Provision of Desalinated Irrigation Water by the Desalination of Groundwater within a Saline Aquifer

Abstract

Irrigated land accounts for 70% of global water usage and 30% of global agricultural production. Forty percent of this water is derived from groundwater. Approximately 20%–30% of the groundwater sources are saline and 20%–50% of global irrigation water is salinized. Salinization reduces crop yields and the number of crop varieties which can be grown on an arable holding. Structured ZVI (zero valent iron, Fe⁰ pellets desalinate water by storing the removed ions as halite (NaCl) within their porosity. This allows an “Aquifer Treatment Zone” to be created within an aquifer, (penetrated by a number of wells (containing ZVI pellets)). This zone is used to supply partially desalinated water directly from a saline aquifer. A modeled reconfigured aquifer producing a continuous flow (e.g., 20 m³/day, 7300 m³/a) of partially desalinated irrigation water is used to illustrate the impact of porosity, permeability, aquifer heterogeneity, abstraction rate, Aquifer Treatment Zone size, aquifer thickness, optional reinjection, leakage and flow by-pass on the product water salinity. This desalination approach has no operating costs (other than abstraction costs (and ZVI regeneration)) and may potentially be able to deliver a continuous flow of partially desalinated water (30%–80% NaCl reduction) for $0.05–0.5/m³.

1. Introduction

Arable agriculture accounts for more than 70% of global water usage [1]. Total anthropogenic water usage is currently estimated at 10,688 km³·a⁻¹ [2]. About 4000 km³·a⁻¹ are abstracted from the riparian and groundwater environment [3]. Irrigation usage accounts for about 70% of this abstracted water [1,3]. Sixty-two percent of water withdrawals for irrigation [4] return to the groundwater and riparian system through a combination of evaporation (23%–30%), infiltration (7%–16%) and overland flow (26%–27%) [4,5]. Seventy-one percent of the global irrigated area is located in areas where the annual depletion of the watershed can exceed 75% [4].

About 24% of the total global harvested cropland is irrigated [6]. This irrigated cropland accounts for more than 40% of the global cereal yield [6] and more than 30% of global agricultural production [7]. Global cereal production would decrease by 20% without irrigation [3]. Global crop production will be required to more than double by 2050 in order to sustain the anticipated growth in global population [6]. The vast majority of this increase in agricultural production is expected to result from an increase in the amount of irrigated land [6] coupled with a more efficient use of irrigation [1,2,3,4,5,6,7,8].

The relative balance of irrigation water abstracted from the riparian system and irrigation water abstracted from groundwater varies regionally [8]. About 40% of global irrigation water is abstracted from shallow groundwater (aquifers) [8]. Locally, this can exceed 70% (e.g., India) [8]. Between 20% and 30% of global groundwater (320–480 km³·a⁻¹) abstracted for irrigation is saline [9].

In some countries (e.g., Uzbekistan) irrigated land may account for less than 15% of arable production, but more than 90% of the arable crop [10]. The groundwater used for irrigation (average 3.7 km³·a⁻¹) has salinities within the range 2.25–3.75 g·L⁻¹ [10].

At least 20% of the global irrigated crop land is adversely affected by salinization [11,12,13,14,15,16,17]. Some estimates indicate that the combination of irrigation and leaching may result in more than 50% of the global irrigated land being affected by salinization [18]. Globally, salinization is associated with between 560 and 1400 km³·a⁻¹ of irrigation water [11,12,13,14,15,16,17,18]. About 405 km² of irrigated land is used for greenhouses [19]. The groundwater used to irrigate these greenhouses is saline in many coastal areas [19].

Salinized irrigation water reduces the associated potential crop yield by between 10% and 90% (relative to irrigation with freshwater) [16,18,19,20,21,22,23,24,25]. Salinization adversely affects the long term viability of the agricultural land. The associated saline irrigation runoff adversely affects: (i) the downstream riparian system; and (ii) the underlying aquifers receiving salinized infiltration water [1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25].

It is estimated that the value of lost agricultural production due to salinization exceeds $27.3 billion/a [26]. The lost value associated with salinization of irrigation water also affects the broader local economy [27]. It is, therefore, reasonable to assume that any strategy, which can cost effectively help stabilize, or reverse, the adverse effects of salinization, will have a major economic impact on the local economy.

The amount of water required for irrigation, varies with crop and region [8]. It is also a function of the net amount of water received by precipitation and the associated evaporation [1,2,3,4,5]. Irrigated water requirements decrease as the irrigation type switches from flood, to drip, to sprinkler, and to more sophisticated irrigation technologies [8].

One possible solution, which can be used to combat salinization, is to use desalinated water for irrigation [28]. Desalinated irrigation water is currently used (on a small scale) in Spain, Israel, Kuwait, Saudi Arabia, UAE, Bahrain, Qatar, Chile, China, Australia [28,29]. The unsubsidized, full cycle cost of providing desalinated water from a modern large scale (>100,000 m³/day) desalination plant varies with installation date and region. It is currently more than $3.5/m³ for most (>100,000 m³/day) plants installed within the last 5 years [16,17]. This is addressed further in Section 9.

In many areas, which use desalinated irrigation water, the delivered water price is heavily subsidized (for strategic and political reasons) and may fall within the range $0.2 to $0.7/m³ [28,29].

1.1. Impact of Irrigation Water Salinity on Crop Yield

Desalination (or partial desalination) of irrigation water, prior to use, is expected to increase the yield of a specific crop [16,18,19,20,21,22,23,24,25]. With any desalination technology used for irrigation, there are two important questions:

(i)

Is the technology likely to produce an adequate amount of water for a reasonable cost over a reasonable timeframe?

(ii)

Is the application, by irrigation, of the desalinated (or partially desalinated) water, likely to make a significant impact on the revenue and profitability of the agricultural holding?

These two questions can be considered with reference to a specific crop. In a freshwater irrigation environment, the estimated crop yield can be approximated as [30]:

Yield (t/ha), Y = (a₁b₁c₁/d₁)(abcde)

(1)

a₁ = plants per hectare (10,000 m²) (this is a function of the number of plants per meter length on a row and the row spacing, m); b₁ = seed pods (or grain heads or fruit) per plant; c₁ = fruit or seeds per pod/grain head; d₁ = seeds or fruit per ton; a = fractional yield change due to fertilizer application; b = fractional yield change due to fungicide application; c = fractional yield change due to pesticide application; d = fractional yield change due to herbicide application; e = fractional yield change due to bactericide application. The fractional yield change is calculated as: (Yield with the application/Yield without the application).

Yield is strongly influenced by the planting strategy, variety, climatic variation, farming practice, fertilizers applied, local soil composition, irrigation water composition and irrigation strategy. The impact of salinity is to reduce the expected crop yield. An indicative relationship for crop yield decrement as a function of salinity is [16,20,21,22]:

Salinity Adjusted Yield (t/ha), Y_S = Y (1 − (_c1s(Salinity, g/L) − c_2s)), where 0 ≤ Y_S ≤ Y

(2)

c_1s and c_2s are regression defined constants (see references [16,20,21,22] for the value of these constants associated with more than 60 different crops). By definition, the expression (_c1a(Salinity, g/L) − c_2a) has the limit values of 0.0 and 1.0. Each of the variables in Equations (1) and (2) can be described by a distribution. The amount of irrigation water required, is location and crop specific, and may fall within the range 0 to 10,000 m³/ha/a [31,32].

1.2. Maximum Sustainable Desalination Cost for Saline Irrigation Water

Most irrigated crops have a low sales value ($/ha, or $/t) and are unable to economically sustain a high desalination cost ($/m³) for irrigated water. The unsubsidized, delivered cost of potable quality desalinated water (from a conventional desalination plant) may fall within the range $3 to $100/m³ [16,17]. The incremental sales value (excluding operating, harvesting and storage costs) added by the use of desalinated water will be site specific. Historically, the high cost of providing desalinated water has prevented its widespread use for irrigation in areas, which currently use salinized irrigation water.

The maximum sustainable cost (M_c, $/m³) of desalination for irrigation water for a specific crop, or agricultural holding, can be estimated as:

M_c, $/m³ = (((Y_S2 − Y_S1)R₁) − O₁ − F₁ − P₁)/W₁,

(3)

Y_S1 = Crop yield using salinized water, t; Y_S2 = Crop yield using partially desalinated water, t; R₁ = Sale price of crop, $/t; O₁ = Incremental increase in operating costs and infra-structure costs associated with the increased yield, $; F₁= Incremental increase in financing costs associated with the increased yield, $; P₁ = Profit required by the agricultural holding on the increased yield, $; W₁ = amount of partially desalinated water which is required to achieve the increased yield, m³.

Irrigation using partially desalinated water is only economically viable if M_c is greater than the actual cost of supplying the partially desalinated irrigation water (A_c, $/m³), i.e., ((M_c − A_c) > 0).

1.3. Zero Valent Iron

It has recently (2010) been discovered that Zero Valent Iron (Fe⁰, ZVI) could be used to partially desalinate water [16,17,33,34,35,36,37,38]. The potential, unsubsidized, delivery cost of this partially desalinated water may be less than $0.1/m³ [16,17].

The two principal ZVI desalination approaches are:

(a)

placement of structured ZVI in pellets. The pellets are then placed in a water body and left for a period of time [16,17];

(b)

placement of structured ZVI in cartridges which are attached to a reactor incorporating fluid recycle [16,17,36].

This study investigates whether this trial technology could be transferred into the subsurface to produce an Aquifer Treatment Zone within a saline aquifer. The Aquifer Treatment Zone is created by the placement of removable ZVI desalination pellets (or cartridges), in infiltration devices (or wells) intersecting a saline aquifer. The abstracted water is used to provide partially desalinated irrigation water.

1.4. Definition and Construction of an Aquifer Treatment Zone

Trial data, from more than 300 ZVI desalination trials (0.0002 to 0.8 m³/batch) [16,17], are integrated with standard hydrological and chemical process engineering models. This integration is designed to identify the principal technical issues associated with a commercial scale-up (within an aquifer). The target proposed commercial scale Aquifer Treatment Zone will produce 1 to 100 m³/day of partially desalinated irrigation water.

The batch desalination trials used a diffusion process (with and without a water recycle) where structured ZVI catalysts (contained in pellets and cartridges) were placed in saline water. The associated methodology and desalination trial data have been placed in the Appendix A, Appendix B, Appendix C, Appendix D, Appendix E, Appendix F and Appendix G.

This study has the following Appendices:

• Appendix A: Trial Materials and Kinetic Methodology; Equations (A1)–(A4); Figure A1 and Figure A2;
• Appendix B: Kinetic Data associated ST series desalination pellets; Equation (B1); Figure B1, Figure B2 and Figure B3;
• Appendix C: Equations and models which are required to provide a control data set for the aquifer and to provide effective monitoring of a proposed Aquifer Treatment Zone following installation; Equations (C1)–(C36b); Figure C1, Figure C2, Figure C3, Figure C4, Figure C5, Figure C6 and Figure C7;
• Appendix D: Classifying the Potential Aquifer Resource;
• Appendix E: Calculating and modeling the dimensions associated with a proposed Aquifer Treatment Zone. Equations (E1)–(E5);

  Table E1. Trial results associated with the operation of semi-closed Aquifer Treatment Zone with Recycle (Figure 20). The recycle ratio, R_rc = y (m³/day)/m (m³/day). R_rt = average residence time (t_d) of the water in the reaction environment. Cartridge casing = MDPE. Each trial example used a different Type B catalysts (Table F1). A single catalyst charge was used in each trial group. Optimised operation assumes that t_d is matched to the onset of B_S. This reduces R_rt (associated with each catalyst) into the range 3 to 12 h.

  Trial Group	Reactor Size, L	Pw, g Fe0/L	Sub-Optimised Rrc	Optimised Rrc	A0, g/L	At = n, g/L	Rrt, h	Log10 (kobserved)	Log10 (kactual)	Water Volume Treated, L	Scale-Up Required to Achieve 20 m3/day	Scale-Up Required to Achieve 20 m3/day with Optimised Operation
  RC1	5.8	23.28	3.60	2.48	4.21	3.03	5.80	−4.80	−6.17	63.8	833.3	574.7
  RC2	240.0	0.50	11.08	3.00	2.97	2.14	22.15	−5.39	−5.09	12000.0	76.9	20.8
  RC3	8.0	28.00	20.25	1.35	2.65	0.79	45.00	−5.13	−6.57	32.0	4687.5	312.5
  RC4	5.8	8.79	302.28	7.45	4.25	2.81	487.00	−6.63	−7.57	34.8	69,971.3	1724.1
  RC5	5.8	4.80	11.75	3.10	36.17	25.71	18.93	−5.30	−5.98	92.8	2719.8	718.4
  RC6	5.8	0.86	12.22	3.72	15.66	8.67	19.68	−5.08	−5.01	75.4	2827.6	862.1
  RC7	5.8	4.66	9.62	3.10	8.84	7.04	15.50	−5.39	−6.06	29.0	2227.0	718.4
  RC8	5.8	0.86	2.84	1.86	6.46	5.30	4.58	−4.92	−4.86	17.4	658.0	431.0
  RC9	800.0	0.07	3.21	0.45	2.10	1.28	42.74	−5.49	−4.34	9600.0	44.5	6.3

  ;
• Appendix F: Concepts and Models Associated with a potential Recycle Strategy; Equations (F1)–(F3); Figure F1, Figure F2, Figure F3, Figure F4 and Figure F5;

  Table F1. Expected Performance of a semi-confined schematic Aquifer Treatment Zone containing 4000 m³ of water. The data values (and scaled volumes) are based on trial observations (Figure B2 and Figure C5, Table E1). The aquifer volumes and feed water salinities have been normalized to demonstrate the operational differences between the catalysts. The duration of all the sub-optimized trials was extended beyond the initial appearance of B_S. Optimized irrigation water production rates assume that t_d approximates to the start of B_S. Catalyst ST* is a Type A Catalyst. Catalysts C to K are Type B Catalysts. ZVI required is based on trial data and has not been optimized.

  Trial Group	Aquifer Volume, m3	Irrigation Water (without Optimisation), m3·day−1	Recycle Water, m3·day−1	Optimised Irrigation Water, m3·day−1	Feed Water Salinity, g·L−1	Product Water Salinity, g·L−1	Mean Desalination	Mean Na+ Removal	ZVI Required, t	Catalyst
  ST	4000	20	0	20	4.00	1.84	54.0%	54.0%	120	ST
  RC1	4000	16,552	59,586	24,000	4.00	2.88	28.0%		93.12	C
  RC1a	4000	18,113	65,208		4.00	2.88	28.1%	38.9%	93.12
  RC2	4000	4334	48,000	16,000	4.00	2.88	27.9%		2.00	D
  RC3	4000	2133	43,200	32,000	4.00	1.19	70.2%		112.00	E
  RC4	4000	197	59,586	8,000	4.00	2.64	33.9%		35.16	F
  RC5	4000	5071	59,586	19,200	4.00	2.84	28.9%		19.20	G
  RC6	4000	4878	59,586	16,000	4.00	2.21	44.6%		3.44	H
  RC7	4000	6194	59,586	19,200	4.00	3.19	20.4%		18.64	I
  RC8	4000	20,961	59,586	32,000	4.00	3.28	18.0%		3.44	J
  RC9	4000	2246	7,200	16,000	4.00	2.43	39.3%		0.28	K
  RC9a	4000	9600	7,200		4.00	2.41	39.6%	50.6%	0.28

  ;
• Appendix G: Modeling a potential Static Strategy when the Aquifer is heterogenous; Equations (G1)–(G4); Figure G1.

A list of abbreviations and symbols used in this study is placed after the Appendices.

Conversion of an aquifer to an Aquifer Treatment Zone requires sequential consideration of three issues: (i) the nature of the desalination product and its potential recovery from the aquifer; (ii) the front-end engineering requirements of a specific site; and (iii) an assessment of the costs associated with the installed Aquifer Treatment Zone.

2. Desalination Product

The different desalination catalysts used in this study (ST, MT, C to K) are described further in Section 7.3, and their performance is detailed in Appendix B, Appendix E and Appendix F. ST, MT, D(i) are identifiers used for different Type A Catalysts. C, D(ii), E to K are identifiers used for Different Type B Catalysts. Notations such as ST3b refer to a trial using the ST catalyst, Trial Group Feed Water Salinity Grouping Identifier = 3 (e.g., 11.05 g NaCl/L), Statistical Trial identifier = b (e.g., Trial 2).

The desalination ZVI pellet trials [16] established that the NaCl was removed from the water (Figure B1, Figure B2 and Figure B3). The removed NaCl was held in ZVI pellets in some form. A number of pellets from the ST Trial Group [16] were allowed to dry in air (at a temperature of 0 to 20 °C) before being cut in half. The surfaces were then polished prior to examination using reflectance microscopy.

The petrography of the desalination pellets defines the desalination process. The petrography and its implications for the design of an Aquifer Treatment Zone are discussed in detail in Section 3, Section 4 and Section 5. Section 4 and Section 5 define the adsorption-desorption mechanisms. These mechanisms are used in Appendix A, Appendix B, Appendix C, Appendix D, Appendix E, Appendix F and Appendix G to demonstrate how desalination can be modeled and monitored within an aquifer.

The petrography is integrated with the front-end engineering associated with installing and operating an Aquifer Treatment Zone in Section 6 and Section 7.

3. Desalination Product Petrography

3.1. Hygroscopic Nature of Cut Pellets

Freshly cut surfaces of the ZVI (ST Trial Group) can develop hygroscopic brine droplets on their surface (see Figure 1). Ion analyses established the presence of Na⁺ and Cl⁻ ions in the hygroscopic and deliquescence brine droplets. NaCl is hygroscopic and has a well-defined deliquescence point (e.g., [39,40]).

3.2. Rough Surface of Cut Pellets Following Desalination

The freshly cut pellet surface (before polishing) is typically covered in white crystalline material (halite [41,42,43,44,45,46,47,48]), e.g., Figure 2.

3.3. Polished Surface of Cut Pellets Following Desalination

The polished ZVI surfaces established that the original porosity had been completely infilled with halite, in some pellets. e.g., Figure 3. Large open pores are infilled with halite, which has grown into the pore from the pore sides, e.g., Figure 4.

3.4. Material Redistribution within the Pellets during Desalination

The porosity of the ZVI is physically expanded by the growth of the NaCl crystals (<1 micron) to form sheets of NaCl within the ZVI, e.g., Figure 5. The borders of the NaCl sheets show that the ZVI matrix may be replaced locally by NaCl to create FeO_x(OH)_y regoliths (e.g., Figure 5 and Figure 6). The halite can contain regoliths of copper associated with the pellet shell, e.g., Figure 6. These regoliths indicate that the copper within the pellet can be affected by dissolution, during the desalination process.

3.5. Reprecipitation of Copper within the Pellets during Desalination

The copper contained within the pellet (Figure 7) can be dissolved and recrystallized as atacamite (Cu₂(OH)₃Cl) [41] and tenorite (CuO) [41] during desalination. The Eh-pH regime (Figure 7) (and Cl⁻ ion regime) associated with these pellets is consistent [50,51,52] with both tenorite and atacamite precipitation (Figure 8 and Figure 9).

3.6. Method of Halite Infill of Porosity within the Pellets during Desalination

The halite fills the pre-existing pores within the ZVI pellets using crystal masses which grow into the pore from the pore edge (Figure 10 and Figure 11).

3.7. Morphology of Halite Infill of Porosity within the Pellets during Desalination

The halite can be present as rounded nodules contained within purple ferrate (Na₂FeO_x, Na_zFeO_x(OH)_y) nodules (and within goethite) (Figure 12). These ferrate nodules can form regoliths (or authigenic nodules) within the halite, or can form as authigenic nodules within the FeO_x(OH)_y matrix (Figure 12).

3.8. Morphology of Halite Infill of Porosity

The NaCl which is concentrated in the pore waters within the ZVI pellets during desalination will go through a number of phases [54]:

L→ L + α → α

(4)

Where L = liquid phase, and α = solid (crystallized) phase. The pore water chemistry during desalination is complex and the relative ratio, or status, of these three potential phases may be influenced by temperature, pressure, ion concentrations, pH and Eh. A simple model of the form expressed by Equation (4) will result in halite formation from a mixture of L + α. Halite will form as either nodular halite (e.g., Figure 12) or by the crystalline infill of porosity by halite.

The iron (or copper) phase can be designated a β phase. In this instance, the following pore-water transitions are possible [54]:

L _{(t = 1)} → L_{(t = 2)} + α,

(5)

L _{(t = 2)} → β + α, or

(6)

L _{(t = 2)} → β,

(7)

L _{(t = 1)} = pore water composition at time, t =1; L _{(t = 2)} = pore water composition at time, t = 2. The final product structure will therefore comprise: α + (β + α). The α mineral will appear as grains of α mineral (e.g., authigenic halite). The β mineral will appear as grains of β mineral (e.g., authigenic FeO_x(OH)_y(H₂O)_z). This allows both phases to precipitate within the pellets as authigenic minerals during desalination.

Alternatively, when the original α mineral is a Fe species, the (β + α) phases can co-precipitate, or the β phase (halite) can precipitate after the α phase. This creates a characteristic “polygonal” mineral distribution of halite surrounding authigenic FeO_x(OH)_y(H₂O)_zNa_dCl_e species (Figure 13). In this instance the β phase forms a characteristic film around the α grains (Figure 13).

This brief analysis has established that the chemistry within the ZVI pellets during desalination is complex and involves dissolution and reprecipitation of Na⁺, Cl⁻, and Feⁿ⁺ species.

3.9. Ironstone Precipitation

The ZVI pellets can develop localized “oolitic” ironstone concretions (Figure 14). These concretions are characterized by a wustite rim [41,44,45,46,47] and contain cuboid authigenic Fe⁰ crystals [41,44,45,46,47].

4. Chemical Implications of the Petrography for Aquifer Desalination

4.1. Dubinin-Astakhov Catalytic Model for ZVI Desalination

The intra-particle porosity of the FeO_x(OH)_y(H₂O)^b+/− particles within the ZVI pellets act as the catalytic site [17]. The particles adsorb Na⁺ and Cl⁻ ions from the water, and then desorb the ions as a [Na⁺Cl⁻] ion adduct (Figure 1, Figure 2, Figure 3, Figure 4, Figure 5, Figure 6, Figure 10, Figure 11, Figure 12 and Figure 13) within the inter-particle porosity (Figure 15).

This type of adsorption-desorption cycle is adopted by the Dubinin-Radushkevich model [55]. This model assumes that the adsorbed species is desorbed to fill the associated nano/meso/macro-dead-end porosity. Under this model [55]:

Ln(q_e) = ln(q_s) − k_adε²,

(8)

q_e = (q_s) exp( − k_adε²),

(9)

q_e = ion removed at equilibrium/unit ZVI (g/g); q_s = theoretical isotherm saturation capacity, g/g; k_ad = Dubinin-Radushkevich isotherm constant (M²·kJ⁻²); ε = Dubinin-Radushkevich isotherm constant, g. While q_e is measured, the other variables are unknown.

Adsorption-desorption associated with goethite (FeO_x(OH)_y(H₂O)_z) [56,57], akaganeite (FeO_x(OH)_y(H₂O)_zCl_n) [58] and ferrate (FeO_x(OH)_y(H₂O)_zNa_nK_m) [59] have been explained elsewhere using a Dubinin-Radushkevich adsorption model or a multilayer film model [55,56,57,58,59,60].

4.2. Significance of Porosity within the ZVI

The volumetric Dubinin-Radushkevich adsorption model is [60,61]:

V_x = V_o·exp[−(RT/β₁E_o·ln(P_s/P_a))²],

(10)

V_x = volume (NaCl) adsorbed by the ZVI and desorbed into the dead-end porosity (V_o) at time t = n, and at a relative pressure of (P_a/P_s) and temperature, T; P_s = initial pressure; P_a = adsorption pressure; V_o = the initial inter-particle nano-meso-macro dead-end pore volume; E_o = characteristic energy of adsorption of a standard adsorbate [61]; β₁ = the affinity coefficient of the adsorbate; E₁ = β₁E_o [61]; E₁ = characteristic energy for a specific fluid-solid system; R = the gas constant.

This model is rewritten as the Dubinin-Astakhov equation where [60]:

V_x = V_o·exp[−(RT/β₁E_o·ln(P_s/P_a))^m],

(11)

m = a factor which is related to pore size distribution and morphology. It follows that [60,61]:

V_x/V_o = exp[−(RT·ln(P_a/P_s)/E)^{2(or m)}],

(12)

Desalination ceases when V_x/V_o = 1. The porosity destruction rate constant associated with (V_x/V_o) is different to the desalination rate constant, k_r (Appendix A, Equations (A1)–(A4)), associated with (C_{t = n}/C_{t = 0}), though both are related.

Equation (12) demonstrates that:

(a)

the desorbtion of Na⁺ and Cl⁻ ions into dead-end porosity will result in their sequestration (Figure 1, Figure 2, Figure 3, Figure 4, Figure 5 and Figure 6); V_o > 0% at t = 0;

(b)

Placement of desorbed Na⁺ and Cl⁻ ions into pores (V_op) which are in direct communication with the main water body will result in the desorbed Na⁺ and Cl⁻ ions returning to the main water body. V_op > 0% at t = 0;

(c)

The inter-particle porosity = ϕ = V_o + V_op, and V_o = a·ϕ; V_op = b·ϕ, where a + b = 1. Once V_x/V_o = 1, {a} = 0, the residual porosity is V_op, i.e., {b} = 1. An irreducible water salinity level, B_S, occurs when {a} reduces to 0.

4.3. Significance of Fluid Pore Pressure within the ZVI Pellet

The fluid pore pressure within the pore, P_p, is different to the bulk (aquifer) pressure, P_s, and can be estimated as [61]:

P_p(d_p) = P_s·(exp(−E_p(d_p)/RT)),

(13)

E_p = average potential energy for the molecules being adsorbed (KJ/mole) [61] = 23.761d_p^−0.8782; d_p = pore width, nm, i.e., the pressure drop between the bulk pressure and the pore pressure (within the ZVI pellet) controls the desalination rate.

It follows that [60,61]:

V_x/V_o = exp[−(RT·ln(P_s/P_p)/E)²],

(14)

These Equations (8)–(14) imply:

• Na⁺ and Cl⁻ ion removal rates will increase with increased pressure differentials between the pressures in the pores and the wider aquifer;
• Na⁺ and Cl⁻ removal can have different rate constants [17];
• Smaller pores will be filled preferentially relative to larger pores.

5. Catalytic Implications of the Petrography

Desalination Model

The detailed catalytic desalination mechanism is unknown. It is believed [17] to involve a cycle of oxidation from Fe^0−II to Fe^III (and possibly Fe^{IV to VIII}) followed by reduction to Fe^0−II (Figure 16).

6. Commercial Implications of the Petrography for Aquifer Desalination

The petrographic analyses demonstrate that the removed Na⁺ and Cl⁻ ions are extracted from the water and are stored as NaCl(s) within the ZVI pellets. These observations indicate that pellets with a large internal structural porosity are able to remove more Na⁺ and Cl⁻ ions. They will have a higher rate constant, k_r, than similar pellets with a lower structural porosity. This is demonstrated in Figure B1.

Commercial scale up of the trial data, combined with a transfer of the reaction environment from a surface based reactor to a sub-surface aquifer, requires a conceptual and front end hydrological engineering and design analysis (FEHED) for each proposed location. This analysis is undertaken prior to the investment decision being made. The commercial investment decision will be controlled by the parameters defined in Equations (1)–(3). The purpose of the FEDEH is to establish that the partially desalinated water product can be delivered for a cost ($/m³) of less than M_c Equation (3).

7. Hydrological and Chemical Engineering Requirements for Commercialization

7.1. Conceptual and Front-End Hydrological Engineering and Design (FEHED)

The primary data requirements for the FEHED process are provided in

Table 1. Example set of primary data required to reconstruct an Aquifer Treatment Zone.

Item	Data Category	Data Subcategory	Example
1	Agricultural Holding Data
	1a	Holding Dimensions and Area	25 ha
	1b	Legal rights to modify water composition in an aquifer
	1c	Irrigation Requirement	20 m3·day−1
	1d	Aquifer Thickness	1 m
2	Target Irrigation Water Salinity
	2a	Aquifer Salinity	4 g·L−1
	2b	Required Irrigation Water Salinity	2 g·L−1
	2c	Crop yield with irrigation using saline aquifer water	1 t·ha−1
	2d	Crop yield with irrigation with partially desalinated water	5 t·ha−1
3	Agricultural Economics Data
	3a	Crop yield decrement as a function of salinity	[16,18,19,20,21,22,23,24,25]
	3b	Crop market value	$300/t
4	Experimental Desalination Data
	4a	Rate constants	Equations (A1)–(A4) and (B1); Figure B1, Figure B2 and Figure B3; Table E1
	4b	Relationship between residence time, td, in the reaction environment and desalination	Equations (A1)–(A4) and (B1); Figure B3
	4c	Rate of Salinity Decline as a function of td	Catalyst ST: (Figure B3)
	4d	Statistical variation of desalination around the mean at the 99% confidence limits	Catalyst ST: 59.1% to 65.8% Catalyst K: 31.7% to 47.0%
	4e	Statistical desalination at the 1st and 3rd quartiles, e.g., Catalyst ST: 54.4% to 70.8%; Catalyst D: 24.8% to 65.8%; Catalyst E: 58.9% to 92.1%; Catalyst K: 33.2% to 45.5%
	4f	Statistical variation of Na+ Removal around the mean at the 99% confidence limits	Catalyst ST: 59.1% to 65.8% Catalyst K: 39.1% to 62.8%

. This process requires the kinetic parameters for desalination to be both known and modeled (Appendix A and Appendix B).

The first stage in the process is an assessment of the available catalyst data in order to establish the amount of catalyst required and the Aquifer Treatment Zone size required by each catalyst. This stage is addressed in Section 7.2.

The second stage in the FEHED process is the selection of the preferred catalyst. This stage is addressed in Section 7.3.

The third stage in the FEHED process is the creation of a control desalination data set (e.g., Figure B1, Figure B2 and Figure B3; Table E1 and Table F1), a kinetic model (e.g., Equations (A1)–(A4)) and a control hydrological model. Section 7.4 and Appendix C detail the equations and models that can be used to create the control hydrological model.

The fourth stage in the FEHED process is analysis of the different development strategies which could be used to create the Aquifer Treatment Zone. In this study two catalyst dependent strategies are considered (in Section 7.5 and Section 7.6): (i) a Static Strategy; (ii) a Recycle Strategy. Each strategy integrates the hydrological and chemical models (Appendix A, Appendix B and Appendix C) to define the design and operating parameters for the proposed Aquifer Treatment Zone. These models are summarized in Appendix E, Appendix F and Appendix G.

The base data set, which is used to illustrate the various hydrological and chemical models in Appendix A, Appendix B, Appendix C, Appendix D, Appendix E, Appendix F and Appendix G, is derived from the desalination trials using the ST catalyst (e.g., Figure 1, Figure 2, Figure 3, Figure 4, Figure 5, Figure 6, Figure 8, Figure 9, Figure 10, Figure 11, Figure 12, Figure 13 and Figure 14; Figure A1; Figure B1, Figure B2 and Figure B3).

These trials have demonstrated that if ZVI catalyst pellets are placed in a body of water, they will desalinate the water (Figure B1, Figure B2 and Figure B3). The two sided statistical 99% confidence limits for the mean desalination after 200 days are:

(a)

15 mm diameter Cu⁰ cased pellets (without operational optimization) = 59.1% to 65.8%, n = 50;

(b)

75 mm diameter MDPI cased pellets (without operational optimization) = 58.2% to 83.4%, n = 15. i.e., 99% confidence limits (BS2846) are: mean ± (Standard Deviation) (t_0.995/(n)^0.5), where n = number of samples; t_0.995 is the value (for n) of the students t test distribution (at the 99% confidence limits).

Similar data sets can be obtained for alternative catalysts (e.g., Table 1, Table E1 and Table F1; Figure B1). This study assumes that the pellets (or cartridges) are placed in removable containers at a number of loci within the aquifer (e.g., in infiltration devices, wells, or boreholes [62,63,64,65]). Removal of the pellets from the aquifer, will allow the NaCl to be periodically removed from the water body.

Once the agricultural land owner has decided that the desalination project may be economically viable (e.g., Equations (1)–(3)), the FEHED process sequentially steps the project through three phases (Appendix D):

• prospective resource;
• contingent resource;
• developed resource.

7.2. Scale Up of Trial Desalination Kinetic Data to Provide an Assessment of the Amount of ZVI Required and the Size of the Aquifer Treatment Zone

The trial data associated with a specific catalyst (Item 4, Table 1) forms the primary database for the conversion of a saline aquifer to a proposed Aquifer Treatment Zone. The initial stages in this process are [66,67,68,69,70,71,72,73,74]:

• Conversion of the salinities (C) at t = 0 and t = n for batch trials to:

  • Rate constants, k_r (Equations (A1) to (A4); Figure B1, Figure B2 and Figure B3);
  • Required residence times, t_d, for the water within the reaction environment (Equations (A1) to (A4); Figure B1, Figure B2 and Figure B3).
• Up scaling the ZVI quantities used in the trials:

  • The petrography demonstrates (Figure 1, Figure 2, Figure 3, Figure 4, Figure 5, Figure 6, Figure 8, Figure 9, Figure 10, Figure 11, Figure 12 and Figure 13) that increasing the amount of ZVI will effectively increase the available ZVI voidage (porosity) available for the sequestered NaCl;
  • The trial data demonstrates that the amount of NaCl removed per unit time is independent of the concentration of ZVI, g/L, P_w (Figure 17a,b);
  • These trial observations indicate that:

    • Each 1 g of ZVI will be expected to remove at least 0.35 g NaCl before it needs to be removed from the reaction environment and is replaced (Figure 17a–c).
    • Scale up by the placement of 10 t ZVI in a saline aquifer would result in the removal of up to 3.5 t NaCl from the aquifer, before the replacement of the ZVI is required (Figure 17a–c). This experimental scale up allows:

      • the anticipated volume of water which can be desalinated by a single ZVI charge to be defined (Figure 17d)
      • the anticipated replacement frequency of the ZVI to be defined (Figure 17e).
• Assessing from the trial data set, the size of the Aquifer Treatment Zone which is required to deliver the target volume of irrigation water:

  • In a commercial development the ZVI will be placed in a number of loci (wells) which intersect the aquifer. The number of loci required is defined by the ZVI concentration data used in the trials (e.g., Figure 17a,b).
  • The required residence time, t_d, to achieve the required irrigation water salinity is defined by the trials kinetic data (e.g., Equations (A1) to (A4); Figure B1, Figure B2 and Figure B3). The residence time is combined with the irrigation water volume required (Table 1) to define the minimum volume of water, and the minimum gross rock volume, contained within a proposed Aquifer Treatment Zone, V_aq (Equations (E1)–(E3) and (E5)).
  • This minimum water volume is then integrated with the known aquifer properties in order to identify an aerial extent for the proposed Aquifer Treatment Zone Equation (E4).

7.3. Selection of Catalyst for the Proposed Aquifer Treatment Zone

The trials [16,17] have established that Fe⁰ (Figure A1) can be used to create two types of structured ZVI desalination catalyst. They are:

• A Type A Catalyst: This catalyst type (catalysts ST and MT) has a rigid porosity (Figure A2) and will, when placed in saline water, gradually remove NaCl from the water over a period of time (e.g., 100–1500 days). It has a low k_r (Figure B1, Figure B2 and Figure B3) and a low B_S. The irreducible (or equilibrium) salinity, B_S, increases with decreasing catalyst porosity (Equations (8)–(14)). Fe⁰ powders, with no effective porosity, do not desalinate water (Figure B1) [36]. The statistical 1^st and 3rd desalination quartiles for t_d = 200 days, for the three Type A catalysts considered in this study are: (i) ST catalyst (Figure A2; Figure B1, Figure B2 and Figure B3), 15 mm diameter pellet: 54.4% to 70.8%; n = 50; (ii) ST catalyst (Figure B3b), 75 mm diameter pellet: 63.2% to 76.8%; n = 13; (iii) MT catalyst (Figure B1), 20–25 mm diameter pellet: 6.1% to 13.7%; n = 24. The amount of NaCl that can be removed is controlled by the porosity (Equations (8)–(14)) and the amount of catalyst placed in the reaction environment (Figure 17). The Type A catalyst will require a residence time in the aquifer of 100–500 days before the maximum (design) equilibrium desalination is achieved (e.g., Figure F1, Figure F2 and Figure F3).
• A Type B Catalyst: This catalyst type (catalysts C to K) has an expandable, or poro-elastic, porosity (e.g., Table E1 and Table F1). The catalyst will, when placed in saline water, in a reaction environment incorporating water recycle, result in a relatively rapid removal of NaCl from water (e.g., 3 to 50 h). It has a high k_r (Table E1 and Table F1) and a relatively high B_S (e.g., 20%–80% (e.g., Table 1, item 4e)). It can preferentially remove Na⁺ or Cl⁻ ions. Once an initial B_S is achieved, or the water recycle ceases, desalination continues to progress at a substantially slower rate (e.g., Figure B1, Figure B2 and Figure B3). The Type B catalyst will require a residence time in the aquifer of 1–30 days before the maximum (design) equilibrium desalination (e.g., F4, F5).

Two subsurface ZVI desalination strategies have been identified. They are:

• A Static Strategy: where the ZVI catalyst is placed in the aquifer and no water is recycled to the aquifer;
• A Recycle Strategy: where some of the product water is recycled back to the aquifer.

A Type A catalyst is designed for use in the “Static Strategy”. A Type B catalyst is designed for use in the “Recycle Strategy”. The choice of catalyst controls the development strategy and costs associated with a proposed Aquifer Treatment Zone.

The desalination strategy used in a proposed Aquifer Treatment Zone can be switched from one approach to the other, during its operating life.

7.4. Control Hydrological Data and Models Required for the Proposed Aquifer Treatment Zone

Conversion of an aquifer to an Aquifer Treatment Zone requires the installation of one or more abstraction wells [75], and the installation of a number of loci (wells) containing the ZVI catalyst. Observation loci (wells) may be placed outside the proposed Aquifer Treatment Zone and a number of reinjection/infiltration loci (wells) may be placed within the proposed Aquifer Treatment Zone.

Each of these loci provides a spatial data point which provides information about: (i) the aquifer geology (thickness, porosity, permeability, mineralogy); (ii) aquifer hydrology and chemistry (piezometric level, water chemistry (pH, Eh, EC, ion concentrations, salinity)).

Three sets of hydrological control data are required prior to placement of the ZVI in the aquifer. They are: (i) aquifer prior to abstraction; (ii) aquifer during a prolonged abstraction test; (iii) aquifer following a prolonged abstraction test.

This information is used during the FEHED to build a hydrological, hydrochemical and geological model of the aquifer and its expected performance during operation.

Periodic monitoring of these loci following installation of a proposed Aquifer Treatment Zone will allow: (i) the performance of the aquifer to be monitored; (ii) any changes in performance to be identified. This will allow an appropriate remedial program to be instigated to mitigate against adverse changes in the aquifer.

The primary hydrological aquifer parameters, which are used to provide the control database, build the hydrological aquifer model, and monitor operation are documented in

Table 2. Hydrological measurements and modeling required in order to: (i) convert an aquifer into a proposed Aquifer Treatment Zone (ATZ); and (ii) monitor the performance of an installed Aquifer Treatment Zone. Further details are provided in Appendix A and Appendix G.

Item	Hydrological Parameter	Measured or Modelled Item	Equations and Figures
1	Control Data Set (Aquifer)
	1a	Permeability	Equations (G1)–(G4), Figure G1
	1b	Porosity	Equations (G1)–(G4), Figure G1
	1c	Thickness	Equations (G1)–(G4), Figure G1
	1d	Piezometric Surface	Equations (C1) to (C4)
2	Control Models
	2a	Isopache (thickness) Model	Equations (G1)–(G4), Figure G1
	2b	Iso-permeability Model	Equations (G1)–(G4), Figure G1
	2c	Iso-potential Model	Equations (C2) and (C3)
	2d	Flow Rate Model	Equations (C1)–(C4)
3	Control Abstraction Test
	3a	Piezometric Surface	Equations (C1)–(C4)
	3b	Iso-potential Model	Equations (C2) and (C3)
	3c	Flow Rate Model	Equations (C1)–(C4)
	3d	Drawdown Model	Equation C5
	3e	Drawdown Permeability Model	Equations (C6) and (C10)–(C12)
	3f	Radius of Influence of the Abstraction Well	Equation (C7)
	3g	ATZ volume as a function of the abstraction rate	Equation (C8)
	3h	Flow rate and flow direction at each loci	Equation (C9)
	3i	Design Parameters for the Aquifer Treatment Zone	Equations (C1)–(C9), Figure C3
4	Operational Monitoring
	4a	Iso-potentiaL, flow rate, and drawdown measurements and modelling	Equations (C1)–(C12)
	4b	Permeability Reductions associated with ZVI	Equation (C13), Figure C4
	4c	Halocline monitoring	Figure C5
5	Diffusion Modelling
	5a	Impact of fluid velocity changes	Equations (C14)–(C21), Figure C6
6	Space Velocity Modelling
	6a	Impact on the desalination rate	Equations (C22)–(C26)
	6b	Impact of water residence time in the ATZ	Equation C27
7	Modifications to Aquifer Properties during Prolonged Abstraction
	7a	Redox parameters	Figure C7
	7b	Permeability reduction due to gas occlusion or mineralization	Equations (C28)–(C30)
	7c	Permeability reductions and redox modifications associated with biofilms	Equations (C31)–(C36)

.

7.5. Construction of the Proposed Aquifer Treatment Zone Using the Static Strategy

A proposed Aquifer Treatment Zone constructed using the Static Strategy will have the following characteristics:

(a)

a radial pattern of ZVI loci;

(b)

a single abstraction well located within the “Aquifer Treatment Zone” ;

(c)

an excess of ZVI is placed in each ZVI loci to increase the life expectancy of each ZVI charge. The life expectancy of a ZVI charge could be designed to exceed 20 years (Figure 17);

(d)

a simple process flow diagram (Figure 18);

(e)

a simple construction (Figure 19).

This type of approach:

(a)

is suitable for most agricultural holdings;

(b)

can be installed using basic, simple, low cost, and widely available equipment and tools. Many of these tools may already be available on the agricultural holding, as the required augers are commonly used to install fence posts; and

(c)

requires no operating energy during its life (apart from power for the abstraction pump).

A side effect of this approach is that any water flowing through the Aquifer Treatment Zone (Figure 19) will be treated by the ZVI. ZVI treatment will remove a variety of cations and anions (e.g., nitrates and organic pollutants) from the water [16,66,68].

Costs Associated with the Static Strategy

The principal costs are the installation of the ZVI loci, ZVI, observation loci, abstraction loci, abstraction well/pump, distribution tank or impoundment, and associated pipe work.

This strategy will require the ZVI pellet charge to be installed with a view to removal and replacement every x years (e.g., Figure C3h).

The ZVI catalyst life expectancy and Aquifer Treatment Zone life expectancy has not been demonstrated in a commercial environment. The unsubsidized amortized cost (capex + opex) of the partially desalinated irrigation water, when this technology is eventually commercialized, may fall within the range $0.05/m³–$0.5/m³.

7.6. Construction of the Proposed Aquifer Treatment Zone Using the Recycle Strategy

The operation of a Type A desalination catalyst (e.g., ST catalyst) can be described and modeled using Equations (8)–(14), (A1)–(A4). The primary control on k_r and B_S (for a specific t_d, P_w, P_i and a_s) is V_o. Modeling demonstrates (Appendix F) that the introduction of a recycle loop in the process flow (Figure 20), can, depending on the location of the reinjection loci, result in the equilibrium B_S increasing or decreasing (Figure F1, Figure F2 and Figure F3; Equations (F1)–(F3)).

The Type B desalination catalyst removes V_o as a constraint, by allowing V_o to expand with time.

7.6.1. Poro-Elastic vs. Rigid V_o

A specific catalyst formulation (catalyst D) was used to construct both a Type A catalyst (Catalyst D(i) with a rigid V_o and a Type B catalyst (Catalyst D(ii)) with a poro-elastic V_o.

The two catalysts were trialed in a 240 L recycle reactor (Figure 20) with a P_w of 0.5 g/L with a t_d of 10 to 24 h (at a temperature of between 5 and 20 °C). The 99% confidence limits on the mean desalination were:

(i)

catalyst D(i) = 8.4% to 13.7%, n = 31;

(ii)

catalyst D(ii) = 9.9% to 65.0%, n = 18.

Both catalysts continued to desalinate over the next 100–700 days using the rate constants demonstrated in Figure B1 (when (y + n) = 0 (Figure 20)). This trial established that a higher k_r and lower B_S is achieved (over a 24 h period) by constructing the catalyst with a poro-elastic V_o.

7.6.2. Recycle Strategy

In a recycle environment, in the presence of a Type B catalyst, a higher k_r occurs when (y + n) ≥ 0 (Figure 20; Table E1 and Table F1), i.e., k_r has a partial dependency with V (Figure F4). Further desalination ceases when a new equilibrium for (V − D) is achieved (Equations (C14)–(C21)). This equilibrium controls B_S and the salinity of the product water. The trial results in Table E1 and Table F1 demonstrate that the equilibrium position for (V − D) can be associated with a rapid desalination of 40%–75%.

These trial observations demonstrate (Table E1 and Table F1) that it may be possible to construct an Aquifer Treatment Zone by incorporating a recycle loop in the process flow (Figure 20). The trial results (Table E1 and Table F1) indicate that this will allow, by comparison with the Static Strategy, the amount of partially desalinated irrigation water produced from a specific Aquifer Treatment Zone volume, to be substantially increased (Table F1).

The Recycle Strategy construction (Figure 20) requires that the volume of water abstracted from the Aquifer Treatment Zone is greater than the amount of water required for irrigation (Figure 20). The excess water is then reinjected into the Aquifer Treatment Zone.

The Recycle Strategy has infiltration loci (for water recycle) and a network of distribution pipes to the infiltration loci (Figure 21). The trial results (Table E1 and Table F1) indicate that P_i for a Recycle Strategy is lower than P_i for a Static Strategy (Figure 17). This reduces the number of ZVI loci required, and the amount of ZVI required, to achieve a specific irrigation water abstraction rate.

The principal additional operating cost of the Recycle Strategy (relative to the Static Strategy) is the increased operating cost of the abstraction well (Figure 20). For a specific target irrigation water abstraction rate, the Recycle Strategy, will have a smaller Aquifer Treatment Zone, and will require less ZVI (Table F1).

The amount of water abstracted for recycle is a function of the catalyst selected. The examples in Table E1 and Table F1 demonstrate that (y + n)/m (Figure 20) is between 0.45:1 and 7.45:1. i.e., the recycle water volume is between 9 and 149 m³/day for an Aquifer Treatment Zone producing 20 m³/day of partially desalinated irrigation water.

In most locations, the air-water contact in the distribution tank will be above the air-water contact in the aquifer (Figure 22). In these circumstances, a recycle pump is unlikely to be required. The recycle water distribution can be effected using buried pipe work, gravity drainage and infiltration (Figure 22). This strategy removes a requirement for a recycle pump and external power source.

8. Commercial Scale-Up Risk

Scaling-up the ZVI desalination trials, using a recycle strategy, to a 20 m³/day commercial Aquifer Treatment Zone involves a throughput scale-up of between 6 and 3000 (Figure C5, Table E1 and Table F1). A larger scale-up factor is required for the static strategy (Figure B1). The commercial risks associated with the scale-up of trial data relate to cost and performance. These commercial risks can increase as the magnitude of the scale-up increases. A commercial scale-up from trial data of between 6 and 3000 is at the lower end of conventional practice [76,77,78,79,80].

Standard chemical process engineering industrial practice is to use the results from small scale trials as a basis for an investment decision, e.g., [76,77,78,79,80]. It is not unusual for the direct commercial, throughput, scale-up (from bench/pilot trials to commercial plants) associated with petrochemical (and chemical) processes to be in the order of 500,000 to 100,000,000.

The scale-up associated with the commercialization of post-1980s ZVI water treatment processes includes:

• Permeable Reactive Barriers (PRB’s) inserted in aquifers, e.g., [81,82]: Small scale (batch and continuous flow) trials containing 0.001 to 5 kg Fe⁰ were scaled-up to >300 PRB’s containing 2 to 1000 t Fe⁰ each;
• Industrial Waste Water Treatment Plant, Shanghai: processing 60,000 m³/day [83,84]. A number of 2 h duration pre-development batch flow trials (containing 0.5 kg Fe⁰) were undertaken [83]. They were succeeded by a single continuous flow pre-development trial (containing 40 kg Fe⁰) at a rate of 120 L/d [83]. This pilot trial was operated for 6 months prior to commercialization [83]. The throughput scale-up factor was 500,000. The scale-up factor in the amount of Fe⁰ used was 22,950.

9. Comparative Desalination Cost Structure

Large (>100,000 m³·day⁻¹) plants (Multi-stage flash distillation (MSFD), reverse osmosis (RO)), which are integrated with a thermal fuel power station have the lowest current desalination water costs ($/m³). The cost of the desalination process is subsidized by the electricity sales associated with the power station. The cost structures associated with large (>90,000 m³·day⁻¹) integrated conventional desalination systems are summarized in

Table 3. Comparison of integrated power station-desalination unit complexes. MSFD = Multi-stage flash distillation + power station; RO = Reverse Osmosis + power station; ZVID = ZVI desalination + power station; ZVI = ZVI desalination without the power station; Assumed Target Capital Investment for the schematic ZVI desalination Aquifer Treatment Zone is $300 MM, (calculated using Equation (15)), when {a} = 1.0. MM = million; $ = US dollars. MSFD, RO, and Power Station Costs and Structures are from Reference [87]. Power Station characteristics [87]: 24.2 MPa Steam Pressure; 6 KPa Condenser Pressure; 25.2 MJ·kg⁻¹ Coal HLV; 92% boiler efficiency; 96.5%–98% turbine efficiency; 10,000 m³·h⁻¹ water supply to the power station. Fuel Costs based on 2014 fossil fuel commodity prices. The power station costs used in the Target ZVID cases are also taken from Reference [87]. Different cost components may have different values of the scale-up constant {a}. The scaled target ZVI examples indicate the water delivery cost range which may be achievable. Power requirements associated with water abstraction and recycle are site and catalyst specific. They are therefore not shown for the ZVID and Scaled Target ZVI cases.

	Actual MSFD	Actual RO	Target ZVID	Scaled Target ZVI	Scaled Target ZVI	Scaled Target ZVI	Scaled Target ZVI	Scaled Target ZVI
Power Plant Capacity, MWh	291.0	291.0	291.0
Power Usage, MWh	9.8	31.1	0.0
Net Power Available for Sale, MWh	281.2	259.9	291.0
Net Power Sales Value, $ MM/a, at $70/MWh	172.4	159.4	178.4
Pressure, MPa		7.0	0.0
ZVI Desalination scale up constant {a}			1.0	1.0	0.95	0.9	0.8	0.7
Sea Water Feed, m3/day	390,921.6	296,712.0	90,196.8	90,196.8	90,196.8	90,196.8	90,196.8	90,196.8
Product Water, m3/day	149,803.2	90,196.8	90,196.8	90,196.8	90,196.8	90,196.8	90,196.8	90,196.8
Reject Brine, m3/day	241,118.4	206,515.2	0.0	0.0	0.0	0.0	0.0	0.0
Capital Investment, $ MM/a	35.2	44.7	29.6	6.8	3.8	2.4	0.8	0.2
Pre-Treatment and Pressurisation Cost, $ MM/a		17.0
Labour Cost, $ MM/a	8.8	6.7	2.0	2.0	1.1	0.7	0.2	0.1
Membrane Replacement Cost, $ MM/a		11.8
Chemical Treatment Cost, $ MM/a	5.3
ZVI Replacement Cost, $ MM/a			10.0	10.0	5.7	3.5	1.1	0.4
Spares Cost, $ MM/a	7.2
Power Station Fuel Cost, $ MM/a	203.2	143.3	120.0	0.0	0.0	0.0	0.0	0.0
Operations and Maintenance Cost, $ MM/a	18.8	24.9	21.0	2.0	1.1	0.7	0.2	0.1
Net Cost, $ MM/a	106.1	89.0	4.1	20.8	11.7	7.3	2.3	0.7
Net Cost, $/m3	1.94	2.70	0.13	0.63	0.36	0.22	0.07	0.02

.

An Aquifer Treatment Zone producing 90,000 m³/day will require access to a large aquifer. The normalized trial data (associated with different catalysts (Table F1)) indicates that the required target volume of water within the Aquifer Treatment Zone may be: (i) between 12,000 m³ and 170,000 m³ (RC1–RC3, RC5–RC9, Table F1; mean = 73,309 m³; 99% upper confidence limit = 169,108 m³), or (ii) <2,000,000 m³ (RC4, Table F1).

The standard chemical engineering cost scale up equation is [85,86]:

C_e = C_k·(P_lc/P_pc)^a,

(15)

C_e = Expected cost, $ (excluding site costs); C_k = Known cost of the constructed unit; P_lc = Larger Plant throughput, m³ day⁻¹; P_pc = Constructed unit throughput, m³ day⁻¹; a = an exponent scale up factor (range = 0.3–1.2 [86]). It is commonly taken as 0.6, range 0.4–0.8 [85,86]. The constructed unit has a known cost and throughput capacity.

Table 3 includes an example target set of costs for a large 90,000 m³·day⁻¹ ZVI in situ desalination aquifer complex. This provides an indication of how economies of scale Equation (15) may impact on the delivered cost of the partially desalinated irrigation water.

Conventional desalination delivers a water product with >97% of the Na⁺ ions removed. By way of contrast, ZVI desalination is only expected to remove between 20% and 70% of the Na⁺ ions (commercial target zone is 35%–60% Na⁺ removal).

10. Conclusions

This study has investigated whether ZVI desalination pellets and cartridges could potentially be used to desalinate a body of water within an aquifer. Two different aquifer reconstruction strategies have been identified and modeled, a static strategy and a recycle strategy.

10.1. The Static Strategy

This requires a relatively long residence time within the aquifer (e.g., 100–300 days). The expected pre-tax cost of this strategy is in the order of $0.05–$0.5/m³, where the bulk of the cost (excluding the costs associated with abstraction) is a capital cost. The principal operating cost is the abstraction pump.

10.2. The Recycle Strategy

This requires a relatively short residence time within the aquifer (e.g., 0.3–3 days). The delivered water cost may be less than $0.1/m³. This strategy will have a requirement for reinjection/infiltration. It has a higher pumped abstraction rate from the aquifer than the static strategy. The recycle strategy allows active management of the aquifer to be undertaken to control the salinity of the abstracted irrigation water.

10.3. Dual Use Strategy

An aquifer, which is initially developed using a Static Strategy, can be subsequently converted to allow operation using a Recycle Strategy, and vice-versa.

ZVI regeneration and replacement intervals are a function of catalyst and design. The replacement interval could potentially be once every 1 to 20 years.

10.4. Next Stage

The next stage will be controlled field testing in an aquifer stepping up through a range of volumes and throughput (e.g., 1 to 100 m³·day⁻¹) sizes. This testing program will be designed to establish, and mitigate, the issues associated with scale-up, prior to commercialization.

A successful commercial scale test program will allow commercial saline aquifer reconstructions (targeting 1–1000 m³·day⁻¹ of partially desalinated product water) to be undertaken.