Desalination Technology for Management Water Issue in California. 15 pages

Desalination 357 (2015) 259–266

Contents lists available at ScienceDirect

Desalination

journal homepage: www.elsevier.com/locate/desal

Environmental and economic assessment of beach well intake versus open intake for seawater reverse osmosis desalination

Maedeh P. Shahabi a,⁎, Adam McHugh a,b, Goen Ho a

a School of Engineering and Information Technology, Murdoch University, Perth, Western Australia, Australia b Infrastructure Advisory, Ernst & Young, 11 Mounts Bay Road, Perth, Western Australia 6000, Australia

H I G H L I G H T S

• Life cycle assessment of the SWRO desalination plant with beach well intake • Life cycle assessment of the SWRO desalination plant with an open intake • Plant with beach well intake results in up to 31% less environmental impact. • Plant with beach well intake results in 13% lower total costs.

⁎ Corresponding author at: Engineering & Inform University, 90 South Street, Murdoch, Western Australia 6

E-mail address: [email protected] (M

http://dx.doi.org/10.1016/j.desal.2014.12.003 0011-9164/© 2014 Elsevier B.V. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history: Received 18 August 2014 Received in revised form 29 November 2014 Accepted 1 December 2014 Available online 10 December 2014

Keywords: Beach well intake Desalination Reverse osmosis Life cycle assessment Open intake Economic analysis

This paper presents a comparative life cycle assessment (LCA) and levelised cost (LC) analysis of two scenarios: an open intake scenario in which a seawater reverse osmosis (SWRO) desalination plant employs an open intake and membrane pre-treatment prior to RO, and a beach well scenario in which feedwater is extracted from the sub- surface using beach well intake and cartridge filtration prior to RO. In both scenarios, desalination plants with 35,000 m3/day capacities were modelled. Results indicate that the beach well intake plant life cycle environmen- tal burdens and LC were as much as 31% and 13% lower respectively, compared with the open intake plant. A de- tailed contribution analysis revealed that the better environmental performance of the beach well intake plant was significantly influenced by its comparatively low electricity use in the simplified pre-treatment process. The better economic performance of the plant with beach well intake was mostly due to savings in chemical use. The results are based on site specific assumptions. However, the LCA and LC framework developed herein could be used to determine the optimum SWRO seawater intake and pre-treatment configuration at plant sites with different characteristics to those modelled herein, provided sufficient data is available.

© 2014 Elsevier B.V. All rights reserved.

1. Introduction

Reliable water supply systems are crucial elements of urban infra- structure, and planning their expansion is challenging. The threat of cli- mate change and global population growth has cast doubt upon the sustainability of traditional water supply sources. This has led to a shift away from reliance on traditional climate dependant supplies such as groundwater and surface catchment dams towards a combina- tion of novel technologies, integrated water sources, water reuse and seawater desalination to provide water security for future urban areas. Seawater desalination provides high quality water. Approximately three billion people — about half of the world's population — live within 200 km of a coastline [1] and 97% of all the water on the planet is saline,

ation Technology, Murdoch 150, Australia. .P. Shahabi).

so seawater is an accessible resource. Currently, reverse osmosis (RO) is the leading technology for desalination [2]. However, there are concerns over its high cost and environmental impacts when compared to tradi- tional water sources.

Life cycle assessment (LCA) has been utilised to explore strategies for reducing the environmental impacts of RO processes, such as mov- ing towards renewable energy inputs [3–8], cleaner fossil fuels [9,10] and plants' size and location optimization [11]. Muñoz and Fernández- Alba [12] quantified the environmental performance improvement that could be obtained by extracting low salinity groundwater instead of seawater for an RO process. Hancock et al. [13] investigated the improvement in environmental performance of coupled seawater desalination and water reclamation by application of new hybrid tech- nologies. The environmental impacts of seawater reverse osmosis (SWRO) with alternative pre-treatment facilities of ultra filtration (UF) and granular media filter have been also reported [14–16]. Howev- er, to the authors' best knowledge, the comparative environmental

260 M.P. Shahabi et al. / Desalination 357 (2015) 259–266

performance of extracting high quality seawater using subsurface in- takes for SWRO has not been previously quantified.

Intake facilities extract and provide feedwater for desalination plants. There are two main categories of intake, namely surface in- takes and subsurface intakes. Surface intakes collect seawater direct- ly from the ocean and deliver seawater to desalination plants while subsurface intakes tap into the saline coastal aquifer under the ocean floor, onshore or off-shore [17]. The quality of seawater ex- tracted from the subsurface is very site specific. There are sites in which seawater extracted via subsurface filtrates naturally through the seabed, and compared to an open intake extraction, the use of the technology at these sites usually results much lower solids, silt, oil and grease, natural organic contamination, and aquatic micro- organisms [18] in the feedwater which leads to simplification of pre-treatment prior to RO and less chemical consumption for mem- brane cleaning [17,19–21]. For example, in Malta and the Caribbean there are numerous subsurface intake technology-based SWRO plants which only use bag filters or cartridge filters ahead of their SWRO membrane systems. This minimal level of pre-treatment is feasible when subsurface intakes are located in a well flushed ocean bottom or shore, away from surface fresh water influence, with seawater collected from a coastal aquifer of uniformly porous structure such as limestone [17]. Without the appropriate site specif- ic conditions, subsurface intake technology could be a costly choice. Factors such as low productivity of the seashore, low subsurface water quality, high concentration of iron or/and manganese or CO2 in feedwater, high variation of source water quality and tempera- ture, and polluted subsurface intake water under influence of con- taminated groundwater all present challenges for SWRO projects with subsurface intakes [19,22]. Thus, to avoid ineffective employ- ment of the technology, site-specific feasibility assessment is essen- tial prior to plant construction [19].

The most common type of subsurface intake for SWRO desalination plants is beach well intake [18,19]. However, there are a number of fac- tors restricting beach well technology as the intake choice for large plants. First, the modular configuration of beach well intake facilities does not deliver the economies of scale enjoyed by open intake facilities [19], making beach well intakes a cost competitive choice only for small- er plants [18]. Second, open intake technology is a better option than beach well technology for large plants due to the limited source water capacity of beach wells [2]. For large plants there is a need for a large number of constructed wells which could disturb a significant area of seashore land and natural habitat, because wells are typically located on seashore within 100 m of the ocean [18]. Despite these technical and economic constraints, previous literature [19] has reported that at numerous sites the environmental performance of beach well intake plants was superior to open intake fed plants due to lower chemical and electricity use in the pre-treatment phase, although these advan- tages were not quantified.

This paper quantifies the environmental and economic perfor- mances of a SWRO plant using beach well intakes under favourable hydro-geological conditions and compares the results to those ob- tained for an open intake plant. Comparative LCA and levelised cost (LC) estimates are made for two SWRO process configurations, these being one open intake scenario and one beach well scenario. For the open intake scenario we focus on a 35,000 m3/day design ca- pacity SWRO desalination plant with an open intake and membrane pre-treatment prior to RO. For the beach well scenario, we again focus on a plant with design capacity of 35,000 m3/day, but in which feedwater was extracted from the subsurface using beach well in- take and filtered by only a cartridge filter prior to RO. Although the results are based on site-specific assumptions, the analytical frame- work detailed below could be readily adapted to assess the compar- ative environmental and economic performances of SWRO intake/ pre-treatment configurations at other sites, such as those with less favourable hydro-geological conditions.

2. Methodology

2.1. Life cycle assessment

The LCA method applied the ISO14040 [23] standard, with the LCA conducted in four stages: goal and scope, life cycle inventory (LCI), life cycle impact assessment, and interpretation. Uncertainty analysis was conducted to assess the influence of variations in process data and model choices on the results. The software SimaPro [24] with connected databases as described in the following sections was used for all LCA modelling and uncertainty analysis.

2.1.1. Goal and scope The goal of this LCA was to quantify and compare the life cycle im-

pacts of the open intake scenario and the beach well scenario. The input and output flows of the SWRO plants were determined by conceptual design and site data. The scope of this study was primarily cradle to gate. The LCA covered the construction and operational phase of both SWRO configurations. The main flows in the operational phase were chemical use, consisting of clean in place (CIP) and chemical enhanced backwash (CEB) processes, materials consumed for membrane replace- ment, and electricity consumption associated with seawater extraction, disc filter (DF), cartridge filter (CF) ultra-filtration (UF) and RO. Disposal of membranes to landfill at the end of their assumed service life was also included in each LCI. Discharged brine to sea was also covered. The same functional unit (1 m3 of desalinated water) was chosen for both scenar- ios to make them comparable. A time boundary of 30 years was selected for both scenarios. The scenarios' system boundaries and input flows are illustrated in Fig. 1.

2.1.2. Life cycle inventory LCI analysis is a key step in LCA, involving the compilation and quan-

tification of inputs and outputs for a given product system throughout its life cycle [23]. Suh and Huppes [25] identify three main categories of LCI method: process-LCI; economic input–output (EIO)-LCI, and hybrid-LCI. Process-LCI tracks the material and energy flows into the system at the process engineering level and is highly detailed, but suf- fers from onerous data requirements that make the modelling of com- plete systems impossible. EIO-LCI quantifies environmental impacts across economic sectors and is able to model a complete economic sys- tem using publicly available data, but lacks engineering detail. In gener- al terms, a hybrid-LCI links together a process-LCI and an EIO-LCI in a manner that removes the weaknesses of each approach while retaining their strengths [25].

In this study, SWRO plant construction phase impacts were accounted with EIO-LCI and operational phase impacts were accounted with process-LCI.

2.1.2.1. Desalination plant construction phase: EIO-LCI. An EIO-LCI model augments a country's economic input–output matrix [26] with a matrix of ecological output of each economy sector to obtain a supply chain of product environmental data. In EIO-LCI, the final inventory vector can be calculated by the following mathematical model [27]:

Q ¼ N � X−1 ð1Þ

A ¼ Z � X−1 ð2Þ

E ¼ Q � I−Að Þ−1 f ; ð3Þ

where N = [nkj] is a matrix of ecological commodity output, nkj indicates the amount of ecological commodity output k associated with the out- put of economy sector j in physical units, X = diag[xi] is a matrix of “Total Output”, xi indicates the total industry output summation of out- put consumed by intermediate industries, final users and exports, X is a

Fig. 1. Graphical overview of the system boundaries and input flows for the SWRO desalination plants, (a) open intake scenario and (b) beach well intake scenario.

261M.P. Shahabi et al. / Desalination 357 (2015) 259–266

diagonal matrix with the elements of the vector x = [xi] strung out along its main diagonal, Z = [zij] is a matrix of interindustry transactions, zij in- dicates the amount of output from industry sector i used by industry sector j in monetary unit, I is an identity matrix, f = [fi] is a vector of final demand, and E = [eij] is an LCI matrix where eij reflects the amount of ecologic output i directly and indirectly associated with delivering a dollar's worth of industry j output to final demand.

In using EIO-LCI, monetary flows are used instead of mass or energy flows as foreground data. The prices (excluding taxes) of components, materials and services need to be obtained for the year in which the country's economic input–output matrix data were collected. The time it takes for national statistics agencies to compile input–output data means that it is normally three to four years old before it can be used for an EIO-LCA. As such, an assumption must be made that the input– output structure of the economy remains stable over the intervening period. In the current study background data was obtained from USA Input output 2002 database for non-residential manufacturing struc- tures sector in SimaPro software [24]. This assumed geographical corre- lation between Australia and the USA is based on Ecoinvent's suggestion that in the case of the lack of LCI data for Australia, North America data- bases could be considered as a replacement because of the countries' similarities in environmental legislations. Construction phase costs were obtained as described in Section 2.2 below and converted to 2002 US$ to be in compliance with USA input–output 2002 database.

2.1.2.2. Desalination plant operational phase: process-LCI. Primary LCI data for electricity, membrane, landfill and brine were generated for two conceptual plant designs using ROSA [28] and ERI [29] software (Fig. 1) for RO processes, manufacturer data, hydraulic calculations for extraction and pre-treatment processes and mass balance equations

for landfill and brine flows. In the open intake scenario, chemical use was based on a SWRO desalination plant with an open intake for feedwater extraction and membrane filtration for pre-treatment locat- ed in Perth, WA [4]. In the beach well scenario, chemical use data was ob- tained through personal communication [30] for a desalination plant in Milos Island, Greece [31] which uses a beach well intake for seawater extraction and a cartridge filter prior to RO. The LCIs for both scenarios' operational phases are provided in Table 1.

LCI background data were obtained from SimaPro databases as follows:

• Electricity production model: Australian high voltage electricity data [32] was used for modelling onsite energy consumption. Grid mix electricity supply and also transmission losses were included in the database.

• Chemical manufacturing: Sugar manufacture was used instead of citric acid. Alkylbenzene sulfonate was used instead of detergent. So- dium tripolyphosphate manufacture was used instead of scale inhibi- tor. Other chemicals were directly selected from the Ecoinvent [33] or Australasian database [32].

• Membrane manufacturing: The processes selected for membrane manufacturing included membrane material production plus extru- sion and injection moulding. All process data were obtained from Ecoinvent [33] or Australasian database [32].

• Transportation: In both scenarios, transportation distances for chemicals, waste and membranes were assumed to be the same as material transportation for a desalination plant in Perth, WA and ob- tained from published document [4]. A twenty tonne articulated truck was selected as the transport vehicle for transportation within Australia [32].

Table 1 Life cycle inventory for both scenarios' operational phases.

Subsystems Input and output flows Design assumptions

Open intake scenario Beach well scenario

Seawater extracted per functional unit (1 m3 of desalinated water)

2.27 m3/m3 2.02 m3/m3 Assumptions included a plant design capacity of 35,000 (m3/day), annual plant capacity factor of 0.85, and annual production of 10.9 × 106 (m3). Recovery ratios were assumed to be 99% for the DF and the CF. The UF recovery ratio was assumed to be 90%. The RO recovery ratio was assumed to be 50%. Feed water quality for designing RO trains is based on Indian Ocean seawater quality [20]; TDS = 40,000 mg/l, temperature = 30 °C. The largest common size for beach well intake plants is selected as plant design capacity.

Electricity use per functional unit (1 m3 of desalinated water) Extraction Open intake (0.05 kWh/m3). Beach well

(0.16 kWh/m3). For the open intake scenario, electricity use was calculated for a single open ocean intake. A 500 m distance between plant and ocean was assumed. Head loss due to friction was assumed to be 3 m/km. The open intake has one duty variable-speed pump and one standby variable-speed pump with efficiencies of 94% and electric motor efficiencies of 82%. For the beach well scenario, electricity use was calculated for five wells with 20 m depth. Five duty wells with daily yield of 15,000 m3/day per well and one standby well were considered. Each well was assumed to be equipped with one variable-speed, 94% efficiency pump and electric motor efficiency of 82%.

Pre-treatment Disc filter (DF), UF, cartridge filter (CF) (0.28 kWh/m3) CF (0.02 kWh/m3) Average pressure drop during operation was assumed to be 0.3 bar for the CF and 0.22 bar for the DF. The UF recovery ratio was assumed to have 3 bar maximum transmembrane pressure. In both scenarios, pumps with an efficiency of 94% and electric motor efficiency of 82% were assumed for UF, DF and CF.

RO Single pass RO with recovery ratio of 50% (2.85 kWh/m3) Single pass RO with recovery ratio of 50% (2.85 kWh/m3)

Single pass RO is the most common practice for SWRO. The RO designed using ROSA software and ERI model [28,29]. Each pressure vessel in the RO system was assumed to contain eight membrane elements connected in series. Both plants were assumed to have 6 RO trains.

Chemical use per functional unit (1 m3 of desalinated water) Extraction and pre-treatment

Intermittent chlorination (sodium hypochlorite; 1.94 g/m3), acid rating (sulphuric acid; 0.69 g/m3), oxidants scavengers (sodium metabisulphite; 0. 07 g/m3), CEB (citric acid; 0.28 g/m3, sodium hypochlorite; 1.63 g/m3) and trace amounts other chemicals.

Scale inhibitor; 2 g/m3. For the open intake scenario, the dosing rate for a desalination plant in Perth, Australia was adopted [4]. The beach well scenario used the dosing rate of the Milos desalination plants in Greece [31], which was obtained through personal com- munication [30].

RO CIP (citric acid; 0.65 g/m3, detergent; 2.72 g/m3, caustic soda; 0.4 g/m3, biocide; 9.86 g/m3).

CIP (citric acid; 0.15 g/m3, detergent; 1.53 g/m3, caustic soda; 0.15 g/m3).

As above.

Membrane use per functional unit

PP; 0.05 g/m3, PE; 0.50 g/m3, PU; 0.14 g/m3, ABS; 1.27 g/m3, PA; 0.14 g/m3.

PE; 0.48 g/m3, PU; 0.08 g/m3, ABS; 0.68 g/m3, PA; 0.14 g/m3.

Polypropylene (PP) is assumed for the production of UF fibre and polyamide (PA) for RO fibre. Polyethylene (PE) was selected for RO and UF protective mesh, acrylonitrile–butadiene–styrene (ABS) for membrane casing and polyurethane (PU) for gluing the elements. The weights of the RO and UF modules were obtained from the manufacturers' data. The weight fraction of these materials in each RO and UF module was adapted from [34] and the authors' engineering judgement. The useful life of the RO and UF membranes was assumed to be 5 years [17].

Landfill and brine Brine; 1.27 m3/m3, disposed membrane to plastic landfill; 2.11 g/m3.

Brine; 1.01 m3/m3, disposed membrane to plastic landfill; 1.38 g/m3.

In both scenarios brine has TDS of 80,000 mg/L.

262 M.P. Shahabi et al. / Desalination 357 (2015) 259–266

2.1.3. Life cycle impact assessment and uncertainty analysis Life cycle impact assessment (LCIA) is the final phase of LCA in which

inventory data is converted into impact results through the use of ap- propriate algorithms or indicators of environmental burdens, aimed at simplified understanding and assessment of the environmental impact of the product system [35]. Based on ISO 14044 [36], LCIA consists of obligatory elements of classification and characterisation and optional elements of normalisation, ranking, grouping and weighting. In this study the characterisation is based on the CML 2001 method [37] which provides ten obligatory impact indicators at a midpoint level.

These impact indicators include: abiotic depletion potential (ADP), acid- ification potential (AP), eutrophication potential (EP), global warming potential (GWP), ozone layer depletion (ODP), human toxicity potential (HTP), fresh water aquatic eco-toxicity (FWAE), marine aquatic eco- toxicity potential (MAETP), terrestrial eco-toxicity potential (TETP) and photochemical oxidation potential (POCP).

In this study, Monte Carlo simulation was employed for uncertainty analysis facilitated by an algorithm in the SimaPro software [24]. A ped- igree matrix was employed for accounting the material and energy flows uncertainty based on the method documented in the Ecoinvent

Fig. 2. Contribution analysis of subsystems to the life cycle impacts, open intake scenario (a) and beach well intake scenario (b).

263M.P. Shahabi et al. / Desalination 357 (2015) 259–266

methodology report [38]. The assigned uncertainty factors of data reli- ability, completeness, temporal correlation, geographical correlation, technological correlation and sample size are listed in supplementary document Table S1.

2.2. Levelised cost

LC is an engineering economics metric that is used for measuring and comparing the costs of alternative projects that deliver similar products. It is the real price at which a long term contract would need to be negotiated in order for a project to breakeven in net present value (NPV) terms. We partition LC into two components, levelised cap- ital cost (LCC) and levelised operational cost (LOC), such that

LC ¼ LCC þ LOC: ð4Þ

LCC is given by

LCC ¼

X n t¼0

Kt 1 þ rð Þt

X n t¼0

Qt 1 þ rð Þt

; ð5Þ

where Kt denotes the capital cost (capex) accruing in year t ($), Qt water production in year t (m3), r the weighted average cost of capital (WACC) — i.e. the rate of return required to service the combined costs of equity and debt — and n the amortisation period.

LOC is given by

LOC ¼

X n t¼0

Vt 1 þ rð Þt

X n t¼0

Qt 1 þ rð Þt

; ð6Þ

Where Vt denotes the operational cost (opex) accruing in year t ($). All capital and O&M costs except electricity, land and labour for both scenar- ios were adopted from the literature [39] and were adjusted to 2013 Aus- tralian dollars (AU$) using exchange rates obtained from Reserve Bank of Australia database [40] and producer price indices obtained from Australian Bureau of Statistics database [41]. Land requirements in hect- ares were obtained from [39] and the unit cost of land was assumed to beAU$300/m2. The amount of electricity use was based on conceptual design (Section 2.1.1) and the wholesale electricity price of AU$143 per MWh was obtained from the literature [42]. The number of full time staff required for routine maintenance and operation was adapted from [39]. Labour cost was calculated based on a 2013 labour rate of AU$68,203 per year. The real WACC of 6.62% proposed by the Western Australian Water Corporation was selected for the LC analysis [43].

3. Results and discussion

3.1. LCA results

3.1.1. Contribution analysis Identifying the key materials and processes with dominant environ-

mental impacts during the life cycle of the system has a significant role in the interpretation phase of LCA [36]. In this study contribution analy- sis is conducted for six subsystems:

1. Electricity use in the treatment process 2. Membrane material use in the pre-treatment and RO processes, in-

cluding material manufacturing and transportation 3. Chemical use in the treatment process, including chemical

manufacturing and transportation 4. Landfill of disposed membrane 5. Brine disposal 6. Construction of buildings and facilities.

As shown in Fig. 2, the LCA results for membrane, landfill and brine disposal indicate that in combination they contributed less than 3% of total environmental impacts in both scenarios across all impact catego- ries. In both scenarios, electricity use in the operational phase contribut- ed more than 75% of the total impacts across all impact categories, with the exception of ODP (~23 to ~26%) where the construction phase was the dominant impact contributor (65% for the open intake scenario and 71% for the beach well scenario). Otherwise, the contribution of plant construction to total environmental impact was relatively small (less than 4% of the total impact), with the exception of the TETP indicator (17% for TETP in the open intake scenario and 22% for TETP in the beach well scenario).

To further identify the differences in both scenarios supply chain, a detailed contribution analysis was conducted at the substance and pro- cess level as illustrated in supplementary document, Table S3. These de- tailed contribution analysis results combined with the process network used for comparing the LCAs of the two scenarios are presented in Table 2.

Zhou et al. [9] conducted a similar, detailed contribution analysis for RO processes with different electricity production models in the United States (US), Spain and Singapore. Comparing results from our study with that previous study revealed that in Spain, the major processes were associated with natural gas and coal burning, which is similar to our results. In the US, producing electricity from hard coal had the highest contribution amongst all impact categories while in Singapore,

Table 2 Main contributors associated with the life cycle impacts of SWRO and benefit of beach well intake over open intake.

Impact categories

Main contributors in supply chain Benefit of beach well intake over open intake

ADP 70% is associated with extraction of black coal and lignite for generating electricity for use in chemical manufacturing and onsite electricity use in SWRO plants.

Less onsite electricity and chemical use in operational phase

AP and EP 87–90% of AP and 73–83% of EP are associated with electricity generation process from black and brown coal fuels, mostly due to the by-product combustion of nitrogen oxides and sulphur oxides in coal fired power plants.

Less onsite electricity and chemical use in operational phase

GWP Greenhouse gas emission emitted from the combustion of black and brown coals in coal fired power plants accounts around 75% of the overall contribution.

Less onsite electricity use in operational phase

ODP The electricity generation process from natural gas power plants accounts for more than 40% of the overall contribution and construction phase (e.g. lighting fixture and machinery manufacturing) accounts for 30% contribution in life cycle impacts.

Less onsite electricity use in operational phase and less construction phase impacts due to simplification in pre-treatment facilities

HTP, FAETP and MAETP

75–80% of HTP, 94–98% of FAETP and MAETP are associated with disposal of hard coal ash to water for electricity production in coal fired power plants.

Less onsite electricity and chemical use in operational phase

TETP 41–57% of TETP is associated with electricity generation process from coal fuels. Emissions of nickel to soil and mercury to air are responsible for more than 50% of TETP. Construction phase (e.g. cement manufacturing, alumina refining and production, and waste management) accounts for 17–20% contribution in life cycle impacts.

Less onsite electricity and chemical use and less construction phase impacts due to simplification in pre-treatment facilities

POCP Emission of sulphur oxides and carbon monoxide to air for electricity production in coal fired power plants accounts for more than 74–78% contribution.

Less onsite electricity use and chemical in operational phase

264 M.P. Shahabi et al. / Desalination 357 (2015) 259–266

heavy fuel oil burning was the major sub-process responsible for envi- ronmental impacts [9].

In summary the contribution analysis revealed that the environmen- tal impacts of the full SWRO supply chain were mainly associated with the emission intensities of the electricity production model. This implies that, by substituting an open intake with beach well intake at suitable sites, life cycle environmental impacts can be reduced as a result of less electricity use in SWRO plant onsite, in upstream chemical manufacturing and in upstream material production for infrastructure construction.

3.1.2. Scenario comparison and uncertainty analysis Fig. 3 shows the normalised impacts of the two scenarios across the

ten impact categories. Each scenario is normalised by the maximum value observed between the two scenarios, so the scenario with the largest environmental impact in each category is normalised as 100%. The absolute values of characterisation results for all impact categories are illustrated in supplementary document, Table S4.

Overall, the results of the LCA indicate that all environmental indica- tors associated with SWRO can be reduced significantly if seawater is extracted through beach well intake instead of open intake facilities. En- vironmental burden reduction of ~10% was observed in ADP, AP, GWP,

Fig. 3. Relative environmental impact of open intake scenario and beach well scenario, with each scenario normalised by the maximum impact value observed across the two scenarios.

MAETP and POCP indicators. These are the indicators which are mostly affected by electricity use in the operational phase. However reductions of up to 31% were observed for the EP, HTP, FEATP, ODP and TETP indi- cators. This is because these indicators are significantly influenced by chemical use in the operational phase.

Uncertainty analysis was conducted to assess the influence of varia- tions in process data and assumptions on the results of comparative LCIA. Fig. 4 shows Monte-Carlo simulation results with a 95% confidence interval for 1000 runs. The percentages illustrate the estimated proba- bility of one scenario achieving a lower impact than the other. The beach well scenario had lower environmental burdens across all impact categories for 57–93% of the runs.

3.1.3. Sensitivity of the environmental benefits to energy source The Australian electricity mix was chosen to model the onsite elec-

tricity use and electricity use for material manufacturing in the LCA analysis as described in Section 2.1.2.2. The Australian energy mix con- sists of approximately 70% coal, 14% natural gas and the remaining 7% is derived from several sources including wind and photovoltaic. In order to investigate the environmental impact sensitivity of the electricity mix assumption, two other electricity mix scenarios were modelled: USA

Fig. 4. Uncertainty analysis results (Monte Carlo simulation with a 95% confidence inter- val, run 1000 times).

Fig. 5. Benefits of beach well intake scenario over open intake scenario in percentage terms for three different electricity mixes.

265M.P. Shahabi et al. / Desalination 357 (2015) 259–266

and Union for the Co-ordination of Transmission Energy (UCTE). Differ- ences between normalised impacts of the two scenarios under three dif- ferent electricity production models across the ten impact categories are presented in Fig. 5. Results suggest that an improved environmental performance may be obtained through the use of the beach well intake configuration in lieu of the open intake configuration under all three en- ergy models. The AP, EP, ODP, TETP, and POCP indicators were more sensitive to changes in the energy production model in than the ADP, GWP, HTP, FATP, and MAETP indicators. The improved environmental performances of the beach well configuration relative to the open intake configuration ranged between 7% and 31% across all environmental in- dicators and energy production models.

3.2. Economic assessment

The LC of produced water was used as an indicator of financial per- formance. Table 3 partitions LC into LCC and LOC sub-categories. For each scenario capex comprises construction, overhead, land and annual membrane replacement costs, while opex comprises all operational and maintenance costs, over a period of 30 years. A more detailed break- down of the scenarios' construction costs is provided in the supplemen- tary document, Table S5.

The LC results range between 2.20 and 2.52 AU$/m3 across the two scenarios. The open intake scenario has a 13% higher LC than the beach well scenario. Moreover, the open intake scenario has a higher LCC and LOC than the beach well scenario. However, the difference in LC across the scenarios is mostly influenced by LOC. The lower LOC in the beach well scenario is due to its lower operational chemical and electricity

Table 3 Cost breakdown of each scenario for functional unit of 1 m3 water of desalinated water.

Levelised cost of activities Open intake scenario

LC (AU$/m3) Ac

Building and facilities' construction (capex) 1.464 5 Overhead cost (capex) 0.162 Land cost (capex) 0.011 Membrane replacement (opex) 0.068 Energy use in treatment (opex) 0.461 1 Labour cost (opex) 0.083 Chemical use (opex) 0.168 Replacement parts and maintenance (opex) 0.068 Insurance (opex) 0.040 LCC (total capex) 1.705 6 LOC (total opex) 0.819 3 LC (total cost) 2.524 10

requirements: a result of the higher quality water extracted from its beach well intake. Contribution analysis shows that LOC is comprised mainly of electricity costs followed by chemical costs in the open intake scenario and labour cost in the beach well scenario.

3.3. Future studies

Given that the defined beach well scenario process configuration is based upon the Milos Island desalination plant in Greece [31] the config- uration we have modelled is likely to be feasible at comparable sites around the globe. However, subsurface feedwater at some sites may have less favourable characteristics than that modelled, such as high concentrations of manganese and/or iron, low dissolved oxygen con- centration, low temperature, high CO2 concentration, MTBE contamina- tion and also the possibility of salinity change over time. Quantification of the environmental and economic performance of SWRO using sub- surface intake under these more challenging conditions deserves fur- ther research.

The current study focuses only on emissions to air, land and water associated with the supply chain of the SWRO desalination process using the CML2001 characterisation method for LCIA. The method ap- plied herein can assist in the development of mitigation strategies targeting those parts of the SWRO supply chain responsible for the larg- est environmental impact contributions. However, there are other po- tential environmental impact categories that may be affected by the employment of beach well or other types of subsurface intakes, for ex- ample, marine life (impingement and entrainment) and infrastructure construction impacts (noise impacts associated with drilling wells off- shore or onshore). Future models would benefit from the inclusion and quantification of such impact categories in the LCA.

Finally, the augmented EIO-LCA method (the top-down approach) was used in the current study to account for the impacts of the SWRO construction phase, during which the ODP and TETP indicators were found to be significant. It is recommended that a detailed and compre- hensive process-LCI be developed for the construction phase of desali- nation plants to help decision makers find possible onsite strategies for reducing these impacts, particularly for subsurface intake construc- tion work such as drilling wells offshore or onshore.

4. Conclusion

Under favourable hydro-geological conditions, SWRO plants com- bining a beach well intake with simplified pre-treatment prior to RO can significantly reduce the environmental impacts of the system at a lower economic cost per unit of desalinated water when compared to a typical open intake and membrane pre-treatment SWRO plant config- uration. Detailed contribution analysis indicates that the better environ- mental performance of the plant with beach well intake is significantly influenced by lower electricity use in its simplified pre-treatment

Beach well scenario

tivities contribution LC (AU$/m3) Activities contribution

8% 1.326 60% 6% 0.147 7% 0% 0.008 0% 3% 0.068 3% 8% 0.432 20% 3% 0.083 4% 7% 0.032 1% 3% 0.068 3% 2% 0.036 2% 8% 1.549 70% 2% 0.651 30% 0% 2.200 100%

266 M.P. Shahabi et al. / Desalination 357 (2015) 259–266

process compared to that required for the membrane pre-treatment process. Moreover, the results suggest that substituting an open intake configuration with a beach well intake configuration can lead to less electricity input requirements throughout the SWRO supply chain (chemical manufacturing, material production and onsite energy use) and therefore a lower environmental impact. Furthermore, LC model- ling suggests that this improved environmental performance can be achieved at a lower economic cost, mainly due to savings in chemical use through the employment of the beach well intake configuration. All of these results are based on site specific assumptions. However, the LCA and LC framework developed herein could be used to determine the optimum SWRO seawater intake and pre-treatment configuration at plant sites with different characteristics to those modelled herein, provided sufficient data is available.

Acknowledgements

The authors would like to thank Dr. Nikos Yfantis from SYCHEM SA who supplied the case study data presented herein and Dr. Martin Anda from Murdoch University for providing valuable comments. Ms. Maedeh P. Shahabi would like to thank the National Centre of Excel- lence in Desalination Australia, funded by the Australian Government through the Water for the Future Initiative (SFR32012MU), for the award of supplementary scholarships. The views expressed in this paper are those of the authors, and are not necessarily the views of the SYCHEM SA, National Centre of Excellence in Desalination Australia and/or Ernst & Young.

Appendix A. Supplementary data

Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.desal.2014.12.003.

References

[1] L. Creel, Ripple Effects: Population and Coastal Regions, Population Reference Bu- reau, Washington, DC, 2003.

[2] L.F. Greenlee, D.F. Lawler, B.D. Freeman, B. Marrot, P. Moulin, Reverse osmosis desa- lination: water sources, technology, and today's challenges, Water Res. 43 (2009) 2317–2348.

[3] M.P. Shahabi, A. McHugh, M. Anda, G. Ho, Environmental life cycle assessment of seawater reverse osmosis desalination plant powered by renewable energy, Renew. Energy 67 (2014) 53–58.

[4] W.K. Biswas, Life cycle assessment of seawater desalinization in Western Australia, World Acad. Sci. Eng. Technol. 56 (2009) 369-375.

[5] R. Raluy, L. Serra, J. Uche, Life cycle assessment of desalination technologies integrat- ed with renewable energies, Desalination 183 (2005) 81–93.

[6] K. Jijakli, H. Arafat, S. Kennedy, P. Mande, V.V. Theeyattuparampil, How green solar desalination really is? Environmental assessment using life-cycle analysis (LCA) ap- proach, Desalination 287 (2012) 123–131.

[7] Z. Norwood, D. Kammen, Life cycle analysis of distributed concentrating solar com- bined heat and power: economics, global warming potential and water, Environ. Res. Lett. 7 (2012) 044016.

[8] R. Salcedo, Multi-objective optimization of solar Rankine cycles coupled with re- verse osmosis desalination considering economic and life cycle environmental con- cerns, Desalination 286 (2012) 358–371.

[9] J. Zhou, V.W.-C. Chang, A.G. Fane, Environmental life cycle assessment of brackish water reverse osmosis desalination for different electricity production models, Ener- gy Environ. Sci. 4 (2011) 2267–2278.

[10] J.R. Stokes, A. Horvath, Energy and air emission effects of water supply, Environ. Sci. Technol. 43 (2009) 2680–2687.

[11] M.P. Shahabi, M. Anda, G. Ho, Influence of site-specific parameters on environmen- tal impacts of desalination, Desalin. Water Treat. (2014) 1–7.

[12] I. Muñoz, A.R. Fernández-Alba, Reducing the environmental impacts of reverse os- mosis desalination by using brackish groundwater resources, Water Res. 42 (2008) 801–811.

[13] N.T. Hancock, N.D. Black, T.Y. Cath, A comparative life cycle assessment of hybrid os- motic dilution desalination and established seawater desalination and wastewater reclamation processes, Water Res. 46 (2012) 1145–1154.

[14] M. Beery, J.-U. Repke, Sustainability analysis of different SWRO pre-treatment alter- natives, Desalin. Water Treat. 16 (2010) 218–228.

[15] M. Beery, G. Wozny, J.-U. Repke, Sustainable design of different seawater reverse os- mosis desalination pretreatment processes, Comput. Aided Chem. Eng. 28 (2010) 1069–1074.

[16] T. Al-Sarkal, H.A. Arafat, Ultrafiltration versus sedimentation-based pretreatment in Fujairah-1 RO plant: Environmental impact study, Desalination 317 (2013) 55–66.

[17] N. Voutchkov, Seawater Pretreatment, Water Treatment Academy, Bangkok, Thailand, 2010.

[18] N. Voutchkov, SWRO desalination process: on the beach–seawater intakes, Filtr. Sep. 42 (2005) 24–27.

[19] T.M. Missimer, N. Ghaffour, A.H.A. Dehwah, R. Rachman, R.G. Maliva, G. Amy, Sub- surface intakes for seawater reverse osmosis facilities: capacity limitation, water quality improvement, and economics, Desalination 322 (2013) 37–51.

[20] M. Wilf, Leon Awerbuch, The Guidebook to Membrane Desalination Technology, Balaban Desalination Publications, L'Aquila, Italy, 2007.

[21] T. Missimer, Water supply development, aquifer storage, and concentrate disposal for membrane water treatment facilities, Methods in Water Resources Evaluation, Schlumberger Water Services, Sugar Land, Texas, 2009. 390.

[22] A.H. Dehwah, T.M. Missimer, Technical feasibility of using gallery intakes for seawa- ter RO facilities, northern Red Sea coast of Saudi Arabia: the King Abdullah Economic City site, Desalin. Water Treat. 51 (2013) 6472–6481.

[23] ISO 14040, Environmental Management—Life Cycle Assessment—Principles and Framework, British Standards Institution, London, 2006.

[24] SimaPro, Pre Consultants, SimaPro 7 Version 7.3.3, PRé Consultants, The Netherlands2011.

[25] S. Suh, G. Huppes, Methods for life cycle inventory of a product, J. Clean. Prod. 13 (2005) 687–697.

[26] W.W. Leontief, Quantitative input and output relations in the economic systems of the United States, Rev. Econ. Stat. (1936) 105–125.

[27] R.E. Miller, P.D. Blair, Input–Output Analysis: Foundations and Extensions, Cam- bridge University Press, 2009.

[28] DOW, The Dow Chemical Company, ROSA System Design Software, Version 9.1, http://www.dowwaterandprocess.com/en/resources/rosa_system_design_soft- ware2013 (accessed 11.04.14).

[29] ERI, Energy Recovery Incorporated, Energy Recovery's Power Model, http://www. energyrecovery.com/power-model2013 (accessed 11.04.14).

[30] N. Yfantis, Personal communication, SYCHEM SA, Greece in: M.P. Shahabi (Ed.), Perth, Australia, 2014.

[31] Ν. Yfantis, D. Yfantis, G. Roupakas, S. Liodakis, Optimization of reverse osmosis desa- lination plants design — case studies, Αqua 08 Water Science and Technology Con- ference, Athens, Greece2008.

[32] LCS, Life cycle strategies, Australasian LCI database, Version 2013.01, http://www. lifecycles.com.au/#!australasian-database/cbm52013 ((accessed 11.04.14) in, Australia).

[33] SCLCI, Swiss Center for Life Cycle Inventories, Ecoinvent Version 2.0 Database, http://www.ecoinvent.org/database/2011 ((accessed 11.04.14), in, Switzerland).

[34] N. Tangsubkul, K. Parameshwaran, S. Lundie, A.G. Fane, T.D. Waite, Environmental life cycle assessment of the microfiltration process, J. Membr. Sci. 284 (2006) 214–226.

[35] R. Horne, T. Grant, K. Verghese, Life Cycle Assessment: Principles, Practice, and Pros- pects, CSIRO, 2009.

[36] ISO 14044, Environmental management–Life cycle assessment–Requirements and guidelines, 2006.

[37] CML, CML-IA Characterisation Factors, 2013. [38] R. Frischknecht, N. Jungbluth, H.-J. Althaus, G. Doka, T. Heck, S. Hellweg, R. Hischier,

T. Nemecek, G. Rebitzer, M. Spielmann, G. Wernet, Overview and Methodology. Ecoinvent Report No. 1, Swiss Centre for Life Cycle Inventories, Dübendorf, 2007.

[39] I.C. Watson, O. Morin Jr., L. Henthorne, Desalting Handbook for Planners, United States Department of the Interior, Denver, 2003.

[40] RBA, Reserve Bank of Australia, Exchange Rate Data, Monthly Data, http://www.rba. gov.au/statistics/hist-exchange-rates/index.html2013 (accessed 11.04.14).

[41] ABS, Australian Bureau of Statistics, Producer Price Indexes, Australia, Table 12, http://www.abs.gov.au/AUSSTATS/[email protected]/DetailsPage/6427.0Dec%202013? OpenDocument2013 (accessed 11.04.14).

[42] ERA, Economic Regulation Authority, Synergy's Costs and Electricity Tariffs, http:// www.parliament.wa.gov.au/publications/tabledpapers.nsf/displaypaper/ 3814966aaf7f801e28d2f56748257a53002ac89a/$file/4966.pdf2012 (accessed 11. 04.14).

[43] ERA, Economic Regulation Authority, Inquiry Into the Efficient Costs and Tariffs of the Water Corporation, Aqwest and the Busselton Water Board, http://parliament. wa.gov.au/WebCMS/WebCMS.nsf/resources/file-tp—water-inquiry/$file/ Water+Inquiry+−+Final+Report+for+Website.pdf2013 ((accessed 11.04.14), in, Western Australia).