Transatlantic Freshwater Aqueduct

Water Resour Manage (2010) 24:1645–1675 DOI 10.1007/s11269-009-9518-y

Transatlantic Freshwater Aqueduct Viorel Badescu · Dragos Isvoranu · Richard B. Cathcart

Received: 5 August 2008 / Accepted: 6 October 2009 / Published online: 22 October 2009 © Springer Science+Business Media B.V. 2009

Abstract This paper offers a technical and geopolitical reappraisal of a macroengineering proposal to plumb Earth’s freshwater, siphoning some of it from a region of surplus (Amazon River Basin) to a region of shortage (arid northern Africa) via his positively buoyant (subsurface floating) seabed-anchored Transatlantic Freshwater Aqueduct. Two different routes for the pipeline, of length 4,317 and 3,745 km, respectively, have been considered. Pipe diameters larger than 60 m are necessary for “reasonable” low pumping power (i.e., less than 20 GW). Using a bundle of smaller size pipes instead of a larger single pipe might overcome technical difficulties but the advantage of simplifying the construction technology might be exceeded by the disadvantage of consuming more power in operation. To keep the number of pumping stations reasonably small (i.e. fewer than 20) a single pipe of diameter higher than 30 m (or bundles of smaller diameter pipes) is required. The Atlantic Ocean currents may be used to provide the necessary power for pumps. The available power possibly provided by the North Brazil Current ranges between 2 and 9 GW. The North Equatorial Current may provide less than 0.3 GW power while the North Equatorial Counter Current provides the largest power availability, ranging between

V. Badescu (B) · D. Isvoranu Candida Oancea Institute, Polytechnic University of Bucharest, Spl. Independentei 313, Bucharest 060042, Romania e-mail: [email protected] D. Isvoranu e-mail: [email protected] R. B. Cathcart Geographos, 1300 West Olive Avenue, Suite M, Burbank, CA 91506, USA e-mail: [email protected]

1646

V. Badescu et al.

2 and about 100 GW. A rough cost estimate of the project is about 20,600 GUSD and 18,400 GUSD, respectively, for two pipeline routes. Keywords South America · Amazon River · Northern Africa · Bulk freshwater transfer · Undersea floating pipeline · Macro-engineering

1 Introduction Freshwater garnishment macro-projects involve artificial large-scale bulk transfers of that vital fluid over great geographical distances under applied forecasting and management conditions, from regions of surplus to regions of deficit, for the economic purpose of social advancement of the drier region by subsequent agricultural and industrial development (Tvedt et al. 2006). In some significant cases, however, the reverse situation exists. For example, the possibility of an Alaska-California Undersea Aqueduct furnishing a reliable long-term freshwater supply solution for California’s chronic urban freshwater shortages was considered from 1965 until 1991 (California Undersea Aqueduct 1975). Also, unsettling early twenty-first century hydrological news that the Colorado River’s runoff is decreasing due to measured regional climate change has positively stimulated the political prospects in California of that old macro-project proposal as a possible urban water supply semi-replacement infrastructure (Barnett and Pierce 2008). The Amazon River accounts for ∼15% of global runoff but this potentially useful freshwater runoff (2 × 105 m3 /s, with fluctuations within its predictable climate regime variability pattern (Milly et al. 2008; Garreaud et al. 2009) loses its social and commercial value as it mixes with the saltwater of the Atlantic Ocean. These basic facts of modern human life inspired Heinrich Hemmer to propose an audacious anchored submarine floating oceanic freshwater pipeline macro-project connecting South America’s water-surplus Amazon River Basin with the arid landscape of northern Africa. “A pipeline stretching from the mouths of the river Amazon [near Macapá, Brazil, at 0◦ 15 0 North Latitude by 51◦ 10 0 West Longitude] to North Africa would be about 4,300 km-long. At a speed of 2 m/s and a capacity of 10,000 m3 /s, it would have a diameter of 80 m. . . . Calculating a demand of 10,000 m3 fresh water [per] hectare per year, 315,000 km2 could be fully irrigated” (Hemmer 1993). Published by the UNO in 1997, the FAO Land and Water Bulletin 4, “Irrigation potential in Africa: a basin approach”, substantiated the imported freshwater Sahara irrigation demand assumption made by Hemmer. Currently, part of the Sahara is irrigated using vertically pumped groundwater but that supply is slow to recharge naturally and hydrological experts in Libya predict that the practice will be forced to cease circa 2060. Therefore, freshwater imports to the Sahara will be required sometime after the mid-twenty-first century. In the geographical context of northern Africa, desalination—currently costing ∼USD 1–3/m3 —is deemed to be the competition for Hemmer’s macro-project and, in this sense, will set the price ceiling when all other factors such as security and environmental impacts have been taken into account (Dore 2005). In most cases, vertical distance is the primary driver of potable water conveyance costs, not the horizontal distance (Zhou and Tol 2005).

Transatlantic Freshwater Aqueduct

1647

However, some irrigation and drainage macro-engineering experts have asserted that the twenty-first century world has exited an “Age of Water Development” and entered an “Age of Water Management”. “Though there is still much undeveloped water in the Amazon and Congo rivers, the total development, pumping, and transportation costs are likely to exceed the value of this water before it gets to the places where it is most needed” (Allison 2003). In this context, a closer look at the Transatlantic Freshwater Aqueduct (TFA) macro-project is useful. In the following, the TFA macro-project is analyzed in some detail from technical, economical and environmental points of view.

2 Macro-Engineering Relevancy Until circa 1965, the conception and management tasks associated with geographically and economically large-scale infrastructure construction macro-projects (also sometimes called “mega-projects”) were typically overseen by civil engineers. As macro-projects become more and more complex to complete, and globalization of the world’s ecosystem-nation economies proceeds further during the twenty-first century, the challenges associated with the management of these macro-projects become more complicated. The importance of global and regional coordination in conservation is increasing and between-country collaboration seems to be a social and economic necessity. Macro-engineers deal mostly with the conception of national, international and extra-terrestrial infrastructures (Badescu et al. 2006; Beech 2008). Modern macro-engineers utilize new and versatile materials, wield honed skills in the prediction of behaviors of built structures under varying loads, employ new technical capacities for moving earth, air and water and, as well, up-to-date organizational techniques for managing the material and power logistics and human (even, sometimes, robotic), labor required of geographically large-scale macroprojects (Singh 2007). Circa 2009, construction is big business with yearly worldwide expenditures totaling ∼4,000 GUSD, with about one eighth of that devoted mainly to water supplies building activity (maintenance, up-grading of obsolete installations, new infrastructure). One of the most remarkable results possible with the TFA macro-project is that an entirely novel kind of intercontinental (South America-northern Africa) and international “watercourse ecosystem” integration will be the ultimate macroengineering product achievement goal. The TFA infrastructure is an oceanic macroengineering project proposal of greater complexity than any tackled previously (Schuiling et al. 2005, 2007; Badescu and Cathcart 2008) and, therefore, required to consider political risks, innovational risks, and organizational risks, weather extremes (in desert, jungle and oceanic), transient workforces, likely human labor construction errors, to name a few. While the climate regime of the Sahara will remain much as it is nowadays for a very long period to come, the Amazon River Basin’s climate regime is projected to change markedly and severely by 2100 if global changes that include atmosphere warming continues at the supercomputer-modeled pace (Williams et al. 2007; Cook and Vizy 2008). Specifically, for the Amazon River Basin, a dramatic ecological shift is forecast, but no reliable useful forecasting details are given. It is a fact that, until recently, water policy makers in Brazil and elsewhere underestimated past human

1648

V. Badescu et al.

exploitation of the Basin’s resources (Heckenberger 2009). Some geoscientist now suggest that reforestation of the Amazon River Basin, abandoned after pandemics instigated by early European explorers and settlers, drastically reduced the numbers of native Amazonians by tens of millions approximately 500 years ago, and may thus have caused the so-called “Little Ice Age” in Europe (Nevle and Bird 2008). Nowadays, deforestation seems to be the main effect of human inhabitation of the Basin. Currently, the Amazon River discharges ∼6,300 km3 annually, politically governed by, in part, the “Treaty for Amazonian Cooperation” signed by appropriate ecosystem-country representatives on 3 July 1978 (Landau 1980). On 23 May 2008, 12 South American ecosystem-nations signed the Union of South American Nations Treaty, UNASUL (Uniao de Nacoes Sul-Americanas, in Portuguese). Just how, or if, this historically recent agreement will affect the in-force 1978 Treaty is unknown. Indeed, one hopes that legal imposition of the TFA macro-project will assist in the ongoing political and scientific effort to preserve the enormous catchment’s forested terrain by imposing a consumer demand limitation factor on the Amazon River Basin’s owners (Kindermann et al. 2008). During June 2008 Brazil and Venezuela signed an agreement to cooperate in their shared part of the Amazon River Basin to help preserve the rainforest, of which Brazil legally controls ∼63%. Note, however, that most of the Amazon River’s freshwater derives from the glaciers and rivers draining the eastern slope of the Andes, especially Peru and Ecuador (Goulding et al. 2003). With lowland deforestation there will, of course, be more fluvial erosion of the land, muddying the waters to some unpredictable future state. There are, however, numerous conflicting reports of Amazon River Basin deforestation rates and some science journal reports state that deforestation has diminished or even stopped. The unlikely possibility of complete Amazon River Basin deforestation is not considered herein. Brazil has become a food-producing and food-exporting Superpower and yet “Brazil could, in principle, triple its area under cultivation over time—without felling any more rain forest” (Omestad 2008). Inter-basin freshwater transfer is sometimes considered in this context, despite Brazilian legislation that does not cover this situation, and future geopolitical conflicts may arise (De Carvalho and Magrini 2006). Whilst some freshwater supply experts would likely suggest the intra-Brazil water diversion from the Amazon River to the semi-arid northeast regions of Brazil, this macro-engineering option remains open even with construction of the TFA. (Similar shorter diversions of freshwater have already been proposed for the Congo River in Central Africa).

3 Transatlantic Freshwater Transport—Bag Trains or a Pipeline? It was after 1850, when the first undersea telegraph cables had been successfully laid, that the Earth’s ocean, and its seafloor, became a human destination rather than merely a barrier or byway to commercial transport and telecommunication (Bouma 1990). During the final decade of the twentieth century, Aquarius Water Transportation became the first private-sector company to tugboat-tow very capacious polyurethane bags—sometimes dubbed “Medusa bags” or “Dracones”—containing drinkable freshwater to various isolated, water-deficit Cyclades Islands offshore of

Transatlantic Freshwater Aqueduct

1649

mainland Greece. Dracone technology of 1956, then consisting of sausage-shape rubberized cotton dinghy fabric bags invented by the UK engineer Sir William Rede Hawthorne, was first commercially deployed to service the Greek islands circa 1962. Freshwater shortages still plague the Greek islands and supplemental means, such as dedicated tanker ships and land-based desalination plants on the islands are being considered by authorities (Kaldellis and Kondili 2007). From 2000, another commercial entity, Nordic Water Supply, commenced deploying similar 20,000 m3 floating bags that transport bulk freshwater from Turkey to northern Cypress. Dracones with capacities of ∼35,000 m3 were proposed as supplements to the inadequate, sometimes catastrophically dwindling, freshwater supplies of coastal cities in India (Cathcart 2005). To make freshwater delivery economically effective by this means, the vital-to-life and industry fluid must be transported in large discrete volumes, ideally in quantities >250,000 m3 . A towable plastic or waterproof tensioned textile floatable bag of ∼250,000 m3 would likely be about 350 m long by 72 m wide with a depth thickness of approximately 14 m. Since the density of seawater is nearly 3% more than freshwater, a Dracone or Medusa bag will float anywhere on the ocean’s surface with a freeboard of ∼2.5–3% of the total fabric/film bag’s thickness. In other words, for a flexible fabric/film freshwater-carrying bag at least 14 m thick the maximal freeboard will be ∼420 mm, resulting in a pressure of ∼4.2 kN/m3 exerted on the container’s membrane at the waterline, reducing to nil at its greatest floatation draft below the seawater’s sub-aerial surface interface with air. An investigation of the prospects of a turn-key macro-project for the bulk transfer of freshwater from Turkey to northern Cypress via a ∼750 km-long submerged buoyant pipeline fabricated of high-density polyethylene, to be anchored to the Mediterranean Sea’s bed by vertical anchor wires, was approved by the Turkish Government Decree No. 98/11202 of 27 May 1998. The emplaced seabed anchors (the design characteristics of which were unspecified or undisclosed) were intended to be spaced about 500 m apart. Turkey assigned leadership for the proposed macroproject to Alsim-Alarko A.S., a holding company of a Turkish group of companies which has been trading as “Alarko” since 1972, and the company offered a finished Feasibility Report to the Government by 1999. The Government required the positively buoyant submarine freshwater pipeline, with an inside diameter of 1.48 m, to have the capability and volume capacity to convey gravitationally 75 × 106 m3 /year— that is, a volumetric flow of 2.38 m3 /s. Investigating engineers found that wall-friction is the dominant cause of hydraulic losses and that temporary under-pressurizations and water hammer caused by valve openings/closures do pose major problems to the sustained, successful installed operation of the post-Dracone era pipeline macroproject facility. As of 2009, the planned bulk freshwater transfer facility in the eastern Mediterranean Sea Basin linking mainland Turkey with Cypress remains unbuilt. The macro-engineering project concept of an extended tubular-shaped submerged floating pipeline carrying only bulk freshwater seems to have been invented by McCammon and Lee (1966). The idea was then adopted by Ellis L. Armstrong (1972), briefly surveyed in 1974 by Yuri M. Savvin, popularly revived by Joseph G. Debanne (1975) and the US Department of the Interior’s Bureau of Reclamation, also in 1975, and finally in an ahistorical context by Ernst G. Frankel (1998). Rather surprisingly, none of these individuals, or teams of macro-engineers, ever mentioned the possibility of additional supplemental carrier uses for very longdistance submerged positively buoyant freshwater oceanic pipelines—uses such as,

1650

V. Badescu et al.

for example, undersea telephonic cables or, later, low-mass and high-capacity fiberoptic telecommunications cables. In terms of twenty-first century telecommunications, optical fiber reduces every other transmission medium to insignificance (Huurdeman 2003). There is only one submarine telecommunications cable connecting Argentina in South America with Senegal in Africa via the Cape Verde Islands— the 8,500 km-long Atlantis-2, operational since 1999.

4 TFA Model The TFA might be constructed with assembled lengths of steel pipe. Alternatively, the TFA may consist of an inflatable large-diameter plastic or modern tensioned textile material (Davenport 2004). Such a barely-submerged hose-like pipe would be a double-walled, honeycombed structure of flexible or textile material manufactured as needed in a dedicated land-based factory and/or aboard an efficient fabrication ship at sea. Freshwater should be injected into the cavity between the twin-walled bendable hose-like freshwater conveyances before, finally, the tubular TFA is fully charged with deliverable freshwater. Pressurizing the freshwater will, thus, engorge the pipeline, producing a stiff structure. The steel-made or hose-like TFA will be positively buoyant, and anchored at a floating depth of ∼100 m beneath the Atlantic Ocean surface (Fig. 1). 4.1 Possible TFA Routes The distance between the starting point of the positively buoyant (subsurface floating) seabed-anchored TFA pipe and a given location on the seashore was evaluated by two analytical procedures, i.e. by considering the orthodrome and loxodrome path, respectively. These procedures (which are described in “Appendix”) are useful when the geographical co-ordinates of both geographical locations are known. A third, combined analytical-numerical procedure, was also used to find the distance between the Amazon River freshwater intake location (described by its latitude and longitude) and the closest unloading location on the African land for a given (loxodrome) direction. This procedure uses the accurate Atlantic Ocean bathymetry

Fig. 1 Part of a submerged floating pipeline with anchor lines arranged at intervals and in a V shape and with buoyancy elements arranged for each anchor point. 1 pipe; 2 anchor line; 3 buoyancy element; 4 anchor point

Transatlantic Freshwater Aqueduct

1651

data given by Smith and Sandwell (1997). The estimates by the three procedures, though close to each other, actually do not coincide. Two different routings for the pipeline have been considered (Fig. 2). Nouakchott, Mauritania (18◦ North Lat. by 15◦ West Long.) might be one appropriate unloading terminal for TFA since its contents could fill an overland pipeline planned to be laid across the Sahara (Charlier 1991). If it is created, Charlier’s trans-Sahara canalpipeline, could water farms, gardens and cities shaded from the sun beneath a Sahara Tent Greenbelt (Cathcart and Badescu 2004). Mauritania’s socio-economics, during the twentieth century, was based mainly on extractive industry (Bennoune 1978); installation of the TFA would introduce a major import component to the national economy. If the TFA pipeline starts in the mouth region of the Amazon River at the Equator in South America (0◦ North Lat. by 50◦ West Long.) and its terminus is at Nouakchott (Route 1) that amounts to a distance (on a great circle) of 4,317.18 km. In case the TFA pipeline terminates at Conakry, Guinea (10.57◦ North Lat. by 17.83◦ West Long.; Route 2), the distance (on a great circle) is 3,745.49 km. Public freshwater supplies in Conakry, Guinea were reformed by the early-twenty-first

Fig. 2 Two possible routes for the Transatlantic Freshwater Aqueduct

1652

V. Badescu et al.

century, yet remain rather inadequate and can be enlarged and improved markedly (Clarke et al. 2002). The difference in length between Route 1 and Route 2 is 621 km, which is about 14% of the length of Route 1. Figure 3 shows the bathymetry of the Atlantic Ocean for the two routes (i.e. Route 1: Amazon River mouth–Nouakchott and route 2: Amazon River mouth– Conakry). One can see that Route 2 is more advantageous from the point of view of the free-falling torpedo anchors necessary to stabilize the TFA pipeline. Indeed, the length of the cables for Route 2 is between 3,500 and 4,500 m while for Route 1 the cable length ranges between about 3,000 m and nearly 6,000 m. 4.2 TFA Using a Single Pipe 4.2.1 Pumping Power As a first scenario a TFA consisting of a single pipe will be considered. The pumping power P pump [W] required to force the movement of the freshwater in the pipe is given by: P pump = Qp pipe

(1)

where Q [m3 /s] is the volumetric flow rate, p pipe [Pa] is the linear pressure drop along the pipe (i.e. p pipe ≡ pinlet − poutlet , where pinlet and poutlet are the freshwater pressure at the inlet and outlet of the pipe). Local pressure drops were neglected in Eq. 1. Given the expected unusually large size of the pipe and the large volumetric flow rate, the common Nikuradse and Moody diagrams are of little help in order to assess a realistic friction coefficient for evaluating the specific pressure drop p pipe . Hence, we preferred to simulate the freshwater flow in the duct based on the incompressible isothermal Reynolds averaged Navier–Stokes equations. Steady-state is accepted here. The scalable k-epsilon scheme has been used as closure turbulence model. A few details about the model are given next. The continuity equation is: ∂   vj = 0 ∂xj

Fig. 3 Atlantic Ocean bathymetry for the two selected routes charted in Fig. 1

(2)

Transatlantic Freshwater Aqueduct

1653

where the bar above a quantity means a Reynolds average, x j( j = 1, 2, 3) [m] are spatial coordinates and v j( j = 1, 2, 3) [m/s] are freshwater velocity components. The Einstein summation rule is accepted here and in the following. The conservation of momentum reads:   ∂p ∂  ∂  ρ v jvi = − + tij − ρvi v j (3) ∂xj ∂ xi ∂xj where ρ [kg/m3 ] and p [Pa] are freshwater density and pressure, respectively, the primes denotes fluctuating quantities. The viscous and turbulent stresses, tij [Pa] and τ ij [Pa], respectively, are defined as:   ∂v j ∂v i tij ≡ μ L + (4) ∂xj ∂ xi  τ ij ≡ −ρvi v j = μt

∂v j ∂v i + ∂xj ∂ xi



2 − ρkδij 3

(5)

2

μt ≡ C μ ρ

k ε

(6)

where μ L [Ns/m2 ] is laminar dynamic viscosity, μt [Ns/m2 ] is the turbulent dynamic viscosity, Cμ is the dimensionless turbulent viscosity coefficient and δ ij is Kronecker’s symbol. The transport equations for the turbulent kinetic energy, k [m2 /s2 ], and turbulent dissipation, ε [m2 /s3 ], are, respectively:   ∂   ∂ μt ∂k ρ v jk = Pk − ρ ε + μL + (7) ∂xj ∂xj σk ∂ x j ∂ ∂   ε ρ ε2 + ρ v jε = Cε1 Pk − Cε2 ∂xj ∂ xj k k



μt μL + σε



∂ε ∂xj

 (8)

where σ k and σε are dimensionless fitting coefficients which are determined by adjusting the results to experimental data and Pk [kg/(m s3 )] is the turbulent kinetic energy production term, defined by: Pk ≡ −ρvi v j

∂v i ∂xj

(9)

Note that a more rigorous compressible approach may also be used (which means taking account of the dependence of freshwater density on pressure and temperature). Some preliminary tests show that the assumption adopted here gives a reasonably accurate useful result. A linear segment of pipe of length Lsample = 500 m has been discretized in order to model the pressure drop and freshwater averaged velocity. The pressure drop for the whole pipe was simply obtained by multiplication of the pressure drop for that segment by the number of such segments the pipe of length L [m] would contain.

1654

V. Badescu et al.

Boundary conditions comprise inlet turbulent intensity and inlet mass flow rate  ˙ inlet = ρ f w Q inlet [kg/s], where ρ f w [kg/m3 ] is freshwater density) and null area (i.e. m averaged relative pressure at outlet (i.e. poutlet = 0 Pa). The standard model dimensionless coefficients (Wilcox 1993) have been adopted in calculations σk = 1, σε = 1.3, Cμ = 0.09, Cε1 = 1.44, Cε2 = 1.92

(10)

These values are appropriate for a large number of fully developed turbulent flows including this one, where Reynolds number is well over 106 . Simulation has been performed with a custom designed 3D finite volume Navier–Stokes code (Ferziger and Peric 1995). The volumetric freshwater flow rate Q = 10,000 m3 /s has been considered during all calculations. The linear pressure drop p pipe decreases significantly by increasing the pipe’s inner diameter Dint [m], as expected (Fig. 4a). There are rather slight differences between the two routes of Fig. 1. The corresponding averaged velocity of the fresh water is illustrated in Fig. 4b. It ranges between 1 and 32 m/s. The required pumping power P pump [W] is depicted in Fig. 4c. Pipe diameters larger than 60 m are necessary for “reasonable” low pumping power (i.e., less than 20 GW). 4.2.2 Maximum Pipe Wall Thickness We shall consider a cylindrical pipe element of length L [m] at depth h [m] below sea level. The element is short enough to allow neglecting the axial variation of various physical parameters (Fig. 5a). The pipe wall’s thickness is δ [m]. The force

Fig. 4 a Linear pressure drop p pipe , b freshwater speed and c pumping power P pump as a function of the pipeline inner diameter Dint , for the two routes of Fig. 1

Transatlantic Freshwater Aqueduct

1655

Fig. 5 a Geometry of the pipe and the forces acting on it; b external and internal pressure, respectively

balance on this single pipe element is (the positive sense is oriented towards the ocean surface, local sea level, in Fig. 5a): F A − F p,ext + F p,int − Fw, f w − Fw, pipe − R = 0

(11)

where R [N] is the resultant force, F A [N] is the Archimedean force, F p,ext [N] and F p,int [N] are the forces due to the external and internal pressure, respectively, while Fw, f w [N] and Fw, pipe [N] are the weights of Amazon River freshwater inside the pipe and pipe’s walls, respectively. We denote by ρ sw [kg/m3 ] and ρ pipe [kg/m3 ] mass density of seawater and the pipe wall’s material, respectively. Two hypotheses are accepted now: (1) the pipe wall thickness is much smaller than pipe inner diameter (δ