LETTER • OPEN ACCESS
Desalination and sustainability: a triple bottom line study of Australia
To cite this article: Michael Heihsel et al 2020 Environ. Res. Lett. 15 114044
View the article online for updates and enhancements.
This content was downloaded from IP address 130.149.177.113 on 02/12/2020 at 10:29
Environ. Res. Lett. 15 (2020) 114044 https://doi.org/10.1088/1748-9326/abbd63
Environmental Research Letters
OPEN ACCESS
RECEIVED
11 July 2020
REVISED
3 September 2020
ACCEPTED FOR PUBLICATION
1 October 2020
PUBLISHED
18 November 2020
Original content from
this work may be used
under the terms of the
Creative Commons
Attribution 4.0 licence.
Any further distribution
of this work must
maintain attribution to
the author(s) and the title
of the work, journal
citation and DOI.
LETTER
Desalination and sustainability: a triple bottom line study of
Australia
Michael Heihsel1, Manfred Lenzen2and Frank Behrendt1
1EVUR, Department of Energy Engineering, Technical University of Berlin, Berlin, Germany
2ISA, School of Physics, The University of Sydney, Sydney, New South Wales 2006, Australia
E-mail: [email protected]
Keywords: LCA, desalination, renewable energy, sustainability, input-output analysis
Supplementary material for this article is available online
Abstract
For many arid countries, desalination is considered as the final possible option to ensure water
availability. Although seawater desalination offers the utilisation of almost infinite water resources,
the technology is associated with high costs, high energy consumption and thus high carbon
emissions when using electricity from fossil sources. In our study, we compare different electricity
mixes for seawater desalination in terms of some economic, social and environmental attributes.
For this purpose, we developed a comprehensive multi-regional input-output model that we apply
in a hybrid life-cycle assessment spanning a period of 29 yr. In our case study, we model
desalination plants destined to close the water gap in the Murray-Darling basin, Australia’s major
agricultural area. We find that under a 100%-renewable electricity system, desalination consumes
20% less water, emits 90% less greenhouse gases, and generates 14% more employment. However,
the positive impacts go hand in hand with 17% higher land use, and a 10% decrease in gross value
added, excluding external effects.
Abbreviations
FTE Full-time equivalents
GHG Greenhouse gas
GVA Gross value added
ha Hectare
hLCA Hybrid life-cycle assessment
IELab Industrial Ecology Virtual Laboratory
IO Input-output
IOT Input-output table
kha Kilo hectare
LCA Life-cycle assessment
MDB Murray-Darling basin
Mha Million hectares
MRIO Multi-regional input-output
RE Renewable electricity
RO Reverse osmosis
TBL Triple bottom line
1. Introduction
Water scarcity affects an increasing proportion of the
world’s population (Greve et al 2018). In Australia,
water shortages have intensified over the past two dec-
ades (Aijm et al 2013). In the ‘granary of Australia’,
the MDB, little precipitation and water over-use has
led to environmental issues like high salinity of rivers
and fish death (Potter et al 2010, Wedderburn et al
2012). Over the past few years, Australia faced heat
records and intense bushfires (Borchers Arriagada
et al 2020). The latter put the country in a state of
emergency for weeks in early 2020 (Komesaroff and
Kerridge 2020).
If using natural resources, improving water man-
agement, and measures like wastewater reuse are
not sufficient to meet water demand, desalination
offers great potential, especially for regions with
access to the sea (Crisp 2012, Bell et al 2018). How-
ever, the technology has some severe drawbacks.
The production cost of desalinated water is about
twice or three times higher than water from con-
ventional sources (Ziolkowska 2015). Furthermore,
effects on the marine ecosystem, high energy con-
sumption and the associated GHG emissions are the
primary ecological challenges (Sadhwani et al 2005,
Stokes and Horvath 2006, Lattemann and Höpner
2008, Shehabi et al 2012, Shahabi et al 2014, Liu et al
2015, Zarzo and Prats 2018, Clark et al 2018, Gude
and Fthenakis 2020). The use of RE could make a
major contribution to environmental sustainability
© 2020 The Author(s). Published by IOP Publishing Ltd
Environ. Res. Lett. 15 (2020) 114044 M Heihsel et al
in particular (Jijakli et al 2012, Baten and Stum-
meyer 2013, Cherif et al 2016, Alhaj and Al-Ghamdi
2019). However, studies that assess total sustainabil-
ity by measuring environmental, social and economic
indicators are missing (Haddad 2013, Gude 2016).
The motivation of this study is to quantify the
three dimensions of sustainability of desalination
depending on the used electricity source. There-
fore, we apply an LCA approach (Malik et al 2016,
Hadjikakou et al 2019) to measure supply chain
effects. We applied an IO-based hLCA for this study
(Joshi 1999). The IO analysis goes back to the research
work of Wassily Leontief, who received the Nobel
Prize for this in 1973 (Leontief 1966). We applied
the Australian IELab to compile tailor-made input-
output MRIO tables (Lenzen et al 2014).
In this study, we simulated fictive desalination
plants at 29 sites around the MDB in southeast
Australia. The desalination plants were designed to
provide the missing water supply in the MDB of
5500 GL in total. The plants were not designed
for continuous operation, but take into account an
oversize factor of 1.5, allowing load-shifting of the
electricity demand. Like all large plants in Aus-
tralia, the plants are designed as RO systems. For
the power supply, we considered five scenarios, with
0, 25, 50, 75 and 100% RE. In the 0% RE scen-
ario, the electricity sector corresponds to the gen-
eration mix of the corresponding year in Australia,
as shown in the IO data. The electricity mix, as
well as the locations of the generators in the 100%
RE scenario, were the results of a GIS-based dis-
patch optimisation model of a previous study (Heih-
sel et al 2019a). The transitional scenarios apply pro-
rata combinations of both electricity mixes. The capa-
cities of the technologies wind, biomass, hydro and
PV are shown in the table in supplementary mater-
ial S1, which can be found online (available online at
https://stacks.iop.org/ERL/15/114044/mmedia).
To the best of our knowledge, this study is the first
comprehensive MRIO TBL study comparing desalin-
ation plants in RE and fossil-fuelled electricity scen-
arios. Thus, we contribute to the research gap in the
area of holistic studies on socio-economic and envir-
onmental impacts of desalination. The study is struc-
tured as follows: In the next chapter, we describe the
methodology and the data used. After that, we present
our results and end with a conclusion. Further tech-
nical details on the methods used can be found in the
supplementary material.
2. Methods and data
Two methods are applicable for carrying out LCAs:
the bottom-up approach (a process-based LCA) and
the top-down approach (based on IOTs) (Finnveden
et al 2009). The bottom-up approach employs phys-
ical process data, allowing very accurate modelling of
immediate upstream stages of the value chain. Since
the processes are explicitly modelled, the value chain
is only reflected to a limited extent due to data avail-
ability, and subordinate levels are not considered.
Therefore, a truncation error occurs. The top-down
approach, on the other hand, uses statistical data on
economic sectors, which allows infinite value chains
to be modelled. A truncation error, therefore, does
not occur, but an aggregation error arises from the
aggregation of different processes in industrial sec-
tors. The combination of top-down and bottom-up
data in a hybrid approach minimises both errors,
which is the advantage of this method (Pomponi and
Lenzen 2018).
Hybrid LCAs are used for carbon footprint stud-
ies of products, companies or sectors by extending
IO tables with physical environmental satellites (Liu
et al 2012, Norwood and Kammen 2012, Rodríguez-
Alloza et al 2019, Heihsel et al 2019b). Numerous
sustainability studies expand the focus on further
economic and social indicators, the so-called TBL
(Elkington 1998, Foran et al 2005, Onat et al 2014,
Malik et al 2016, Hadjikakou et al 2019). The present
framework builds on previous research from Heihsel
et al (2019b).
We extended an existing IO model as follows:
In addition to the GHG satellite, we implemented
additional indicators, namely water use, land use,
employment, and GVA. Furthermore, we extended
the regional resolution up to 46 regions. The highly
detailed regional resolution enables to aggregate the
results according to different regional classifications
(Australian states and territories, water catchment
and rainfall areas). We increased the time series to
the years 1990–2018. Since the existing IO data do
not contain RE sectors, we augmented the tables with
additional process data. By using hybridised process-
data from Yu and Wiedmann (2018), we modelled
RE sectors for wind, solar, hydro and biomass tech-
nologies. Hereby, we analysed the TBL impacts of
the construction and operation period of desalination
plants. In this study, we followed the IO-based hLCA
approach by Malik et al (2014) and Suh and Huppes
(2005). A projection of the IO data and the process
data into the future would be possible in principle but
would be subject to considerable uncertainty when
analysing investments spanning decades.
2.1. Input-output and renewable electricity data
To complement the IO data with RE specific data, we
used process data from AusLCI. (2020) and Ecoin-
vent (2014), which Yu and Wiedmann (2018) have
hybridised. Yu uses an integrated hLCA framework
that includes monetary and physical data. Yu’s hLCA
framework contains 4463 processes (physical data)
and 1284 monetary IO sectors, both in matrix form.
For our study, we extracted data from the process-
coefficient matrix (the primary life-cycle inventor-
ies (LCI) of the processes). Moreover, we utilised the
2
Environ. Res. Lett. 15 (2020) 114044 M Heihsel et al
cut-off matrix, which supplements the pure process-
based data with IO data. Both matrices describe
the production recipe of the technology. While the
process-coefficient matrix shows upstream physical
pre-processes as inputs to produce a functional unit
of the process, the cut-off matrix primarily contains
upstream services. Each process is represented by an
individual column in the matrices. S2 in the supple-
mentary material shows the structure of the integ-
rated hLCA framework by Yu. The processes extracted
for this study are shown in S3. In our hLCA frame-
work, we augmented four RE technologies, namely
wind, biomass, hydro and photovoltaic.
Furthermore, adding new sectors to the IO-
framework requires physical data for the satellite
accounts. Therefore, we collected data for the envir-
onmental indicators GHG, water use and land use as
well as for the social indicator employment. The elec-
tricity generation processes in Yu’s LCI have a func-
tional unit of 1 MJ. Hence, we normalised the satellite
intensities accordingly. Table 1summarises the dir-
ect intensities of the considered RE technologies. The
RE vectors represent the intermediate consumption
of the electricity generated and thus take into account
both operation and maintenance as well as construc-
tion, the latter weighted according to expected life-
time. The construction of RE plants is thus reflected
in the indirect effects on an annual basis. The dir-
ect intensities of RE thus only refer to direct electri-
city generation. In the supplementary material S1, we
show the LCI data preparation process before aug-
menting the IO-tables.
2.2. Input-output data
We used the Australian IELab to compile tailor-made
IO-tables for our study (Lenzen et al 2014). The IELab
offers a unique disaggregation level of 1284 IOPC sec-
tors and 2214 SA2 regions. For our study, we cre-
ated a framework of 64 IO-sectors, including the
four augmented RE sectors. Our framework consists
of 46 Australian regions. The framework contains
both industries and commodities, which means that
the intermediate demand framework has a size of
2×64 ×46 =5888 rows and columns. The IELab
uses statistical data from ABS (2015f), ABS (2015c),
ABS (2015b), ABS (2015d), ABS (2015e) for the IO
data and AGEIS (2015), ABS (2015g), ABS (2015a),
ABS (2015h) for the social and environmental data,
respectively. We created time series IOTs for the years
1990–2018.
2.3. Sector augmentation and re-balancing
Australian IOTs do not include separate RE sectors.
Therefore, we post-augmented the IOTs with process
data. Columns in IOTs show the input of the pro-
duction of a particular industry; in other words, it
is the recipe of the manufactured commodities. The
rows describe the intermediate sales structure of com-
modities to other industries. While we supplemented
columns with the LCI data, we adopted the rows (i.e.
sales structure) from the structure of the existing elec-
tricity sectors. The location of the generators resulted
from Heihsel et al (2019a).
We compiled separate IOTs for each of the five
scenarios with 0, 25, 50, 75 and 100% RE generation.
The RE sectors were scaled accordingly. The conven-
tional electricity sectors were scaled according to the
complementary value. A schematic diagram of the
augmented IOT-framework can be found in the sup-
plementary material S4.
Due to the augmentation of the IOTs, they were
no longer balanced, which means that input and out-
put were no longer equal. We used a RAS-type bipro-
portional numerical algorithm to post-balance the
IOTs (Lahr and de Mesnard 2004, Malik et al 2014).
2.4. Desalination process-data
Heihsel et al (2019b) analysed Australia’s largest 20
desalination plants, which correspond to 95% of the
Australian seawater desalination capacity. For their
study, they used process-data from the desaldata data-
base (Global Water Intelligence 2016). We used these
data as a weighted-average proxy for the 29 desal-
ination plants. We applied results from Heihsel et al
(2019a) to specify the capacity and the locations
of the plants. We assumed the construction period
between 1990 and 1992. We allocated 25% of the cap-
ital expenditures (capex) each to the first and the
last year. Thus 50% of the capital costs were taken
into account for the second year. The operation and
maintenance period runs continuously from 1993.
Since the desalination plants were used for load shift-
ing scenario in this study, we considered an oversize
factor of 1.5.
2.5. Calculation of the triple bottom line impacts
The TBL framework describes an accounting concept
in which all three fields of sustainability are examined.
The term was first introduced by Elkington (1998). In
our analysis, we assessed water use, land use and GHG
emissions as environmental indicators. The social
indicator is employment and the economic indicator
GVA. To measure the supply chain impacts of the
choice of electricity source on the sustainability of
seawater desalination, we used the standard IO meth-
odology, which goes back to Leontief (1966). In the
following section, we present the basic principles of
the methodology.
Let Q be the matrix of the physical satellite
accounts containing the social and environmental
indicators. The same calculations were applied
accordingly to GVA as an economic indicator. Each
row within the matrix represents another indicator.
For each indicator row i, we got the vector of the
intensities by dividing by outputs with
qi=Qi
b
x−1,(1)
where b
xrepresents the diagonal matrix of the output.
3
Environ. Res. Lett. 15 (2020) 114044 M Heihsel et al
Table 1. Direct intensities of renewable electricity generation.
Wind Biomass Hydro Photovoltaic
Water use (L MJ−1) 0.00 1.41 7.22 1.05E-03
Land use (sqm MJ−1) 3.28E221204 1.75E-05 2.16E-03 1.08E-04
Greenhouse gases (g CO2,e MJ−1) 0.00 2.10 1.60 0.00
Employment (FTE MJ−1) 4.53E-08 3.15E-07 2.92E-08 3.85E-07
The standard Leontief equation estimates the out-
put xof the sectors depending on the final demand y
by
x= (I−A)−1y,(2)
where Iis the identity matrix and Arepresents the
technical coefficient matrix. The matrix of technical
coefficients representing the production recipe of the
sectors is defined by A=Tb
x−1The Leontief inverse
L= (I−A)−1contains the supply chain multipliers.
We obtained the supply chain impacts Qdi(k×l)
from (1) and (2) by
Qd
i=b
qi(I−A)−1b
y.(3)
Each value in row kand column lin Qdishows the
contribution of industry kor the product lto the total
magnitude of indicator i.
The GVA impacts were determined accordingly
to (3), using the GVA intensities vinstead of q.
The results can be aggregated into two different rep-
resentations. The formula 1Q′Qd
ishows the impacts
caused by the production of commodities. Sum-
ming the rows by Qd
i1Qaggregates the impacts by
emitting industries. The operator ‘transposes the
summation vector 1QQdi. Electricity is distributed
to the conventional electricity sector and to the
RE sectors by the electricity penetration ratio λ=
0,25,50,75or100%. We allocated the total electri-
city demand ewith ec= (1−λ)e to the conventional
electricity sector and with et=λegt
∑4
t=1gtto the RE
sector in the demand vector y. Here, g is the gener-
ation of the RE technology t.
2.6. Uncertainties
Prior LCAs similar to ours (e.g. Malik et al 2018)
have shown that measurement uncertainty contained
in information on the satellite account Qand the
supply-use data Trepresent the main origins for
uncertainties in aggregate results, such as the TBL
scores calculated in this work (Malik et al 2019).
Because of nonlinearities in Leontief’s equation (2),
the latter are usually determined using Monte-Carlo
simulation (Bullard and Sebald 1977,1988, Len-
zen et al 2010). Because of particular features in
the propagation of errors in the Leontief system
(Heijungs and Lenzen 2014), high errors of individual
matrix elements cancel each other out because of their
stochastic nature (Quandt 1958, Lenzen 2000), and
the uncertainties of aggregate results are usually much
smaller than matrix errors. Whilst the latter occupy a
wide range between typically a few and a few hundred
percent (Bullard and Sebald 1977,1988, Lenzen et al
2010), the former hover between about 5% and 20%
(see e.g. Lenzen et al 2018 and Lenzen et al 2020).
Given the high similarity of data sources and math-
ematical procedures, these uncertainty magnitudes
also apply to this study.
3. Results and discussion
3.1. Holistic triple bottom line impacts
In our study, we investigated the TBL impacts of sea-
water desalination depending on the electrical energy
supply on the environmental indicators water use,
land use and GHG. We used GVA as an economic
indicator and employment as a social indicator.
Table 2shows the overall results of the TBL foot-
prints from 1990–2018, for the scenarios with only
conventional electricity and 100% RE. The first three
years represent the construction period of all plants,
the years from 1993 onwards represent the operating
and maintenance period over 26 yr, which is in the
average range of the concession periods of large Aus-
tralian plants (Global Water Intelligence 2016). The
spider plots in figures 1and 2show the relative per-
formance of the different scenarios in relation to the
0% RE scenario. The values of the charts are calcu-
lated by summing the indicator values of each scen-
ario and relating them to the sum of the indicator
values of the 0% RE scenario. The values of the 0%
RE scenario are thus always 1, i.e. they set the bench-
mark. In order to show better performance consist-
ently greater than 1, the reciprocal value is formed for
indicators for which less is better (for example, GHG,
land use, water use). Hence, the normalised index nk,s
for key figure kand scenario s(where scenario 0 is
the base scenario with 0% RE) in the spider plots res-
ult from nk,s=∑29
y=1ik,s,y
∑29
y=1ik,0,yfor the key figure employ-
ment and GVA (where more is better) and from nk,s=
∑29
y=1ik,0,y
∑29
y=1ik,s,yfor water use, land use and GHG emissions
(where less is better) with yas year of investigation
and ias the indicator values. Hence, the spider plots
indicate better performances compared to the base
scenario outside and weaker performances inside the
blue base scenario circle.
We see that RE have a positive impact on water
consumption during the entire life cycle. In both
scenarios, the construction phase accounts for around
4
Environ. Res. Lett. 15 (2020) 114044 M Heihsel et al
Table 2. Overall results of the TBL footprints of desalination with 0% RE and 100% RE.
Water use Land use Greenhouse gas emission Employment Gross value added
RE-ratio 0% 100% 0% 100% 0% 100% 0% 100% 0% 100%
Year (GL) (GL) (kha) (kha) (Mt CO2, e) (Mt CO2, e) (103FTE) (103FTE) (AU$ bn) (AU$ bn)
1990 443 414 5747 5643 50 15 273 272 13 13
1991 912 852 11 668 11 470 105 31 533 531 26 26
1992 439 411 5613 5511 51 15 255 253 13 13
Total 1794 1677 23 029 22 623 206 60 1061 1056 52 52
Construction
Contribution 43% 51% 77% 65% 23% 70% 72% 63% 42% 47%
1993 114 73 582 842 32 1 24 29 2.1 1.7
1994 109 70 534 785 31 1.4 23 29 2.1 1.6
1995 105 69 493 752 30 1.3 22 29 2.0 1.6
1996 103 68 461 723 30 1.3 20 28 2.1 1.7
1997 101 66 434 663 30 1.2 20 28 2.1 1.7
1998 97 65 370 625 29 1.2 18 26 2.0 1.7
1999 88 64 327 592 27 1.1 16 25 1.9 1.6
2000 89 63 299 552 27 1.1 16 25 2.0 1.6
2001 85 63 273 563 26 1.1 15 25 1.9 1.7
2002 94 66 276 620 29 1.1 15 25 2.0 1.7
2003 87 62 272 576 27 1.0 12 22 2.1 1.8
2004 85 56 156 378 29 1.0 12 22 2.1 1.8
2005 64 57 174 377 19 1.1 14 24 2.2 2.1
2006 63 58 171 339 20 1.1 14 23 2.3 2.2
2007 59 43 190 237 18 0.9 13 20 2.4 2.2
2008 60 42 187 234 19 0.9 12 20 2.6 2.4
2009 62 44 191 237 20 0.9 13 20 2.8 2.6
2010 56 41 199 246 20 0.8 15 21 3.0 2.6
2011 67 54 94 241 24 0.9 11 18 3.2 2.7
2012 103 61 94 273 27 0.8 12 20 3.3 2.8
2013 115 70 129 269 31 0.7 15 22 4.0 3.0
2014 113 66 136 280 29 0.7 16 22 4.1 3.1
2015 96 65 133 287 28 0.7 13 22 4.0 3.1
2016 90 70 190 473 25 0.7 14 21 3.6 2.9
2017 108 73 219 499 31 0.7 16 23 3.8 3.0
2018 121 71 239 494 35 0.8 18 23 4.2 3.2
Total 2335 1604 6821 12 156 692 26 409 613 70 58
Operation and maintenance
Contribution 57% 49% 23% 35% 77% 30% 28% 37% 58% 53%
5
Environ. Res. Lett. 15 (2020) 114044 M Heihsel et al
1.00
1.02
1.04
1.06
1.08
1.10
0.980.991.001.011.021.03
1.00
1.50
2.00
2.50
3.00
3.50
0.98
0.99
1.00
1.01
1.02
1.03
0.98
0.99
1.00
1.01
1.02
1.03
Water
Land Use
Greenhouse GasesEmployment
Gross Value Added
0% RE
25% RE
50% RE
75% RE
100% RE
Figure 1. Construction period TBL performances of desalination utilising different electricity mixes.
0.50
0.70
0.90
1.10
1.30
1.50
0.500.600.700.800.901.00
1.00
7.00
13.00
19.00
25.00
31.00
1.00
1.20
1.40
1.60
1.80
2.00
0.50
0.60
0.70
0.80
0.90
1.00
Water
Land Use
Greenhouse GasesEmployment
Gross Value Added
0% RE
25% RE
50% RE
75% RE
100% RE
Figure 2. Operation and maintenance period TBL performances of desalination utilising different electricity mixes.
half of the total water consumption. The use of 100%
RE would reduce water consumption by 31% in the
operating phase and by 7% in the construction phase
of desalination plants. Thermal power plants require
vast quantities of water due to their technical concept.
In particular, the cooling required for the thermody-
namic process is highly water-intensive. In coal-fired
power plants, the amount of water required to gen-
erate electricity can, therefore, account for up to ten
times the weight of the coal required (Gleick 1994).
Further water is required for the post-treatment of
ash and waste disposal. Indirectly, coal mining and
recultivation of the landscape are water-intensive. In
contrast, the direct water consumption of PV and
6
Environ. Res. Lett. 15 (2020) 114044 M Heihsel et al
wind is negligible. Although the water consumption
of hydro and geothermal electricity by evaporation
and cooling is also substantial, PV and wind dominate
the 100% RE system in this case study, which is why
water consumption would be significantly reduced.
For the construction, operation and maintenance
of the 29 seawater desalination plants, approx. 3300
GL of water would be needed over the entire period
of 29 yr when using RE. In contrast, the plants would
produce 5500 GL of water per year, i.e. over the estim-
ated life cycle of 29 yr, water consumption would
amount to 2.3% of the total amount of water pro-
duced. With conventional energies, the water con-
sumption share is 2.9%. Over the total period, 21%
of water use could be saved by using RE.
Within the construction phase, the land use of
desalination would average around 7.5 Mha per year.
Moreover, the construction phase is the driver in land
use with 77% contribution when using conventional
electricity. The use of RE during the construction
phase, which would reduce land consumption by 2%,
has only a marginal impact. If RE is used in the opera-
tional phase, land use would increase significantly by
78%.
RE would have the most significant positive
impact on carbon emissions from desalination plants,
both during construction and operation. In detail,
206 Mt CO2,e would be generated by the construction
using the Australian electricity mix. Only 60 Mt CO2,e
would be emitted if Australia had a 100% RE grid.
The savings correspond to a reduction of 71%. An
even more significant reduction in emissions could be
achieved in the operating phase. While 692 Mt CO2,e
is emitted in the 0% RE scenario, we see a reduc-
tion to 26 Mt with 100% RE. Consequently, carbon
emission could be reduced by 96% in the operational
period. In the base scenario, the operating phase con-
tributes 77% to carbon emissions. In the RE scen-
ario, the construction phase is the main contributor,
with 70% of the total emissions. The results thus show
that further measures are needed in other sectors
to further reduce emissions during the construction
phase.
Regardless of the power source, the average
employment during the construction period would
be over 350 000 FTE per year. The construction
period therefore contributes significantly to total
employment, accounting for around 72% of total
jobs. While employment in the 100% RE scen-
ario would decrease slightly during the construc-
tion phase, employment during the operation phase
would increase significantly to 50%. Hence, jobs
would increase by 50% from an average of around
16 000 FTE to around 24 000 FTE per year when using
100% RE.
The construction and operating periods have a
similarly high contribution to GVA. In both scen-
arios, the construction of the 29 plants generates GVA
of around AU$52 bn (current prices). If 100% RE
were used, GVA would be significantly reduced by
17.5% during the operation and maintenance period.
Only AU$58 bn instead of AU$70 bn would be gen-
erated. Overall, this leads to a reduction of around
10% of GVA over the entire life cycle of 29 yr. This
decline can be explained by the fact that Australia is
mining a large proportion of its conventional fuels
domestically. However, this calculation does not take
the external costs of using conventional energy into
account.
3.2. Sectoral implications
In the following section, we identified the sectors
that have a significant impact on the indicators.
Figure 3shows the percentage contribution of the
different sectors to the overall impact. The upper
diagram shows the contribution of each industry
where the impacts occur. The lower diagram shows
the commodities that trigger the impacts through
their demand. The bars compare the 0% RE with the
100% RE scenario. The electricity industry is mainly
responsible for water consumption in both scenarios.
Furthermore, the decrease in water consumption in
the 100% RE scenario is mainly caused by the elec-
tricity industry and to a lesser extent by the min-
ing industry. In contrast, the water consumption of
the agricultural industry increases with the increasing
share of RE in the electricity grid, mainly due to the
use of biomass. Moreover, electricity is the most relev-
ant commodity determining water consumption. The
decrease in water consumption with 100% RE use is
also driven by this commodity.
Land use in both scenarios is dominated by the
agricultural industry. The same industry causes addi-
tional land use when increasing RE share. While the
contribution of the commodity electricity to land use
is still relatively small when fossil-based electricity is
used, it would more than double if RE were used. The
main contributing commodities for land use are civil
engineering services and installations, the construc-
tion of intakes and outfalls and the manufacturing of
equipment and materials.
The electricity industry is the driving force for
GHG emissions when desalination is operated and
built by utilising conventional electricity. This pic-
ture changes with a higher share of RE so that the
manufacturing industry plays the most considerable
role. If we examine the commodities, a similar picture
emerges. Electricity as a commodity, but also the con-
struction of intake and outlet play a significant role in
GHG emissions. Both commodities reduce their con-
tribution in increasing the share of RE.
The manufacturing industry substantially con-
tributes to employment. A transition to 100%
RE would not change the contribution. However,
the electricity sector becomes more critical in the
100% RE scenario, making it the second most job-
relevant industry. When looking to commodities, the
production of equipment and materials is primarily
7
Environ. Res. Lett. 15 (2020) 114044 M Heihsel et al
Figure 3. Percentage impact contributions of industries and commodities, comparing the 0% RE and 100% RE scenario.
responsible for job creation. The contribution of the
electricity commodity would also increase signific-
antly due to a higher RE-share.
The electricity industry is the largest contrib-
utor to GVA, but the contribution decreases as RE
increases. With an increasing share of RE, the mining
industry would also reduce its contribution to GVA.
On the other hand, the construction industry would
increase its GVA contribution by increasing the share
of RE. In all scenarios, the second most important
industry sector in terms of GVA is the manufacturing
industry. In both scenarios, the commodity electri-
city is the main driver of GVA, although the share is
reduced by increasing the RE share.
3.3. Regional implications
Figure 4shows the percentage change in the indic-
ator totals for each of the 46 regions over the entire
life cycle when switching from conventional electri-
city to 100% RE. Red indicates a reduction, green
an increase of the indicator. The relative change for
each region is given by pk,r=∑29
y=1ik,RE100,r,y
∑29
y=1ik,RE0,r,ywhere i
is the total amount of the indicator of the key fig-
ure k, RE100 is the 100% RE scenario and RE0 is the
0% RE scenario. Furthermore, rindicates the region
and ythe year. It should be noted that especially
regions that previously achieved relatively low values
and have now experienced a high relative increase
may still show low values in absolute terms. The
diagram only shows the relative changes to the val-
ues in the 0% RE scenario. The key messages of the
chart are the relative change in the indicator values
in individual regions and, in particular, the shift in
the indicator values, but not the absolute level of these
values. The maps on the bottom show the classifica-
tion of the 46 regions, aggregated to Australian states
and territories, water catchments (with the MDB)
and rainfall areas. The map of the states and territ-
ories also shows the locations of the desalination
plants.
Throughout the Eastern Region, water consump-
tion would increase, particularly in MDB areas,
through greater integration of RE. As the desalina-
tion plants in our case study were built to address
water scarcity in this area, this is an unsatisfactory res-
ult. However, we have seen in section 3.1 that water
consumption is only about 2% of total production,
so the additional consumption is relatively small. By
contrast, water consumption is reduced in the direct
coastal areas of the east coast, where most of the eco-
nomic activity and population is located. Although
these regions face higher rainfall than other areas off
the coast, these regions are subject to long-term water
stress due to high population density and economic
activity. Victoria, in particular, would have to cope
8
Environ. Res. Lett. 15 (2020) 114044 M Heihsel et al
Figure 4. Relative changes of the TBL indicators in the analysed 46 regions.
with a significant increase in water consumption at
100% RE.
The densely populated areas on the east coast
would not face any significant change in land use
when shifting to RE. In order to provide water
in the MDB, demand is generated at the desal-
ination plant sites shown in the states and ter-
ritories diagram. Compared to centralised power
plants, which are mostly located near the coast,
the demand for the construction and operation of
decentralised RE plants also generates a decentral-
ised demand. Decentralisation is also reflected in
land use, which is why areas in eastern Australia
and in the MDB in particular are increasingly used
in the 100% RE scenario. In central Australia, on
the other hand, land use would tend to decrease,
9
Environ. Res. Lett. 15 (2020) 114044 M Heihsel et al
while on the west coast, land use would increase
significantly.
On the east coast, where economic activities take
place, we see an obvious reduction in carbon emis-
sions at 100% RE. The entire east coast with its high
population density and economic activity would sig-
nificantly reduce carbon emissions from desalination
by switching to RE. In areas where decentralised RE
increase economic activity, emissions would increase,
e.g. inland in eastern Australia or on the west coast.
If desalination is built and operated on the basis
of RE instead of conventional electricity, fewer jobs
would be created in a few regions in the north
and south, including Tasmania. On the other hand,
employment would increase in the inland areas. All
areas in central and western Australia would see an
increase in employment, as RE is much more decent-
ralised than conventional power stations. Diversific-
ation of employment is beneficial for a country like
Australia, where population and economic activity
are concentrated in a few areas, while there are large
unused areas in the hinterland.
Like employment, GVA would be regionally diver-
sified due to the increasing use of RE. However, GVA
would decrease in the economic areas of the east coast,
especially in the north, but also in Tasmania. In the
Northern Territory, GVA would increase significantly.
From an economic point of view, such a development
would be beneficial as economic activity diversifies
into areas with lower population density and less eco-
nomic activity.
4. Conclusion
In our IO-based hLCA assessment, we showed the
comprehensive TBL sustainability impacts of seawa-
ter desalination, depending on the utilised electricity
source. Using social, economic ecological indicators,
we explained in detail which sectors and regions con-
tribute positively or negatively to the sustainability of
desalination through a 100% RE grid.
With higher RE penetration, we measured rising
employment but also falling GVA. Even if there is a
trade-off between employment and GVA, the assess-
ment shows that a higher RE share would contrib-
ute to regional diversification of economic activity. In
fact, the spatial diversification of GVA is itself a value,
as the local concentration of GVA increases the cost of
land use. Therefore, decentralised GVA prevents price
increases caused by land scarcity. While desalination
with conventional electricity generates higher GVA
through domestic fuel mining, this economic benefit
is not sustainable in the light of the Paris Agreement.
Furthermore, the environmental performance of
desalination with RE is highly beneficial. The applic-
ation of a 100% RE system reduces the carbon emis-
sions of desalination during construction by 71% and
during operation by 96%. The benefit becomes even
more evident when we monetise the reduced external
costs. Let us assume an external cost of AU$100 per
tonne of CO2and 25 Mt as an average amount of
carbon emissions reduced by RE during an operat-
ing year. Then the reduction in external costs through
avoided emissions would result in savings of AU$3.75
bn per year. In contrast, the total loss of GVA over the
same period is some hundred million dollars. How-
ever, our analysis shows that about 30% of the car-
bon emissions during the construction period of the
desalination plants cannot be reduced by 100% RE.
Hence, further measures are still needed.
The results clearly show that RE and desalination
benefit from synergy effects. Due to the high energy
consumption, this applies in particular to desalina-
tion, but similar effects can also be expected for other
water supply technologies due to the technical simil-
arities. Due to the consequences of climate change, it
is to be predicted that Australia’s water problems will
increase in the future. Policymakers should, there-
fore, aim for a common strategy for Australia’s water
and energy supply. The scope of consideration should
be as broad as possible and should also include, for
example, other related problems such as the discharge
of brine with increased desalination use.
Our holistic assessment approach has several
strengths for the analysis of infrastructure projects.
Since we examined all three areas of sustainability
using an hLCA framework, the results are directly
comparable. We used the same system boundaries,
the same assumptions and framework conditions,
and the same economic linkages. Trade-offs, such as
GVA and GHG emissions, can be compared directly.
The methodology is particularly useful for practition-
ers to estimate the impact of infrastructure projects
in the early planning phase. The political value of the
findings is substantial. Our MRIO approach allows
detailed regional and sectoral conclusions. The highly
disaggregated analysis enables long-term economic
policies, e.g. due to changes in regional water use,
employment or economic activities.
The sustainability of seawater desalination plants
depends in particular on regional factors such as local
energy supply or industrial interconnections. In order
to make these influences apparent, regional economic
data are indispensable. The hLCA approach with the
use of the IELab offers an efficient and effective pos-
sibility to break down statistical data to create tailor-
made IOTs. The achievable granularity depends in
particular on local data availability, whose quality is
increasing worldwide. In recent years, IELabs for Aus-
tralia, China, Indonesia, Taiwan, Japan, and the USA
have been created that can be used for these assess-
ments (Geschke and Hadjikakou 2017).
Further research is needed on the effects of large
quantities of brine intake on marine biology. Further-
more, the development of new membranes promises
remarkable increases in desalination efficiency. The
associated effects on sustainability also require further
research.
10
Environ. Res. Lett. 15 (2020) 114044 M Heihsel et al
Acknowledgments
This work was financially supported by the Friedrich
Naumann Foundation for Freedom. We acknowledge
support by the German Research Foundation and
the Open Access Publication Fund of TU Berlin.
The Authors further acknowledge financial support
by the National eResearch Collaboration Tools and
Resources project (NeCTAR) through its Industrial
Ecology Virtual Laboratory. NeCTAR is an Australian
Government project conducted as part of the Super
Science initiative and financed by the Education
Investment Fund. We thank Man Yu for providing
the RE process data. We would also like to thank Ka
Leung Lam, Syed Muhammad Hassan Ali and Bonnie
McBain for their advice on processing the input data.
We would like to acknowledge Arunima Malik for her
advice on the post-balance procedure. Finally, we also
thank the anonymous reviewers for the valuable com-
ments, which have improved this study.
Data availability statement
The data that support the findings of this study are
available upon reasonable request from the authors.
ORCID iDs
Michael Heihsel https://orcid.org/0000-0001-
9886-0006
Manfred Lenzen https://orcid.org/0000-0002-
0828-5288
Frank Behrendt https://orcid.org/0000-0002-
7282-8806
References
ABS 2015a Agricultural Commodities, cat. no. 7121
(www.abs.gov.au/ausstats/[email protected]/mf/7121.0)
ABS 2015b Australian National Accounts: input-output tables,
DetailsPage/5209.0.55.0012014-15?OpenDocument)
ABS 2015c Australian National Accounts: input-output tables
(Product Details), cat. no. 5215. (www.abs.gov.au/
ausstats/[email protected]/mf/5215.0.55.001)
ABS 2015d Australian National Accounts: national Income, cat.
ABS 2015e Australian National Accounts: state Accounts, cat. no.
5220 (www.abs.gov.au/AUSSTATS/[email protected]/
DetailsPage/5220.02017-18?OpenDocument)
ABS 2015f Household Expenditure Survey, cat. no. 6530.0
(www.abs.gov.au/AUSSTATS/[email protected]/Lookup/6530.
0ExplanatoryNotes12015-16?OpenDocument)
ABS 2015g Manufacturing Industry, cat. no. 8221
(www.abs.gov.au/ausstats/[email protected]/mf/8221.0)
ABS 2015h Water Use on Australian Farms, cat. no. 4618
(www.abs.gov.au/ausstats/[email protected]/mf/4618.0)
AGEIS 2015 National Greenhouse Gas Inventory (https://ageis.
climatechange.gov.au)
Aijm V D, Beck H E, Crosbie R S, Ram D J, Liu Y Y, Podger G M,
Timbal B and Viney N R 2013 The Millennium Drought in
southeast Australia (2001–2009): natural and human causes
and implications for water resources, ecosystems, economy,
and society Water Resour. Res. 49 1040–57
Alhaj M and Al-Ghamdi S G 2019 Integrating concentrated solar
power with seawater desalination technologies: a
multi-regional environmental assessment Environ. Res. Lett.
14 74014
AusLCI. 2020 The Australian Life Cycle Inventory Database
Initiative. (www.auslci.com.au)
Baten R and Stummeyer K 2013 How sustainable can desalination
be? Desalin. Water Treat. 51 44–52
Bell E M, Stokes-Draut J R and Horvath A 2018 Environmental
evaluation of high-value agricultural produce with diverse
water sources: case study from Southern California Environ.
Res. Lett. 13 25007
Borchers Arriagada N, Bowman D M J S, Palmer A J and
Johnston F H 2020 Climate change, wildfires, heatwaves and
health impacts in Australia Extreme Weather Events and
Human Health ed R Akhtar (Cham: Springer International
Publishing) pp 99–116
Bullard C W and Sebald A V 1977 Effects of parametric
uncertainty and technological change on input-output
models Rev. Econ. Stat. 59 75–81
Bullard C W and Sebald A V 1988 Monte Carlo sensitivity analysis
of input-output models Rev. Econ. Stat. 70 708–12
Cherif H, Champenois G and Belhadj J 2016 Environmental life
cycle analysis of a water pumping and desalination process
powered by intermittent renewable energy sources Renew.
Sustain. Energy Rev. 59 1504–13
Clark G F, Knott N A, Miller B M, Kelaher B P, Coleman M A,
Ushiama S and Johnston E L 2018 First large-scale ecological
impact study of desalination outfall reveals trade-offs in
effects of hypersalinity and hydrodynamics Water Res.
145 757–68
Crisp G J 2012 Desalination and water reuse-sustainably drought
proofing Australia Desalin. Water Treat. 42 323–32
Ecoinvent 2014 Ecoinvent Database 2.2 (Zurich, Switzerland)
Elkington J 1998 Cannibals with Forks: The Triple Bottom Line of
21st Century Business (Gabriola Island, BC: New Society
Publishers)
Finnveden G, Hauschild M Z, Ekvall T, Guin´
ee J, Heijungs R,
Hellweg S, Koehler A, Pennington D and Suh S 2009 Recent
developments in life cycle assessment J. Environ. Manage.
91 1–21
Foran B, Lenzen M, Dey C and Bilek M 2005 Integrating
sustainable chain management with triple bottom line
accounting Ecol. Econ. 52 143–57
Geschke A and Hadjikakou M 2017 Virtual laboratories and
MRIO analysis—an introduction Econ. Syst. Res. 29 143–57
Gleick P H 1994 Water and Energy Annu. Rev. Energy Environ.
19 267–99
Global Water Intelligence 2016 Desaldata Database
(www.desaldata.com)
Greve P et al 2018 Global assessment of water challenges under
uncertainty in water scarcity projections Nat. Sustain.
1486–94
Gude V G 2016 Desalination and sustainability - an appraisal and
current perspective Water Res. 89 87–106
Gude V G and Fthenakis V 2020 Energy efficiency and renewable
energy utilization in desalination systems Prog. Energy 2
Haddad B M 2013 A case for an ecological-economic research
program for desalination Desalination 324 72–78
Hadjikakou M, Stanford B D, Wiedmann T, Rowley H V,
Kobayashi Y, Ishii S, Gaitan J P A, Johns G, Lundie S and
Khan S J 2019 A flexible framework for assessing the
sustainability of alternative water supply options Sci. Total
Environ. 671 1257–68 (https://linkinghub.elsevier.com/
retrieve/pii/S004896971931280X)
Heihsel M, Ali S M H, Kirchherr J and Lenzen M 2019a
Renewable-powered desalination as an optimisation
pathway for renewable energy systems: the case of Australia’s
Murray-Darling Basin Environ. Res. Lett. 14 124054
Heihsel M, Lenzen M, Malik A and Geschke A 2019b The carbon
footprint of desalination: an input-output analysis of
seawater reverse osmosis desalination in Australia for
2005–2015 Desalination 454 71–81
11
Environ. Res. Lett. 15 (2020) 114044 M Heihsel et al
Heijungs R and Lenzen M 2014 Error propagation methods for
LCA—a comparison Int. J. Life Cycle Assess. 19 1445–61
Jijakli K, Arafat H, Kennedy S, Mande P and Theeyattuparampil V
V 2012 How green solar desalination really is?
Environmental assessment using life-cycle analysis (LCA)
approach Desalination 287 123–31
Joshi S 1999 Product environmental life-cycle assessment using
input-output techniques J. Ind. Ecol. 395–120
Komesaroff P and Kerridge I 2020 A continent aflame: ethical
lessons from the Australian Bushfire Disaster J. Bioeth. Inq.
17 11-14
Lahr M L and de Mesnard L 2004 Biproportional techniques in
input-output analysis: table updating and structural analysis
Econ. Syst. Res. 16 115–34
Lattemann S and Höpner T 2008 Environmental impact and
impact assessment of seawater desalination Desalination
220 1–15
Lenzen M 2000 Errors in conventional and input-output-based
life-cycle inventories J. Ind. Ecol. 4127–48
Lenzen M et al 2014 Compiling and using input–output
frameworks through collaborative virtual laboratories Sci.
Total Environ. 485–486 241–51
Lenzen M et al 2020 Global socio-economic losses and
environmental gains from the Coronavirus pandemic PloS
One 15 e0235654
Lenzen M, Sun Y Y, Faturay F, Ting Y P, Geschke A and Malik A
2018 The carbon footprint of global tourism Nat. Clim.
Change 8522–8
Lenzen M, Wood R and Wiedmann T 2010 Uncertainty analysis
for multi-region input-output models—a case study of the
UK’S carbon footprint Econ. Syst. Res. 22 43–63
Leontief W W 1966 Input-output Economics (Oxford : Oxford
University Press)
Liu C H, Lenzen M and Murray J 2012 A disaggregated emissions
inventory for Taiwan with uses in hybrid input-output life
cycle analysis (IO-LCA) Nat. Resour. Forum 36 123–41
Liu J, Chen S, Wang H and Chen X 2015 Calculation of carbon
footprints for water diversion and desalination projects
Energy Procedia 75 2483–94
Malik A, Lenzen M and Geschke A 2016 Triple bottom line study
of a lignocellulosic biofuel industry GCB Bioenergy 896–110
Malik A, Lenzen M, Mcalister S and Mcgain F 2018 The carbon
footprint of Australian health care Lancet Planet. Health
2e27–35
Malik A, Lenzen M, Rˆ
o N E and Dietzenbacher E 2014 Simulating
the impact of new industries on the economy: the case of
biorefining in Australia Ecol. Econ. 107 84–93
Malik A, Mcbain D, Wiedmann T O, Lenzen M and Murray J 2019
Advancements in input-output models and indicators for
consumption-based accounting J. Ind. Ecol. 23 300–12
Norwood Z and Kammen D 2012 Life cycle analysis of distributed
concentrating solar combined heat and power: economics,
global warming potential and water Environ. Res. Lett.
744016
Onat N C, Kucukvar M and Tatari O 2014 Integrating triple
bottom line input-output analysis into life cycle
sustainability assessment framework: the case for US
buildings Int. J. Life Cycle Assess. 19 1488–505
Pomponi F and Lenzen M 2018 Hybrid life cycle assessment
(LCA) will likely yield more accurate results than
process-based LCA J. Clean. Prod. 176 210–5
Potter N J, Chiew F H S and Frost A J 2010 An assessment of the
severity of recent reductions in rainfall and runoff in the
Murray-Darling Basin J. Hydrol. 381 52–64
Quandt R E 1958 Probabilistic errors in the leontief system Nav.
Res. Logist. Q. 5155–70
Rodríguez-Alloza A M, Heihsel M, Fry J, Gallego J, Geschke A,
Wood R and Lenzen M 2019 Consequences of long-term
infrastructure decisions—the case of self-healing roads and
their CO2emissions Environ. Res. Lett. 14 114040
Sadhwani J J, Veza J M and Santana C 2005 Case studies on
environmental impact of seawater desalination Desalination
185 1–8
Shahabi M P, Mchugh A, Anda M and Ho G 2014 Environmental
life cycle assessment of seawater reverse osmosis desalination
plant powered by renewable energy Renew. Energy
67 53–58
Shehabi A, Stokes J R and Horvath A 2012 Energy and air
emission implications of a decentralized wastewater system
Environ. Res. Lett. 724007
Stokes J and Horvath A 2006 Life cycle energy assessment of
alternative water supply systems Int. J. Life Cycle Assess.
11 335–43
Suh S and Huppes G 2005 Methods for life cycle inventory of a
product J. Clean. Prod. 13 687–97
Wedderburn S D, Hammer M P and Bice C M 2012 Shifts in
small-bodied fish assemblages resulting from
drought-induced water level recession in terminating lakes
of the Murray-Darling Basin, Australia Hydrobiologia
691 35–46
Yu M and Wiedmann T 2018 Implementing hybrid LCA routines
in an input–output virtual laboratory J. Econ. Struct. 733
Zarzo D and Prats D 2018 Desalination and energy consumption.
What can we expect in the near future? Desalination 427 1–9
Ziolkowska J R 2015 Is desalination affordable? Regional cost and
price analysis Water Resour. Manage. 29 1385–97
12
