F., Turner, I., Anh, V.: A finite volume unstructured mesh method for modeling saltwater intrusion into aquifer systems. In: First International Conference on Saltwater Intrusion and Coastal Aquifers— Monitoring, Modeling, and Management. Essaouira, Morocco, 23–25 April 2001 Meiri, D.: Two-phase flow simulation of air storage in an aquifer. Water Resour. Res. 17, 1360–1366 (1981) Moridis, G.J., Finsterle, S., Heiser, J.: Evaluation of alternative designs for an injectable subsurface barrier at the Brookhaven National Laboratory site, Long Island. New York. Water Resour. Res. 35, 2937–2953 (1999) Moridis, G.J., Reddell, D.L.: Secondary water recovery by air injection: 1. The concept and the mathematical and numerical model. Water Resour. Res. 27, 2337–2352 (1991) Newport, B.D.: Salt water intrusion in the United States. Special Report No. EPA-600/8-77-011, U.S. Environmental Protection Agency, Ada, Oklahoma (1977) Oldenburg, C.M., Pruess, K.: Dispersive transport dynamics in a strongly coupled groundwater-brine flow system. Water Resour. Res. 31, 289–302 (1995) Pinder, G.F., Cooper Jr, H.H.: A numerical technique for calculating the transient position of the saltwater front. Water Resour. Res. 6, 875–882 (1970) Pruess, K.: EOS7, an equation-of-state module for the TOUGH2 simulator for two-phase flow of saline water and air. Technical report LBL-31114, Lawrence Berkeley Laboratory, Berkeley, California (1991) Rao, D.: Gas injection EOR-A new meaning in the new millennium. J. Can. Petroleum Technol. 40, 11–19 (2001) Rastogi, A.K., Choi, G.W., Ukarande, S.K.: Diffuse interface model to prevent ingress of seawater in multilayer coastal aquifers. J. Spatial Hydrol. 4, 1–31 (2004) Reddy, K.R., Adams, J.A.: Effects of soil heterogeneity on airflow patterns and hydrocarbon removal during in situ air sparging. J. Geotech. Geoenviron. 127, 234–247 (2001) Segol, G., Pinder, G.F., Gray, W.G.: A Galerkin finite element technique for calculating the transient position of the saltwater front. Water Resour. Res. 11, 343–347 (1975) Sherif, M.M., Hamza, K.I.: Mitigation of seawater intrusion by pumping brackish water. Transp. Porous Media 43, 29–44 (2001) Todd, D.K.: Salt-water intrusion and its control. J. AWWA 66, 180–187 (1974) Voss, C.I., Souza, W.R.: Variable density flow and solute transport simulations of regional aquifers containing a narrow freshwater-saltwater transition zone. Water Resour. Res. 23, 1851–1866 (1987)

123