You are on page 1of 8

Journal of Cleaner Production 86 (2015) 201e208

Contents lists available at ScienceDirect

Journal of Cleaner Production


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

Water footprinting of electricity generated by combined cycle gas


turbines using different cooling technologies: a practitioner's
experience
Jan Mertens a, *, Anne Prieur-Vernat b, Dominique Corbisier a, Elsa Favrot c, Gustaaf Boon d
a
Laborelec, GDF-SUEZ, Rodestraat 125, 1630 Linkebeek, Belgium
b
CRIGEN, GDF-SUEZ, 361 Avenue du Pr esident Wilson, B.P.33, 93211 Saint Denis La Plaine, France
c
Environmental and Societal Responsibility Division, GDF-SUEZ, Troonplein, 1000 Brussels, Belgium
d
Research & Innovation Division, GDF SUEZ, Troonplein 1, 1000 Brussels, Belgium

a r t i c l e i n f o a b s t r a c t

Article history: Efforts undertaken for reducing environmental impacts of energy production have been primarily
Received 5 March 2014 focused on carbon reduction while the fact that energy production also requires water has been largely
Received in revised form overlooked. During the last decade, and despite the fact that global warming still remains today the focus
30 July 2014
of many environmental evaluations, water scarcity issues have increasingly received attention. Despite
Accepted 16 August 2014
Available online 27 August 2014
the fact that recently, an increasing demand for large industrial companies to calculate and report on
their water footprint exists, water resources have only recently been addressed in life cycle assessment
(LCA) and their assessment still lacks wide application. The paper presents a practitioner's experience
Keywords:
Water footprinting
with respect to the application of three recently developed water footprinting methodologies that are
Life cycle assessment considered as current-state-of-the-art. The methods are applied with as objective the estimation of the
Energy water footprint of combined cycle gas turbines with different cooling technologies. The study reveals that
absolute values of water footprints (Leq. kWh1) are very different among methods and therefore results
are not directly comparable in terms of their absolute results. In contrast, ranking among power plants
agree for the different methods when large differences in water consumption/impact exist. However, the
ranking may differ between different methods when differences are small. It therefore remains
impossible to select one method as the preferred method to use. This study therefore serves as support to
the recently emerging working groups (eg. WULCA) that aim at harmonizing the different existing
methodologies.
2014 Elsevier Ltd. All rights reserved.

1. Introduction fact that energy production also requires water has been largely
overlooked (Tzimas, 2011). Based on 2007 data, water abstraction
Over the last decade, and despite the fact that global warming for cooling in energy production accounts for 44.8% of the total
still remains today the focus of many environmental evaluations, water abstraction in Europe (Tzimas, 2011). Hydro-electric plants
water scarcity issues have increasingly received attention. The and cooling water use, both relevant for power production, have
combination of population growth and economic development been identied as globally important water consumers (Mila  i
leads to increasing human freshwater use (Kounina et al., 2013). Canals et al., 2009). Currently, efforts are on-going within Europe
While research has primarily focused on agricultural water use, in the eld of water abstraction mainly concerning the develop-
industrial water use has recently become more prominent (Pster ment of advanced cooling technologies. This lead to a reduction in
et al., 2011). In particular for electricity production systems, the energy-related cooling water abstractions of 10% over the last
focus has been primarily on carbon emission reduction while the 10e15 years in Europe (Tzimas, 2011). Although signicant
amounts of water are abstracted, only a small fraction is consumed.
In this study, evaporated water through a cooling tower is consid-
Abbreviations: LCA, Life Cycle Assessment; CCGT, Combined Cycle Gas Turbine. ered as consumed since it is unlikely that it is returned in the form
* Corresponding author. Sustainable Process Technology, Laborelec, Rodestraat of precipitation into the same water body (catchment). However,
125, 1630 Linkebeek, Belgium. Tel.: 32 2 382 05 71; fax: 32 2 382 02 41. this is very much function of the catchment size and prevailing
E-mail address: mertensja@yahoo.co.nz (J. Mertens).

http://dx.doi.org/10.1016/j.jclepro.2014.08.046
0959-6526/ 2014 Elsevier Ltd. All rights reserved.
202 J. Mertens et al. / Journal of Cleaner Production 86 (2015) 201e208

wind speeds and thus may not always be entirely correct. Even (ii) share experiences and formulate recommendations for wa-
when considering power plants with a cooling tower, still most of ter footprinting method developers.
the water is returned into the river (and thus only a small fraction is
evaporated) albeit at a higher temperature than its intake tem-
perature (Verones et al., 2010). For electricity production in Europe,
1.1. Material and methods
Tzimas, 2011 estimate a 75% reduction in greenhouse gases, 90%
reduction in freshwater abstraction but a 30% increase in fresh-
1.1.1. Power plant description
water consumption compared to 1990. The reduction in water
The four different GDF SUEZ power plants (CCGT plants) modelled
abstraction is mainly due to the increasing share of power plants
in this study are listed in Table 1. Two of them are located in Belgium,
with a cooling tower (instead of open cycle power plants) and
one in France and one in Italy. The details on their capacity and
because of the increasing efciencies of modern power plants. The
conversion efciency can also be found in Table 1. These four were
increasing water consumption is also due to the increasing share of
chosen because they all have a different cooling technology and they
power plants with a cooling tower as well as because of the in-
will be referred to as follows throughout this study:
crease in overall electricity production. Freshwater consumption
should therefore be the main criterion for setting research &
1. Aero CCGT: This CCGT uses an aerocondensor as cooling tech-
development and demonstration priorities.
nology as depicted in Fig. 1.
Water resources have only recently been addressed in life cycle
2. Surface CT: This CCGT disposes of a surface water (ie. river
assessment (LCA) and their assessment still lacks wide application
water) cooling tower with a concentration factor of 1.95. The
(Pster et al., 2011). LCA databases, such as the ecoinvent v2.1 data-
concentration factor is the relation between the amount of
base (Ecoinvent Centre, 2007), do provide basic data on water use.
water taken in and the amount of water released back into the
Some of the available life cycle impact assessment methods also
surrounding water body:
differentiate different types of water use (Pster et al. 2009; Mil ai
Canals et al., 2009; Bayart et al. 2010). Despite these efforts, Kounina
et al., 2013 conclude that the existing databases should be (1) water in evaporation water out
completed with input and output freshwater ow differentiated ac- concentration factor
water out water out
cording to water types based on its origin (surface water, ground-
(1)
water, and precipitation water stored as soil moisture), (2)
regionalized, and (3) if possible, characterized with a set of quality A concentration factor of 2 for example means that the amount of
parameters. To tackle this major environmental concern, various the water evaporated in the cooling tower is half of the water intake. A
initiatives were recently launched in order to develop and standardize concentration factor of 1 corresponds to an open cycle or once through
analytical tools to measure and assess freshwater use at regional and cooling system where no water is evaporated from the cooling tower
global scale (Kounina et al., 2013). In particular, there is the United and all water is released into the surrounding water body.
Nations Environment Programme (UNEP)/Society of Environmental
Toxicology and Chemistry (SETAC) Life Cycle Initiative's working 3. Sea CT: this CCGT disposes of sea water cooling tower with a
group on the assessment of freshwater use and consumption in life concentration factor of 1.19. This lower concentration factor as
cycle assessment (LCA) called WULCA (Koehler and Aoustin, 2008). compared to the surface CT implies that less water is evaporated
In many cases, the water footprint concept is referred to when through the cooling tower as compared to the surface CT case.
only quantities of consumed water are being assessed rather than 4. Open cycle: this power plant uses sea water in an open cycle (ie.
its related impact. The sense of localization, and even more so of concentration factor 1) or thus all water that is taken into the
water type and quality, is missing in most studies: ie. the con- plant is rejected back into the sea and thus no water is
sumption of 1 L of water in a water scare area is considered to have consumed as such.
the same impact as 1 L of water in a water abundant area. Main
reason is the fact that only few methodologies exist that allow For each of the power plants, the technology for the production
estimating the impact of water use ie. that relate quantities to local of the demineralized water (Ion exchange or Reverse osmosis) is
environmental impact. Moreover, the existing methods were only also listed in Table 1 as well as the water consumed in this process.
recently developed. Many of the studies that estimate water foot- Demineralised water is used in the wateresteam cycle to drive the
prints are related to water consumption quantities of agricultural electricity generation turbine. In some power plants, municipal
products (eg. Herath et al., 2013; Page et al., 2012; Jefferies et al., tapwater is used for the production of demineralised water while in
2012) and biofuels (eg. Hagman et al., 2013;Su et al., 2014). This is others, surface water or a combination of surface- and groundwater
also the case in the study from Herath et al., 2011 where quantities (local sources) is used. The right column in Table 1 presents the type
of consumed water are estimated according to three methods for and amount of water taken in and released for the production of 1 L
hydro-electric power in New Zealand. In fact, no water footprinting of demineralised water. Furthermore, from these numbers, the
study that estimates impacts, taking into account the localization amount of water needed for the production of 1 kWh of electricity
and water type, was found in literature for electricity production is calculated and presented. The differences between the amount of
systems. This study is therefore novel in this sense. demineralised water consumed by the plants is mainly related to
The driving force behind the study is the fact that recently, an differences in number of start-ups and shutdowns the plants have
increasing demand for large industrial companies to calculate and per year. Increasing this number generally increases the deminer-
report on their water footprint exists. Therefore, the paper shares a alised water consumption. However, Table 1 shows that the dem-
practitioner's experience with respect to the application of three ineralised water consumption is a small fraction of the cooling
waterfoot printing methodologies, which are current-state-of-the- water consumption for plants that have a cooling tower.
art. The aim is twofold:
1.2. Power plant modelling using SIMAPRO and QUANTIS SUITE
(i) estimation of the water footprint of electricity generated by a
CCGT using different cooling technologies for future use in the The functional unit used throughout the study is dened as one
GDF SUEZ action plans related to reducing its water footprint kWh of electricity from the power plant delivered to the grid. A
J. Mertens et al. / Journal of Cleaner Production 86 (2015) 201e208 203

Table 1
Details on the modelled GDF SUEZ CCGT plants.

Name Capacity Type & Country Watershed Cooling water (per kWhe) Demineralised water
(Mwe) efciency

Aero 350 CCGT, 51% Belgium Schelde Aerocondensor 3.25 L groundwater in (IX), 2.25 L out to
surface water for the production of 1 L
demin water, > 0.013 L kWh1
consumed
Surface CT 437 CCGT, 55% Belgium Maas CT, 2.9 L surface water in, 1.5 L out to surface water, conc. 1 L surface & 0.07 L tapwater in (RO) for
factor 1.95,> 1.4 L consumed the production of 1 L demin water,
0.07 L out to surface water, >
0.031 L kWh1 consumed
Sea CT 390 CCGT, 54% Italy Unsp. Italy CT, 7.1 L sea water in, 6 L out to sea water, conc. 2.1 L tapwater in (RO), 1.1 L out to sea
factor 1.19, > 1.1 L consumed water for the production of 1 L demin
water, > 0.014 L kWh1 consumed
OC 528 CCGT, 54% France Unsp. France OC, 175 L sea water in, 175 L out to sea water, conc. 2.8 L tapwater in (RO), 1.8 L out to sea
factor 1, > 0 L consumed water for the production of 1 L demin
water, > 0.0066 L kWh1 consumed

CT: Cooling Tower.


OC: Once Through Cooling (Open Cycle).
IX: Ion Exchange.
RO: Reverse Osomis.

SIMAPRO model (www.pre-sustainability.com/SimaPro, accessed The natural gas combusted in the BE CCGT plants is modelled
19/06/2014) was constructed, using real life data from a 350 MWe according to the Belgian supply mix in 2011: 18% Liqueed Natural
power plant located in Belgium. This model is used as a basis for all Gas from Qatar and the rest is pipeline gas from Norway, The
CCGT plants studied here. SIMAPRO software (System for Inte- Netherlands, Russia and United Kingdom. Since no specic in-
grated environMental Assessment of PROducts) was developed by ventory data of the natural gas extraction and liquefaction in Qatar
the Dutch PRe  Consultants rm and is a well-known, internation- are available, corresponding datasets of Algerian liqueed natural
ally accepted and veried LCA modelling tool. The software allows gas production are applied (Ecoinvent Centre, 2007). Since Qatar is
to model and analyse complex life cycles in a systematic and not connected to Europe with a long-distance pipeline, the natural
transparent way, following the recommendations of the ISO 14040 gas is transported as liqueed natural gas in an LPG tanker. The
series of standards. transport distance from the Ras Laffan port in Qatar to Zeebrugge in
This model is used to model each of the power plants. However, Belgium amounts to 11,277 km. In Belgium the liqueed natural gas
the site specic efciency, water use (demineralisation cooling) is evaporated in an evaporation plant modelled according to the
and fuel supply chain are individually implemented. The primary operation of an Italian plant (Ecoinvent Centre, 2007). The high
sources of inventory data in this study are: pressure natural gas at consumer is modelled according to corre-
sponding ecoinvent dataset including energy requirements, emis-
 Real life inventory data of each of the CCGT power plants based sions and infrastructure of the high pressure distribution network
on the environmental reporting data of GDF SUEZ in Belgium. For the natural gas supply of the other CCGT (in France
 The ecoinvent database v2.01 (Ecoinvent Centre, 2007), that and Italy), the eco-invent countries natural gas supply is used. A
contains inventory data for many basic materials and services plant life span of 30 years is adopted.
and is used for the modelling of the infrastructure and fuel The model was designed in such a way that it is possible to
supply. separately evaluate the water footprint of: (i) fuel supply chain, (ii)

Fig. 1. Sketch of a Combined Cycle Gas Turbine with an aerocondensor as cooling technology.
204 J. Mertens et al. / Journal of Cleaner Production 86 (2015) 201e208

Table 2
Division between the different power plant processes into direct and indirect con-
tributions to the power plant's water consumption.

Direct Indirect
contribution contribution

Cooling at the power plant x


Demineralized water production at PP x
PP operation, incl. chemicals use x
and waste treatment
Fuel supply chain x
Infrastructure x

operation of the Power Plant (PP), (iii) cooling water, (iv) demin-
eralisation installation and (v) infrastructure building (ie. the con-
struction of the power plant). The model for the CCGT PP operation
includes the consumption of chemicals periodically or continuously
used such as NH3, HCl, NaOH and carbohydrazine for the water
steam cycle, H2 for the generator cooling etc. Table 2 shows the split
that was selected for processes that are considered to be direct or
indirect (ie. upstream). Power plant operation was classied as
being indirect since it mainly consists of the consumption of
chemicals that are manufactured outside of the power plant.
The model was developed using SIMAPRO software (version 7).
It was then imported into QUANTIS SUITE 2.0 software because this
modelling tool has the Quantis water database (Quantis, 2011) Fig. 2. Relation between water footprint and Global Warming Potential (GWP) for the
incorporated and allows the denition of user specied water modeled CCGT power plants.
footprint characterisation factors as explained below. The Quantis
To convert litres to litres equivalent, characterization factors are
water database will be incorporated in SIMAPRO software version 8
provided by each of the three methods at different scales. In this
so that exporting from SIMAPRO and importing in the QUANTIS
study, the characterization factors of the different methods at the
SUITE 2.0 will be no longer needed.
smallest available scale (0.5 longitudinal by 0.5 latitudinal
extension or watershed values) are used and implemented in
1.3. Water footprinting methods Quantis software. These values are obtained for each of the three
methods through a Google earth map made available in each of the
Three current state-of-the-art water footprinting methods are three methods. All methods are based on the Water GAP Global
applied and the results compared. An in-depth description of each Hydrology Model (WGHM) (Do ll et al., 2012). This implies that
of the methods is not given here, however their main properties are when comparing the water footprint of different process; we would
listed in Table 3. This table shows that differences exist among the expect to obtain the same ranking using the different methodolo-
methodologies. All methodologies provide characterization factors gies but the absolute numbers (in terms of Leq.) will vary as a result
that allow converting litres (consumed water) to Leq. (impact of the different scaling used.
related to this water use as a function of the (i) type of water and (ii) It is important to mention that for the on-site water consumption
where the water was consumed). This is analogous to what is done of the power plants, site-specic (i.e. watershed level) characteriza-
when calculating Global Warming Potentials (GWP) of different tion factors are implemented in Quantis software from the available
gases. For example, emitting 1 g of methane is attributed a value of Google earth maps for all three methods. However, in Quantis soft-
23 grams equivalent CO2 since emitting 1 g of CO2 is arbitrarily set ware, for the upstream processes, only country specic data are
to a GWP value of 1 g equivalent. Similarly, high Leq. values refer to implemented for each of the methods. Moreover, it is not straight-
the fact that the consumption of 1 L of water has a high environ- forward to nd out which characterization factors are used in each of
mental impact. There is however a large difference between the the processes in Quantis software; which makes upstream water
methods in the different scaling that they apply to go from litres to footprint interpretation difcult as explained in more detail below.
litres equivalent implying that absolute numbers will not be
comparable. For the Pster et al., 2009 and the Boulay et al., 2011b 2. Results and discussion
methods, a scaling (referred to as S-curve in Table 3) between 0 and
1 is used. This means that in regions with great water scarcity, the 2.1. Water consumption and Global Warming Potential
consumption of 1 L of water equals the maximum value of 1 Leq. In
contrast, in regions where water is abundant, Leq. values corre- Fig. 2 presents the relation between GWP and water footprint
sponding to the consumption of 1 L of water will be smaller than 1. for the modeled CCGT power plants. The GWP for each of the power

Table 3
Main assumptions of the three methodologies used in this study (adapted from Boulay, 2013).

Pster et al., 2009 Boulay et al., 2011b (simplied) Frischknecht et al., 2006

Indicator Withdrawal to availability ratio Consumption to availability ratio Withdrawal to availability ratio
Spatial resolution and coverage 0.5  0.5 eCountryeWorld Watershedeworld WatershedeCountryeworld
Modelling S-curve from 0.01 to 1 S-curve from 0 to 1 Normalized ratio based on legislation
Inventory ow Withdrawal or consumption Consumption Consumption
Background data Water GAP model Water GAP model Water GAP model
Other specicities Seasonal variability considered Split groundwater/surface water
J. Mertens et al. / Journal of Cleaner Production 86 (2015) 201e208 205

Fig. 3. Relation between water consumption and water footprint calculated using three different methods for the modeled CCGT plants (gure on the right is a zoom into the lower
part of the gure on the left).

plants is calculated by Quantis software using the IPPC (2007) generated but has a water footprint close to the Open Cycle (OC)
method implemented in software. The gure reveals a factor 5-10 and aerocondensor CCGT plants (between 0.02 and
difference between water consumption in a CCGT with an aero- 0.05 Leq. per kWh,. depending on the method used). The reason is
condensor or an open cycle as compared to a CCGT with a cooling that the use of sea water is not penalized (ie. characterization factor
tower. It is clear that the direct water consumption at the power of 0) and thus free of footprint in all methods. This is a short-
plant dominates the water footprint of CCGT, but still variations coming of the methods since the possible water quality change and
exist and these will be discussed further in this text. The variation rise in water temperature as the sea water moves through the plant
in GWP is rather small and is a function of the GWP related to the is not accounted for. The CT with surface water CCGT shows the
upstream gas supply chain and the power plant's efciency. The gas largest impact of all with a water footprint between 0.2 and 3
supply chain eco-invent model for gas supply in Italy and France (Leq. per kWh), again depending on which method is used.
has a large GWP associated to it. This is the reason why, despite the
lower efciency of the CCGT with aerocondensor (51% as compared
to 54e55% for the other CCGT) located in BE, it still has a lower GWP 2.3. Comparison of water footprinting methodologies
than the French and Italian CCGT. This result shows that care must
be taken interpreting results in cases where upstream processes are The comparison between the water footprints calculated using
different. It also suggests that more research into upstream pro- the three different methods is presented in Fig. 3. This comparison
cesses must be carried out to further rene our modelling. conrms that one should not compare the absolute values obtained
In many cases, there is a trade-off between water consumption using the different methods. The main reason is that all methods
and GWP. This is the case when choosing an aerocondesor or use a different scaling, ie. some use characterization factors that lie
cooling tower as power plant cooling system. An aerocondensor between 0 and 1 (eg. Pster et al., 2009 and Boulay et al., 2011),
needs electricity to operate and thus will lower the conversion while others use characterization factors that can be larger than 1
efciency of the power plant which increases its GWP. However, it (eg. Frischknecht, 2006). All methods agree that the surface CT
will have a huge positive effect on the power plant's water con- CCGT has the largest water footprint. However, Fig. 3 does show a
sumption. This trade-off between GWP and water consumption is difference in ranking among the power plants when differences in
also discussed in the study by Page et al., 2012 for the case of to- water footprint are smaller: eg. Boulay et al., 2011 rank the aero-
mato production. These authors propose an endpoint indicator to condensor CCGT as having the smallest impact of all CCGT plants
evaluate this trade-off and show that in their case the GWP dom- while Pster et al., 2009 considers the aerocondensor CCGT to have
inates the environmental impact. the highest impact of all CCGT (not considering the surface CT).
Fig. 4 relates the water consumption to water footprint calcu-
lated using the three different methods for each of the modeled
2.2. Water footprint CCGT plants in a separate gure. Moreover, both the consumption
and the impact are split-up in a direct and indirect contribution
Fig. 3 presents the relation between water consumption (in li- (see denition above) for each of the plants. As was already seen in
tres per kWh) and water footprint (in litres equivalent per kWh) Fig. 3, all methods agree that the surface CT has the highest impact
calculated according to three different methods. A high water but the methods do not agree on the ranking of the other CCGT
consumption in terms of litre per kWh of electricity does not plants. Moreover, the division between direct and indirect impacts
necessarily imply a high water footprint (expressed as litre equiv- varies widely among the methods as well. The origin of the char-
alent per kWh). One example of this is the sea Cooling Tower (CT) acterization factors used for the indirect activities is not very
power plant which consumes around 1.4 L per kWh of electricity transparent and it turns out to be impossible to actually verify
206 J. Mertens et al. / Journal of Cleaner Production 86 (2015) 201e208

Fig. 4. Comparison of the water consumption (direct and indirect) and corresponding water footprint (direct and indirect) calculated using three methodologies for each of the
modeled CCGT.

which factors are used for the different upstream processes. in ranking (not in absolute values) is therefore not to be expected a-
Moreover, in Quantis software used for the modelling, the charac- priori and we hypothesize that they are possibly due to:
terization factors are only available at the country level.
These discrepancies between methods are observed despite the 1. Withdrawal to/or Consumption to/availability ratio as underly-
fact that we implemented in the modelling all the available char- ing environmental impact
acterization factors all at catchment level. On top of that, the same 2. Seasonal variability taken into account or not
underlying model (water GAP) is used for the calculation of the 3. Distinction between the use tapwater, surface water and
characterization factors in all methods. The observed discrepancies groundwater (only Boulay et al., 2011)
J. Mertens et al. / Journal of Cleaner Production 86 (2015) 201e208 207

Fig. 5. Effect of the location of a CCGT plant on the water footprint calculated according to three different methodologies (note the log scale).

2.4. Impact of location on water footprint of a CCGT plant is not transparent and they are only available at country level. This
latter comment is valid for all methods. The main conclusion
Fig. 5 illustrates the effect of moving the surface CT from the therefore from this study is that working groups (eg. WULCA
Meuse catchment (BE) to different catchments on its water (Koehler and Aoustin, 2008)) that aim at harmonizing the different
footprint. Also the effect of taking country level characterization existing methodologies are important.
factors instead of catchment characterization factors is presented. This conclusion is very much in line with the conclusions from
It is clear that the location has a large impact on its water foot- the studies carried out by Jeswani and Azapagic, 2011 and
print. Also the effect of whether one considers the impact factor Zonderland-Thomassen et al., 2014. These authors applied some of
at country level or at catchment level inuences greatly the re- the existing methodologies to calculate respectively the water
sults. Surprisingly, when going from country to catchment level, footprint of a corn-derived ethanol produced in 12 different
there is no consistent effect on the CCGT's water footprint when countries and beef cattle and sheep produced in New-Zealand.
comparing different methodologies: eg. for France and Italy: Jeswani and Azapagic, 2011 report a huge variation in results be-
Pster et al., 2009 impact increases when going from country to tween different methods whereas Zonderland-Thomassen et al.,
catchment level while the other two methods show a decrease. 2014 stress the need to integrate catchment variability into water
This inconsistency may result in a different ranking of PP's footprinting. Both studies conrm the need for a standardized
depending on whether country or catchment characterization methodology for assessing the impacts of water use on a life cycle
factors are considered and the method used. Again, this analysis basis. This should be done in the short-term since the pressure on
calls for harmonization between the different methods to make international industrial companies to estimate their water footprint
available standard characterization factors at the different scales is increasing rapidly.
(ie. country, catchment, ). Apart from this, the current study and studied methodologies
miss out on two important effects that power plants related to their
water footprint: (i) changes in water quality and (ii) increase of
3. Conclusions temperature as water moves through the cooling circuit of the
power plant are not taken into account. There is also a clear need
The study conrms the initial thinking that absolute values of for method development to account for these two effects.
water footprints (Leq. kWh1) are very different among methods
and therefore results are not directly comparable in terms of their
absolute values. However, this difference in absolute values is not Acknowledgements
problematic. The reason is that when comparing water footprints of
power plants, we are usually interested in rankings and not in This research was feasible thanks to the GDF SUEZ research
absolute values. In general, the ranking among power plants agree program which funded this research.
for the different methods when large differences in water con-
sumption/impact exist. However, rather disturbing from a user's
point of view is that fact that the ranking may differ between References
different methods when differences are small. It therefore remains
^nes, L., Margni, M., Pster, S., Franois, V., Koehler, A.,
Bayart, J.-B., Bulle, C., Desche
impossible to select one method as the preferred method to use. 2010. A framework for assessing off-stream freshwater use in LCA. Int. J. LCA 15,
Apart from this, the origin of the factors used for indirect activities 439e453.
208 J. Mertens et al. / Journal of Cleaner Production 86 (2015) 201e208

Boulay, A.-M., 2013. In: Diving into Water Scarcity Indicators: Model Components, Kounina, A., Margni, M., Bayart, J.-B., Boulay, A.-M., Berger, M., Bulle, C.,
Choices and Uncertainties, Presentation at the SETAC Europe Annual Meeting, Frischknecht, R., Koehler, A., Mil a i Canals, L., Motoshita, M., Nn ~ ez, M.,
Glasgow, Scotland. Peters, G., Pster, S., Ridoutt, B., van Zelm, R., Verones, F., Humbert, S., March
Boulay, A.-M., Bouchard, C., Bulle, C., Desche ^nes, L., Margni, M., 2011a. Categorizing 2013. Review of methods addressing freshwater use in life cycle inventory and
water for LCA inventory. Int. J. LCA 16, 639e651. impact assessment. Int. J. LCA 18 (3), 707e721.
Boulay, A.-M., Bulle, C., Bayart, J.-B., Desche ^nes, L., Margni, M., 2011b. Regional Mila i Canals, L., Chenoweth, J., Chapagain, A., Orr, S., Anto  n, A., Clift, R., 2009.
characterization of freshwater use in LCA: modeling direct impacts on human Assessing freshwater use impacts in LCA: part I e inventory modelling and
health. Environ. Sci. Technol. 45, 8948e8957. characterisation factors for the main impact pathways. Int. J. LCA 14, 28e42.
Do ll, P., Hoffmann-Dobrev, H., Portmann, F.T., Siebert, S., Eicker, A., Rodell, M., Jefferies, D., Mun~ oz, I., Hodges, J., King, V.J., Aldaya, M., Ercin, A.E., Mila
 i Canals, L.,
Strassberg, G., Scanlon, B., 2012. Impact of water withdrawals from ground- Hoekstra, A.Y., 2012. Water Footprint and Life Cycle Assessment as approaches
water and surface water on continental water storage variations. J. Geodyn. 59- to assess potential impacts of products on water consumption. Key learning
60, 143e156. points from pilot studies on tea and margarine. J. Clean. Prod. 33, 155e166.
Ecoinvent Centre, 2007. Ecoinvent Data v2.01. ecoinvent reports No. 1-25. Swiss Page, G., Ridoutt, B., Bellotti, B., 2012. Carbon and water footprint tradeoffs in fresh
Centre for Life Cycle Inventories, Duebendorf, Switzerland. retrieved from: tomato production. J. Clean. Prod. 32, 219e226.
www.ecoinvent.org (accessed 19.06.14.). Pster, S., Koehler, A., Hellweg, S., 2009. Assessing the environmental impacts of
Frischknecht, R., Steiner, R., Braunschweig, A., Egli, N., Hildesheimer, G., 2006. Swiss freshwater consumption in LCA. Environ. Sci. Technol. 43, 4098e4104.
Ecological Scarcity Method: the New Version 2006. Swiss Federal Ofce for the Pster, S., Saner, D., Koehler, A., 2011. The environmental relevance of freshwater
Environment (FOEN), Switzerland. consumption in global power production. Int. J. LCA 16, 580e591.
Hagman, J., Nerentorp, M., Arvidsson, R., Molander, S., 2013. Do biofuels require Quantis, 2011. Quantis Water Databasedtechnical Report. Lausanne, Switzerland.
more water than do fossil fuels? Life cycle-based assessment of jatropha oil Available at: www.quantis-intl.com/waterdatabase.php (accessed 19.06.14.).
production in rural Mozambique. J. Clean. Prod. 53, 176e185. Su, M.H., Huang, C.H., Li, W.Y., Tso, C.T., Sheng Lur, H.S., 2014. Water footprint
Herath, I., Deurer, M., Horne, D., Singh, R., Clothier, B., 2011. The water footprint of analysis of Bioethanol Energy Crops in Taiwan. J. Clean. Prod. http://dx.doi.org/
hydroelectricity: a methodological comparison from a case study in New Zea- 10.1016/j.jclepro.2014.06.020. In Press.
land. J. Clean. Prod. 19, 1582e1589. Tzimas, 2011. Sustainable or not? Impacts and Uncertainties of Low-carbon energy
Herath, I., Green, S., Singh, R., Horne, D., van der Zijpp, S., Clothier, B., 2013. Water technologies on water. In: Presentation at the AAAS Annual Meeting, 17e21
footprinting of agricultural products: a hydrological assessment for the water February, Washington, USA.
footprint of New Zealand's wines. J. Clean. Prod. 41, 232e243. Verones, F., Hanaah, M.M., Pster, S., Huijbregts, M.A.J., Pelletier, G., Koehler, A.,
IPPC, 2007. Climate Change 2007: Synthesis Report. Valencia, Spain, 12e17 2010. Characterization factors for Thermal Pollution in freshwater Aquatic En-
November 2007, Available from: www.ipcc.ch/pdf/assessment-report/ar4/syr/ vironments. Environ. Sci. Technol. 44, 9364e9369.
ar4_syr.pdf (accessed 19.06.14.). Zonderland-Thomassen, M.A., Lieffering, M., Ledgard, S.F., 2014. Water footprint of
Jeswani, H.K., Azapagic, A., 2011. Water footprint: methodologies and a case study beef cattle and sheep produced in New Zealand: water scarcity and eutrophi-
for assessing the impacts of water use. J. Clean. Prod. 19, 1288e1299. cation impacts. J. Clean. Prod. 73, 253e262.
Koehler, A., Aoustin, E., 2008. UNEP/SETAC Life Cycle Initiative SETAC Europe Annual
Meeting, Project Group on Assessment of Use and Depletion of Water Resources
Within LCA. Warsaw, Poland.

You might also like