Journal of Environmental Management 91 (2010) 11931201

Contents lists available at ScienceDirect

Journal of Environmental Management

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

Multiple-objective evaluation of wastewater treatment plant control alternatives

Xavier Flores-Alsina a, c, Alejandro Gallego b, Gumersindo Feijoo b, *, Ignasi Rodriguez-Roda a, d
a

Laboratory of Chemical and Environmental Engineering, University of Girona, Montilivi Campus s/n 17071, Girona, Spain
Department of Chemical Engineering, School of Engineering, University of Santiago de Compostela, 15782 Santiago de Compostela, Spain
c
modelEAU, Departement Genie Civil, Pavillon Adrien-Pouliot, 1065, avenue de la Medecine, Quebec G1V 0A6, QC, Canada
d
ICRA, Catalan Institute for Water Research, H2O Building, Emili Grahit 101, 17003 Girona, Spain
b

a r t i c l e i n f o

a b s t r a c t

Article history:
Received 16 July 2009
Received in revised form
15 December 2009
Accepted 10 January 2010
Available online 18 February 2010

Besides the evaluation of the environmental issues, the correct assessment of wastewater treatment
plants (WWTP) should take into account several objectives such as: economic e.g. operation costs;
technical e.g. risk of suffering microbiology-related TSS separation problems; or legal e.g. accomplishment with the efuent standards in terms of the different pollution loads. For this reason, the main
objective of this paper is to show the benets of complementing the environmental assessment carried
out by life cycle assessment with economical, technical and legal criteria. Using a preliminary version of
the BSM2 as a case study, different combinations of controllers are implemented, simulated and evaluated. In the following step, the resulting multi-criteria matrix is mined using multivariate statistical
techniques. The results showed that the presence of an external carbon source addition, the type of
aeration system and the TSS controller are the key elements creating the differences amongst the
alternatives. Also, it was possible to characterize the different control strategies according to a set of
aggregated criteria. Additionally, the existing synergies amongst different objectives and their consequent trade-offs were identied. Finally, it was discovered that from the initial extensive list of evaluation
criteria, only a small set of ve are really discriminant, being useful to differentiate within the generated
alternatives.
2010 Elsevier Ltd. All rights reserved.

Keywords:
Wastewater
Multi-criteria
Life cycle assessment
Cluster analysis
Principal component/factor analysis
Discriminant analysis

1. Introduction
The increasing pace of industrialization, urbanization and population growth that our planet has faced over the last one hundred
years has increased considerably environmental pollution and
habitat destruction, and negatively affected water, air and soil
quality. In this context, wastewater treatment has become one of
the most important environmental issues, insofar as it reduces
pollution of natural water resources i.e. inland surface waters,
groundwater, transitional water and coastal water promotes
sustainable water re-use, protects the aquatic environment, and
improves the status of aquatic ecosystems. However, wastewater
treatment is based on a variety of processes with new environmental impacts due to energy consumption, use of chemical
reagents, sludge production and environmental emissions. Therefore, it is also necessary to analyze the system environmentally to
determine the overall pollution associated to these activities.

* Corresponding author. Tel.: 34 981 563100x16020; fax: 34 981 547168.

E-mail addresses: xavi@lequia.udg.es, xavier.ores@gci.ulaval.ca (X. FloresAlsina), alejandro.galleo.schmid@gmail.com (A. Gallego), gumersindo.feijoo@usc.es
(G. Feijoo), ignasi@lequia.udg.es, irodriguezroda@icra.cat (I. Rodriguez-Roda).
0301-4797/$ see front matter 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jenvman.2010.01.009

Life Cycle Assessment (LCA) has proved to be a helpful tool to

assess the environmental performance of wastewater treatment
facilities (see for example Karrman and Jonsson, 2001; Lundin et al.,
2000; Tillman et al., 1998; Pasqualino et al., 2009) or even total
wastewater treatment network systems (Lima and Park, 2009). LCA
can be dened as a compilation and evaluation of inputs, outputs
and environmental impacts of a system through its life cycle: from
the production of raw materials to the disposal of waste generated
(ISO 14040, 2006). Some authors have used LCA methodology in
the wastewater engineering eld for comparing design alternatives
for organic carbon and nitrogen removal from the environmental
point of view (Gallego et al., 2008; Renou et al., 2008; Vidal et al.,
2002), treatment processes for sludge (Hospido et al., 2005) or
oz et al., 2007).
emerging technologies (Mun
Nevertheless, the evaluation of wastewater treatment plants
(WWTP) involves the consideration of multiple objectives simultaneously (Flores-Alsina et al., 2008). The main purpose of wastewater treatment is to remove pollutants that can harm the aquatic
environment. Thus, the selected alternative needs to comply with
current regulatory standards as well as to minimize the environmental impact on the receiving water body (Gernaey et al., 2004).
Furthermore, operating costs have to be minimized; in particular,

1194

X. Flores-Alsina et al. / Journal of Environmental Management 91 (2010) 11931201

energy savings must be examined e.g. aeration, pumping, heating

and mixing. Chemicals, such as metal salts for phosphorus precipitation, the external carbon source to enhance denitrication efciency and the costs related to the collection and disposal of sludge
(Vanrolleghem and Gillot, 2002) must also be considered. Finally, it
is important to include all naturally occurring microbiology-related
solid separation problems, generally caused by microorganism
population imbalances between lamentous and oc-forming
bacteria. These problems lead to bulking, foaming and rising sludge
episodes, thereby causing undesirable operating conditions in the
activated sludge system (Jenkins et al., 2003; Wanner, 1994).
Previously published works have emphasized the need to
combine LCA with other types of information such as operational
costs (Tan et al., 2004) or technical reliability (Basson and Petrie,
2007) in order to conduct more realistic evaluations. Despite the
importance of considering different types of objectives i.e.
economic, technical and legal during the environmental assessment of WWTP alternatives, this topic has been treated scarcely in
the wastewater treatment eld. Hence, there is a clear need to ll
this gap, conducting integrated assessments that combine the
environmental impact on water, air and soil provided by LCA with
other sources of information to quantify the economical feasibility,
technical reliability and accomplishment of the efuent discharge
limits of the different WWTP alternatives.
The objective of this paper is, therefore, to complement the
environmental assessment carried out by LCA during the evaluation of WWTP alternatives with economic, technical and legal
criteria. The evaluation of twelve control strategies at plant level is
used as a case study using a preliminary version of the International
Water Association (IWA) Benchmark Simulation Model No 2
(Jeppsson et al., 2006). The multi-criteria matrix obtained during
the evaluation of the alternatives is mined using multivariable
statistical techniques to nd groups or clusters of alternatives with
a similar behaviour, nd the main correlations between alternatives and criteria, and nally identify which are the most useful/
important criteria that allow differentiating between alternatives.

treated during one year in all the different alternatives. This functional unit selection seems to be in agreement with other WWTP LCA
studies (Renou et al., 2008; Roeleveld et al., 1997).
The system considered is a gate to gate analysis including all
the processes from the entrance of the inuent to the WWTP until
its exit as an efuent. The system is expanded to include the
treatment and disposal of the sludge generated. The consumption
of electricity and chemicals is also included, that is to say, their
production and transportation upstream.
The activated sludge considered is a modied LudzackEttinger
conguration comprised of 5 continuous stirred tank reactors
(CSTRs) in series and a secondary settler. The 1st and the 2nd reactor
(ANOX1 and ANOX2) are anoxic with a total volume of 2000 m3,
while the 3rd, 4th and 5th (AER1, AER2 and AER3) tanks are aerobic
with a total volume of 3999 m3 (1333 m3 each). The circular
secondary settler (SEC) has a surface area of 1500 m2 with a total
volume of 6000 m3. The BSM2 plant contains a primary clarier
(PRIM), a sludge thickener (THK), an anaerobic digester (AD),
a storage tank (ST) and a dewatering unit (DH). Further information
about BSM2 layout and the description of the process models can be
found in Jeppsson et al. (2006). The inuent wastewater to be treated,
which comprises 609 days dynamic inuent data with samples taken
every 15 min, has an average ow rate of 20 648 m3 day1 with
a carbon and nitrogen load of 12 234 kg COD$day1, 493 kg N day1
and 231 kg P day1 according to the principles outlined in Gernaey
et al. (2006a).
The default open loop (A1) conditions involve constant external
and internal recirculation, waste ow and aeration ow. Reactors
ANOX1 and ANOX2 are not aerated but fully mixed. Then, eleven
alternative control strategies [A A2 ,., A12] have been implemented and compared to the default open loop base case. These
strategies correspond to the combination of different loops to
control
 the dissolved oxygen concentration (DO) in the aerobic reactors (AER1, 2 and 3) manipulating the aeration ow (KLa).
 the DO set point based on ammonia (NH4) or oxygen uptake
rate (OUR) measurements in AER1,2 and 3 respectively,
 the total suspended solids (TSS) in AER3 manipulating the
waste ow (QW),
 the control nitrate concentration (NO
3 ) in ANOX2 manipulating internal recycling ow rate (Qintr) or external carbon
source (Qcarb).

2. LCA of the case study

2.1. Goal and scope
The main purpose of this case study is to compare the behaviour
and impact of different controllers implemented in a preliminary
version of the activated sludge section of the IWA Benchmark
Simulation Model No 2 (BSM2). The function of a WWTP is the
treatment of an inuent (with the objective of organic load, nutrient
and suspended solids reduction) in order to reach satisfactory values
before release in natural water courses. Therefore, the functional unit
selected for comparison of the WWTP with the different congurations is the treatment of 753.3 Hm3 of water, which is the amount

The details of the twelve resulting control strategies are

summarized in Table 1.
When considering which parts of the system are going to be
considered in the LCA (system boundaries), very different choices
can be made in wastewater systems and these choices will inevitably
affect results (Lundin et al., 2000). In this sense, there are several

Table 1
Control strategies evaluated in this case study.
Characteristics

3 DO

Ammonium controller

Qintr controller

Qcarb controller

TSS controller

Surmacz controller

Reference

Vanrolleghem and
Gillot (2002)
DO in AER1, 2 and 3
DO in AER1, 2 and 3
2, 2 and 2 g
(COD) m3

Vrecko et al. (2006)

Copp (2002)

Vrecko et al. (2006)

Vrecko et al. (2006)

NH
4 in AER3
NH
4 in AER1, 2 and 3
1 g N m3

NO
3 in ANOX2
NO
3 in ANOX2
1 g N m3

NO-3 in ANOX2
NO
3 in ANOX2
1 g N m3

DO set point in 3
DO strategy
Cascaded PI
A5, A6, A7 and A8

Qintr

Qcarb

TSS in AER3
TSS in AER3
4400 g TSS m3
(if T < 15  C) 3400 g TSS m3
(if T > 15  C)
Qw

Vanrolleghem and
Gillot (2002)
OUR in AER1
DO in AER1, 2 and 3
1850 g COD m3 d1

PI
A3, A5, A7, A9
and A11

PI
A4, A6, A8, A10
and A12

Measured variable(s)
Controlled variable(s)
Set point/critical value

Manipulated variable

KLa

Control algorithm
Applied in control
strategies (Aj)

PI
A2, A3, A4, A5, A6, A7, A8,
A9, A10, A11 and A12

Cascaded PI
A7, A8, A11, and A12

DO set point in 3
DO strategy
ON/OFF cascaded PI
A9, A10, A11 and A12

X. Flores-Alsina et al. / Journal of Environmental Management 91 (2010) 11931201

studies (Dennison et al., 1998; Mels et al., 1999; Roeleveld et al., 1997)
that consider that the sludge is a waste. However, other studies
(Hospido et al., 2005; Lundin et al., 2000; Tillman and Ekvall, 1994)
consider a positive impact of the recycling of nutrients in sludge,
because the production of fertilizers is avoided. As mentioned above,
the selection of either option will affect the results. In this study, the
sludge produced in the WWTPs is considered to be applied as
fertilizer for agriculture because it is the most used alternative for
sludge management in Europe (Padmanabhan, 2005).
Finally, the impact of the construction was not included because
the different congurations imply negligible variations in the
construction of the WWTP. The impact of wastewater pumping to
the treatment plants was not considered, as this action is not
affected by WWTP operation.
2.2. Inventory
To carry out the inventory, data were collected using dynamic
simulation. Further information about the BSM models can be
found in Jeppsson et al. (2006). All the dynamic simulations were
preceded by a steady state simulation (200 days). This ensures
a consistent starting point and eliminates partiality due to the
selection of initial conditions in dynamic modelling results. Only
the data generated during the last 365 days of the simulation were
used for plant performance evaluation.
Some of the necessary data to carry out the inventory were not
provided by the models and was taken from literature and databases, considering Europe and the most recent data for geographical and time requirements:
 Electricity: For electricity production data from the Ecoinvent
Database (Hischier et al., 2007)) have been used, updating
them with the European production prole for 2008 (International Energy Agency, 2009)
 Chemicals: Data of the production for external carbon source
(methanol) to enhance denitrication were obtained from
Ecoinvent database (Nemecek and Kagi, 2007).
 Fertilizers avoided: The application of sludge to soil as fertilizer
reduces the need of using synthetic fertilizers which results in
environmental benet if the content of heavy metals in the
sludge is kept within admissible values. The substitutability
was assumed to be 70% for phosphorus and 50% for nitrogen
(Bengtsson et al., 1997). Entries from the Ecoinvent database
(Nemecek and Kagi, 2007) for the most consumed N-based and
P-based fertilizers in Europe (calcium ammonium nitrate and
triple superphosphate; Eurostat (2008)) were used to represent the avoided fertilizer. Sludge brings also organic matter
but this effect has not been considered environmentally due to
lack of data.
 Heavy metals in the sludge: Heavy metals can be at high
variable concentration depending normally on the amount of
specic industrial wastewaters in the overall ow. In this case
study, the results come from an analytical study carried out in
several European Countries that summarizes the average
concentrations that can be found in dehydrated sludge (European Commission, 2000).
 Methane and nitrogen compound (N2, NOx and NH3) emissions: Data from a variety of sources were used to calculate
emissions resulting from sludge application for agricultural
purposes (Hobson, 2000; Lundin et al., 2000; Mossier, 1993).
 Phosphorus: The load of P in the efuent was calculated by
subtracting the P available in the inuent by the P present in
the sludge, which was obtained also from the above mentioned
study of dehydrated sludge in several European Countries
(European Commission, 2000).

1195

2.3. Impact assessment

CML 2 baseline 2000 methodology developed by Institute of
Environmental Sciences (CML) of Leiden University was used
(Guinee et al., 2001) to quantify the degree of satisfaction of the
environmental objectives. In particular, the following impact
categories were considered because they are usually used in
wastewater LCAs: abiotic depletion (criterion X11), global warming
(X12), ozone layer depletion (X13), terrestrial ecotoxicity (X14),
photochemical oxidation (X15), acidication (X16) and eutrophication (X17) (Table 2).
The consumption of electricity is the major process contributing
to abiotic depletion (between 96 and 98% of the impact), global
warming (7791%), ozone layer depletion (7599%) and photochemical oxidation (8197%) and is a signicant contributor to
acidication (3342%). Abiotic depletion increases due to the
consumption of coal, natural gas and oil for the production of energy.
The CO2 emission from the combustion of these raw materials is the
main contributor to global warming. Emission of gases that reduce
the ozone layer (principally Halon1211 and Halon 1301) is minimal,
mainly associated with the production of electricity with oil. The
emission of SOx caused by the production of energy is behind
the correlation with photo-oxidant formation and acidication. The
application of sludge to agricultural purposes is the major contributor to terrestrial ecotoxicity (9899%) due to the presence of heavy
metals and to acidication (5665%) due to the emission of NH3.
Eutrophication is attributed to the discharge of the treated efuent
(9899%), being the emissions of PO4
3 the major contributor. The
results obtained are in agreement with other wastewater treatment
LCA (Gallego et al., 2008; Hospido et al., 2008; Machado et al., 2007).
3. Quantication of economical, technical and legal criteria
This section describes the additional objectives used during the
evaluation of the alternative scenarios. In this case study, three
Table 2
Evaluation criteria used.
Criterion (XI)

UNITS (year1)

X11: abiotic depletion (Heijungs et al. (1992)

X12: global warming (World Meteorological
Organisation, 1999)
X13: ozone layer depletion (Solomon et al., 2007)
X14: terrestrial ecotoxicity (Huijbregts, 1999)
X15: photochemical oxidation (Jenkin and Hayman, 1999)
X16: acidication (Guinee et al., 2001)
X17: eutrophication (Huijbregts et al., 2001)

kg Sb eq.
kg CO2 eq.
kg
kg
kg
kg
kg

X21:
X22:
X23:
X24:
X25:
X26:
X27:
X28:
X29:

Tg TSS
Mw.h
Mw.h
Tg
Mw.h
Mw.h
Tg
Mw.h

sludge production (SP) (Vrecko et al., 2006)

aeration energy (AE) (Vrecko et al., 2006)
pumping energy (PE) (Vrecko et al., 2006)
carbon source (CS) (Vrecko et al., 2006)
mixing energy (ME) (Vrecko et al., 2006)
heating energy (HE) (Vrecko et al., 2006)
methane production (MP) (Vrecko et al., 2006)
net energy consumption (E) (Vrecko et al., 2006)
operating cost (Vrecko et al., 2006)

CFC-11 eq.
1,4-DCB eq.
C2H4 eq.
SO2 eq.
PO4 eq.

X31: overall bulking risk (Comas et al., 2008)

X31,1: low C/N bulking risk (Comas et al., 2008)
X31,2: low DO bulking risk (Comas et al., 2008)
X31,3: low F/M bulking risk (Comas et al., 2008)
X32: overall foaming risk (Comas et al., 2008)
X32,1: low FM foaming risk (Comas et al., 2008)
X32,2: high Ss/Xs foaming risk (Comas et al., 2008)
X33: rising sludge risk (Comas et al., 2008)

days
days
days
days
days
days
days
days

X41:
X42:
X43:
X44:

days
days
days
days

TIV
TIV
TIV
TIV

COD (Copp, 2002)

TSS (Copp, 2002)
BOD (Copp, 2002)
TN (Copp, 2002)

1196

X. Flores-Alsina et al. / Journal of Environmental Management 91 (2010) 11931201

additional objectives are taken into account, specically economical, technical and legal issues. Thus, the degree of satisfaction of
the economical objectives was quantied by the following criteria:
sludge production (X21), aeration energy (X22), pumping energy
(X23), carbon source (X24), mixing energy (X25), heating energy
(X26), methane production (X27) and net energy consumption
(X28). Finally, an aggregated operational cost (X29) was calculated
as a weighted sum of the previous criteria, based on the equation
dened by Vrecko et al. (2006). The degree of satisfaction of the
technical objectives was quantied by means of the risk of microbiology-related solid separation problems (Comas et al., 2008). It
includes bulking (overall risk X31, and the different specic causes:
low C/N ratio X31,1, low DO X31,2 and low F/M X31,3), foaming
(again overall risk X32 and two different causes: low F/M X32,1 and
high SS/XS X32,2) and rising sludge (X33). Finally, the percentage
of time that the plant violates the limits xed by the law
(TIV) measures the accomplishment of the legal objectives for COD
(X41), TSS (X42), BOD (X43) and TN(X44) (Copp, 2002). The list of
the evaluation criteria with the corresponding units used in this
case study, as well as the list of references, is summarized in Table 2.
4. Results
The result of the assessment is a huge and complex matrix
consisting of a large number of quantied environmental,
economical, technical and legal criteria, which are often difcult to
interpret and thus drawing meaningful conclusions (Table 3). In
order to facilitate both the analysis and interpretation of this multicriteria matrix, the resulting database is mined using multivariable
statistical techniques (cluster analysis, principal component/factor
analysis and discriminant analysis).
4.1. Cluster analysis
Cluster analysis (CA) is an unsupervised pattern recognition
technique that uncovers intrinsic structure or underlying behaviour
of a data set without making a priori assumptions. Classication of
the objects or a system into categories or clusters is based on the
nearness or similarity of data points; see for example Hair et al.
(1998). The result is the assignment of a set of observations into
classes (called clusters) so that observations in the same cluster are
similar in some sense. In this paper hierarchical clustering is
performed on the data set after scaling the variables between
0 and 1 by means of the Wards method using the Euclidian
distance as a measure of similarity.
Cluster Analysis (CA) rendered a dendogram where all the
proposed alternatives are grouped into two main statistically
signicant clusters (Fig. 1). The rst (alternatives A4, A6, A8, A10
and A12) and the second cluster (Alternatives A1, A2, A3, A5, A7, A9
and A11) correspond to control strategies with and without
external carbon source addition (cluster 2.1 and 2.2). If these
clusters are further classied, a new group of alternatives can be
found based on TSS and the aeration controller. Thus, on the one
hand the rst cluster is subdivided in a subgroup containing
alternatives (A4, A10, A6 and A12) and (A8) i.e. with external carbon
source, TSS and SNH controller, named cluster 4.1 and 4.2
respectively. On the other hand, if the second cluster is subdivided, two more groups are obtained comprising strategies A7
and A11 (cluster 4.3) and A1, A2, A3, A5, and A9 (cluster 4.4) are
obtained. Thus, clustering analysis indicates that there are four
main types of control alternatives, where the presence of external
carbon source addition, the type of aeration control and the TSS
controller are the key elements creating the differences between
the clusters.

4.2. Principal component/factor analysis

Principal Component Analysis (PCA) extracts the eigenvalues
and eigenvectors from the covariance matrix of the auto scaled
variables. The principal components (PCs) are the uncorrelated
(orthogonal) variables obtained by multiplying the original correlated variables with the eigenvectors. Each eigenvector consists of
a vector of coefcients (loadings). PCA allows reducing the
dimensionality of the original data using a new set of aggregated
indicators with a minimum loss of information. Factor analysis (FA)
further reduced the contribution of less signicant variables
obtained from PCA and resulted in the new groups of variables
known as varifactors (VF) extracted through rotating the axis
dened by PCA.
Principal Component and Factor Analysis (PCA/FA) is applied to
the auto scaled simulation output to compare the evaluation
criteria between the implemented control strategies and to identify
the most inuential factors. PCA of the entire data set evolved to
four principal components (PCs) with eigenvalues lower than one.
A varimax rotation of the PCs to four different varifactors (VFs)
explained about 93.4% of the total variance. The values of the PCs
are further cleaned up with this technique and in VFs original
variables are contributing more clearly (see Table 4). The factor
loadings allows us to know the grade of relationship of the different
criteria which can be classied as strong, moderate and weak
corresponding to absolute loading values higher than 0.7, between
0.5 and 0.7 and lower than 0.5 respectively (Liu et al., 2003). The
variable X31,1 is excluded from this analysis because it exhibits
a constant value (i.e. variance zero).
VF1, which explains 40.89% of the total variance, has strong (in
bold) positive loadings in X13, X14, X21, X24, X26, X27, X29 and
negative loadings for X17, X33, X44. This VF correlates the operating costs and denitrication efciency. The periodic addition of
an external carbon source (X24) implies the subsequent increase of
sludge production (X21), the heating energy (X26) and the overall
operating cost (X29), although the methane production is
increased (X27) resulting from the additional activity in the
digester. On the other hand, this addition enhances the overall
plants denitrication efciency because an extra-electron donor is
provided; reducing the percentage of time that the efuent
nitrogen is in violation (X44) as well as reducing the risk of nitrate
that potentially can be denitried in the secondary settler (X33). In
terms of environmental impact, the improvement of the efuent
nitrogen reduces the potential eutrophication (X17) as a trade-off
of increasing the ozone layer depletion (X13), due to the gases
emitted in the production of the external carbon source, and
terrestrial ecotoxicity (X14) due to the increase of the sludge
production with heavy metals. VF2 explains 24.94% of the total
variability and presents strong positive loadings for X11, X12, X15,
X16, X22, and X28, and negative load for X25. This VF highlights
that high aeration energy (X22), and thus high net energy
consumption (X28), is the main responsible of abiotic depletion
(X11), global warming (X12), photochemical oxidation (X15) and
acidication (X16). VF3 (explains 17.82% of the total variance)
basically identies the low F/M ratio as the main cause of risk of
foaming. Also, it can be observed that the operation conditions that
potentially can drive to foaming problems present low organic
matter pollution removal efciency. Thus, criteria X32,1 (low F/M
foaming), X32 (overall risk of foaming), X41 (TIV COD), X42 (TIV
TSS) and X43 (TIV BOD5) are well represented in this varifactor.
Finally, VF4 explains a 9.83% of the total variance, and presents
strong loadings for low F/M bulking risk (X31,3) and overall risk of
bulking (X31). To sum up, it can be said that low F/M ratios are the
main responsible of both bulking and foaming problems. Nevertheless the operating conditions that potentially can lead to

Table 3
Values of the evaluation criteria for the 12 tested control strategies.
A2

A3

A4

A5

A6

A7

A8

A9

A10

A11

A12

abiotic depletion
global warming
ozone layer depletion
terrestrial ecotoxicity
photochemical oxidat.
acidication
eutrophication

11 854
1 809 227
0.18
993 638
360
22 834
271 487

10 946
1 687 175
0.17
993 205
334
22 166
271 892

10 624
1 643 563
0.16
991 811
324
21 911
272 003

11 494
2 042 888
0.24
1 054 650
414
23 935
250 109

12 936
1 954 496
0.20
993 693
391
23 623
263 962

14 269
2 406 153
0.28
1 048 665
491
25 858
246 498

13 126
1 980 281
0.20
994 848
396
23 778
263 894

15 152
2 576 472
0.30
1 094 815
528
27 237
248 272

10 415
1 615 482
0.16
991 617
318
21 756
272 163

11 149
1 973 142
0.23
1 049 961
399
23 576
255 991

10 578
1 637 750
0.16
992 986
323
21 895
269 609

10 037
1 812 974
0.21
1 068 369
364
23 001
258 200

X21:
X22:
X23:
X24:
X25:
X26:
X27:
X28:
X29:

sludge production
aeration energy
pumping energy
carbon source
mixing energy
heating energy
methane production
net energy consumpt.
operating cost

968.89
3120.17
145.16
0.00
236.52
1552.35
413.03
2575.99
3930.32

968.82
2927.52
145.16
0.00
236.52
1552.20
413.03
2383.19
3737.60

967.58
2914.60
89.58
0.00
236.52
1551.72
412.96
2314.66
3665.70

1028.72
3117.43
145.27
435.74
236.52
1589.65
430.96
2503.13
5306.74

968.53
3402.24
93.60
0.00
236.52
1552.31
413.18
2805.59
4158.81

1021.78
3705.48
145.26
422.60
236.52
1588.99
430.74
3091.83
5835.99

969.59
3448.12
90.56
0.00
236.52
1552.53
413.62
2846.01
4202.25

1066.64
3960.62
146.94
490.01
236.52
1625.42
447.89
3282.15
6326.55

967.47
2864.89
92.42
0.00
239.14
1551.65
412.96
2270.33
3621.20

1024.26
3036.47
145.26
399.42
238.66
1587.31
429.68
2429.63
5113.29

968.75
2904,0.96
89.42
0.00
238.88
1551.83
413.33
2305.14
3659.49

1042.73
2836.34
146.58
375.33
239.29
1605.31
438.80
2194.69
4843.55

X31: overall bulking risk

X31,1: low C/N bulking risk
X31,2: low DO bulking risk)
X31,3: low F/M bulking risk
X32: overall foaming risk
X32,1: low FM foaming risk
X32,2: high Ss/Xs foaming risk
X33: rising sludge risk

130.83
0.00
23.22
108.53
126.37
125.99
0.39
328.92

112.11
0.00
4.01
108.10
125.93
125.56
0.36
334.31

112.69
0.00
4.95
107.74
127.48
127.04
0.44
319.41

139.86
0.00
3.02
136.87
162.22
161.80
0.42
226.54

121.94
0.00
14.95
107.24
127.73
127.28
0.45
325.28

156.23
0.00
10.79
145.60
170.21
169.78
0.43
236.28

191.53
0.00
5.36
187.45
154.45
153.95
0.50
333.62

129.07
0.00
2.56
127.26
103.77
103.16
0.61
240.85

113.17
0.00
6.26
106.96
127.05
126.60
0.45
323.14

136.53
0.00
3.90
132.66
159.03
158.66
0.38
218.05

190.76
0.00
5.90
186.64
154.15
153.60
0.54
329.76

138.24
0.00
4.91
134.73
114.33
114.03
0.30
202.41

X41:
X42:
X43:
X44:

0.00
0.02
0.00
293.23

0.00
0.02
0.00
313.78

0.00
0.02
0.00
310.47

1.01
6.19
1.98
101.81

0.00
0.02
0.00
231.26

2.04
7.58
3.36
82.29

0.00
2.18
0.11
247.86

0.00
2.19
0.18
109.02

0.00
0.02
0.00
282.47

0.94
5.73
1.89
105.98

0.00
2.14
0.09
281.69

0.00
2.14
0.18
140.48

TIV
TIV
TIV
TIV

COD
TSS
BOD
TN

X. Flores-Alsina et al. / Journal of Environmental Management 91 (2010) 11931201

A1
X11:
X12:
X13:
X14:
X15:
X16:
X17:

1197

Reescaled distance cluster combine

1198

X. Flores-Alsina et al. / Journal of Environmental Management 91 (2010) 11931201

20

Cluster 4.1 Cluster 4.2

Cluster 4.3

Cluster 4.4

10

A4

A10

A6

A12

A8

A7 A11 A3
Alternative

A9

A1

A2

A5

to the high aeration ow rate necessary to reduce the efuent

ammonium concentration and thus reach the desired set point. On
the other hand, alternatives with an ON/OFF cascaded PI controller
(A9, A10, A11 and A12) present lower values in VF2 because this
control strategy implies lower aeration costs, but higher mixing
costs. This is due to the intermittent anoxic zones need to be
maintained fully mixed when the aerobic zone is used to denitrify.
The low F/M ratios, the risk of foaming and the decrease in overall
organic matter pollution removal efciency give A4, A6 and A10
(controller strategies with an external carbon source and without
TSS controller) high scores in VF3. Finally, the increase in the TSS set
point during the winter periods decreases the F/M ratio of alternatives A7 and A11 and subsequently the risk of bulking is increased.
For this reason, cluster 4.3 gets high scores in VF4. Thus, it can be
said that alternatives with external carbon source and TSS
controller (cluster 4.3) and without external carbon source and
with TSS controller (A4, A6, A10) presents both high risk of suffering
microbiology-related TSS separation problems.

Fig. 1. Dendrogram obtained of the cluster analysis.

4.3. Discriminant analysis

foaming problems are accompanied with poorer organic matter
removal efciency.
Once the principal components are identied and labelled, the
scores obtained by the implemented control strategies can be
calculated as a linear combination of the original variables. The
representation of the scores of the different control strategies is
depicted in Fig. 2. As expected, the results of the PCA/FA are in good
agreement with the cluster analysis (CA). Alternatives with an
external carbon source addition (clusters 4.1 and 4.2) present high
scores in VF1, and are characterized by having a low potential
eutrophication risk as a trade-off for high operating costs, terrestrial ecotoxicity and ozone layer depletion. Alternatives with
ammonium controller (A5, A6, A7 and A8) present higher scores in
VF2 associated with abiotic depletion, global warming, photochemical oxidation and acidication potential. This is attributable

Table 4
Loadings of the evaluation criteria on the rst ve rotated PC for the complete
data set.
Criterion (XI)

VF1

VF2

VF3

VF4

X11:
X12:
X13:
X14:
X15:
X16:
X17:

abiotic depletion
global warming
ozone layer depletion
terrestrial ecotoxicity
photochemical oxidation
acidication
eutrophication

0.21
0.60
0.75
0.98
0.62
0.61
L0.81

0.97
0.78
0.63
0.19
0.77
0.78
0.43

0.06
0.15
0.20
0.01
0.16
0.08
0.36

0.07
0.04
0.03
0.02
0.04
0.06
0.08

X21:
X22:
X23:
X24:
X25:
X26:
X27:
X28:
X29:

sludge production
aeration energy
pumping energy
carbon source
mixing energy
heating energy
methane production
net energy consumption
operating cost

0.98
0.28
0.68
0.95
0.23
0.97
0.97
0.22
0.85

0.18
0.95
0.08
0.17
0.74
0.21
0.21
0.97
0.49

0.01
0.03
0.18
0.25
0.10
0.05
0.06
0.06
0.20

0.02
0.11
0.47
0.02
0.30
0.03
0.04
0.07
0.01

X31: overall bulking risk

X31,2: low DO bulking risk
X31,3: low F/M bulking risk
X32: overall foaming risk
X32,1: low F/M foaming risk
X32,2: high Ss/Xs foaming risk
X33: rising sludge risk

0.08
0.46
0.02
0.09
0.09
0.06
L0.95

0.11
0.28
0.05
0.02
0.02
0.60
0.07

0.38
0.09
0.34
0.94
0.94
0.28
0.26

0.82
0.47
0.89
0.30
0.30
0.61
0.07

X41:
X42:
X43:
X44:

0.38
0.58
0.44
L0.87

0.20
0.13
0.17
0.24

0.86
0.78
0.86
0.38

0.13
0.16
0.10
0.04

TIV
TIV
TIV
TIV

COD
TSS
BOD
TN

Discriminant Analysis (DA) is used to determine the variables

which allow discriminating between two or more naturally
occurring groups (Johnson and Wichern, 1992). It operates on raw
data and the technique constructs a discriminant function (Eq. (1))
for each group.

Dz Ci;k

n
X

wi;k Xi;k

(1)

k1

where z is the number of the function, Ck is the constant inherent to

each function, n is the number of parameters used to classify a set of
data into a given group and wi is the weight coefcient assigned by
DA to a given performance evaluation parameter (Xi). In this case
study, the number of groups are the classes obtained by CA and the
parameters are the evaluation criteria.
Finally, Discriminant Analysis (DA) is performed aiming at
dividing the original data set into the four groups (cluster 4.1, 4.2,
4.3 and 4.4) obtained by CA. The control strategy is the grouping
variable, while all the evaluation criteria are the independent
variables. DA is performed using all the evaluation criteria except
X31,1, and it has rendered the corresponding classication matrixes
(CM) assigning 100% of the cases correctly. The stepwise DA shows
that carbon source, heating energy, low F/M bulking risk, rising
sludge risk and time in violation for BOD (X24, X26, X31,3, X33,
and X43) are the most discriminant criteria. The correct grouping
pattern of DA coincides with the clusters obtained in CA. The
discriminant functions are listed in Table 5.
According to DA, the external carbon source addition and the
TSS controller show a clear effect on the overall plant performance.
As mentioned before, the immediate effect of an external carbon
source addition (X24) is the increase of the biomass in the reactor
and consequently arriving to the digester and thus increasing
heating energy (X26). As a side effect of the improved overall
nitrogen removal efciency, there is a decrease of the nitrate
potentially denitriable in the secondary settler (X33) but an
increase the bulking risk due to low F/M ratio (X31,3) and a worse
efuent biochemical organic demand (X43).
Fig. 3 represents the scores of each control strategy to a certain
discriminant function (Dk). Thus, D1, with the highest discriminant
ability (95.6%) separates cluster 4.1 and 4.2 from cluster 4.3 and 4.4.
This is mainly due to the effect of the external carbon source
controller in the overall plant performance. The risk of separation
problems occurring and the better organic matter removal efciency
is the main reason to separate strategies without external carbon

X. Flores-Alsina et al. / Journal of Environmental Management 91 (2010) 11931201

a 2,0
1,0

cluster 4.1
cluster 4.2
cluster 4.3
cluster 4.4

0,5

A7

A5
A1

A8

1,0

-0,5
-1,0

A2
A 11

A4

A 10

A9

-2,0
-1,5

A 11
A7
A8

0,5

A 12

0,0

A 12

-1,5

-0,5

0,0

0,5

1,0

1,5

2,0

A9
A 10

-0,5

A3

-1,0

A5
A2

-1,5
-1,0

cluster 4.1
cluster 4.2
cluster 4.3
cluster 4.4

2,5

1,5

0,0

A3

3,0

2,0

A6

VF4

VF2

1,5

1199

-2,0
-2,0

A6

A4

A1
-1,5

-1,0

-0,5

0,0

0,5

1,0

1,5

2,0

2,5

3,0

VF3

VF1

Fig. 2. Principal component scores for the control strategies implemented for principal component 1 and 2 (a) and for principal component 3 and 4 (b).

source and TSS controller (cluster 4.4) to the rest of the alternatives
(clusters 4.1, 4.2 and 4.3) as shown by D2. Finally D3, with the lowest
variance explanation clearly separates cluster 4.2 from 4.1, although
cluster 4.3 and 4.4 obtained medium scores. This is mainly due to the
relatively better organic matter pollution efciency, but lower risk of
rising of the alternative with a TSS, SNH and external carbon source
controller compared to A2, A4, A6 and A12.
5. Discussion of the results
The results of this analysis provided several points of discussion.
In the rst place, from a process engineering point of view, it was
possible to evaluate the multi-objective behaviour of the different
implemented controllers complementing the evaluation provided
by LCA. Secondly, the use of multivariate statistical techniques
facilitated the analysis of the results, providing with additional
information not available at rst glance. In this sense, the combined
use of CA and PCA/FA was useful to characterize the initial set of
alternatives from different points of view e.g. eutrophication
potential, operating costs, risk of suffering microbiology-related
TSS separation problems. Additionally, the existing synergies
amongst different objectives and their consequent trade-offs were
identied. For example, all the alternatives included in cluster 4.3
and 4.4, i.e. with external carbon source addition, were characterized for presenting similar reduction of the eutrophication potential (X17), rising risk (X33) and accomplishment of the efuent TN
(X44) when they were compared to the alternatives included in
cluster 4.1 and 4.2. Nevertheless, the results of the analysis pointed
out some of the undesirable effects of these improvements such as
the increase of the operating costs (X29), terrestrial ecotoxicity
(X14) and ozone layer depletion (X13).
Another interesting example was the evaluation of the plants
overall process performance when the ammonium controller was
implemented (A5, A6, A7 and A8). In spite of improving the nitrication efciency, the results of the analysis surprisingly showed no
substantial environmental gain with this controller. In fact, both
simulations and multivariate analysis showed that high aeration
Table 5
Canonical classication functions for discriminant analysis of the implemented
WWTP alternatives.
Criterion (XI)

D1

D2

D3

X24: carbon source

X26: heating energy
X31.3: low F/M bulking risk
X33: rising sludge risk
X43: TIV BOD

2.06
11.93
9.47
0.29
15.08

1.17
1.81
2.40
1.88
2.48

0.43
0.78
0.06
1.56
1.27

energy (X22) was necessary to achieve the desired set point. As

a result, the controller turned to increase the environmental impact
i.e. abiotic depletion (X11), global warming (X13), photochemical
oxidation (X15) and acidication potential (X16). Thus, the results
of this study may lead us to conclude that for this particular case
there is no environmental benet of including this controller.
It is important to point out that the results of this base case
analysis depend strongly on the model selection prior to performing the simulations. When modelling activated sludge plants, there
is often disagreement on the best model to apply for a given case.
The representation of biomass decay (Siegrist et al., 1999), the
modelling of nitrogen removal (Henze et al., 2000) and the oversimplication of the settling models (i.e. non-reactive in most cases,
despite the fact that a signicant amount of biomass is often stored
at the bottom of the secondary clarier, e.g. Gernaey et al., 2006b)
are key issues that are still under discussion. For this reason, the use
of activated sludge model including phosphorus removal (Henze
et al., 2000) or dummy states acting as heavy metals (Jeppsson et al.,
2006) that would suppose better estimates of some of impact
categories, could result in different results and conclusions.
The main difference between this work and previous research
studies is the combined evaluation of different types of objectives
at the same time. For example the case study showed that some
special features of the TSS controller would have not been able to be
discovered with a regular LCA approach. From the generated
results, it could be seen that the TSS controller plays a different
technical role when combined or not with a nitrate controller i.e.
depending on the circumstances, there can be an increase in the
risk of suffering TSS separation problems. If the control strategies
are evaluated taking into account only environmental issues
(Fig. 2a) there is no substantial difference between cluster 4.3 and
4.4. Nevertheless, when an additional dimension is added (Fig. 2b)
alternatives A7 and A11 (cluster 4.3) are clearly disfavoured.
All in all, the results of the analysis presented a straightforward
fashion of characterizing alternatives. Thus, it is possible to know
which control strategy has the most signicant impact in the
different evaluation criteria For example, in case of being necessary
an environmentally friendly alternative, one would go for one of
the control strategies within cluster 4.1 and 4.2 i.e. more expensive
to operate but with lower eutrophication potential. Additionally,
the results of the analysis revealed that no substantial improvement is due to the implementation of the ammonium controller. In
fact, as mentioned before there were several collateral environmental damages derived from its implementation. For this reason,
it would be recommended to go for an alternative with just a DO
(A4) or an OUR (A10 and A12) controller, because it reduces several
impact categories related to aeration. Finally, if the analysis is taken

1200

X. Flores-Alsina et al. / Journal of Environmental Management 91 (2010) 11931201

40
30
20

cluster 4.1
cluster 4.2
cluster 4.3
cluster 4.4

A2
A1

A3

10

A5

A9

D3

D2

-30
-300

A6
A11
-200

A2
-100

A8

A7

A11

A5

A3
A2
A1

-2

A4

A7

-4

A12
100

A8

-10

A7

cluster 4.1
cluster 4.2
cluster 4.3
cluster 4.4

10

-20

12

200

-6
-300

A6
A2

-200

-100

A 12
A4

100

200

D1

D1

Fig. 3. Discriminant analysis of the WWTP control strategies implemented for the discriminant function 1 and 2 (a) and 1 and 3 (b).

even further, it can be seen that the implementation of TSS

controller can reduce the risk of having some microbiology-related
TSS separation problems substantially, being A12 the alternative
with more change of success. A similar analysis, but driving to
different conclusions could be done when some other of the
objectives are prioritized.
Finally, the impact that the initial list of evaluation criteria has
during the whole evaluation procedure should be mentioned. The
results of the PCA/FA and DA analysis showed that redundant
information is included during the analysis, and only few of the
initial set of criteria present a clear variation from one alternative to
another. Nevertheless, the reader should be aware that it is
impossible to know a priori which would be the main correlations
between evaluation criteria and the generated alternatives. Each
multivariate model is really case specic and some changes from
one study to another when different control strategies and evaluation criteria are involved have to be expected. For this reason, the
authors advocate for the use of techniques such as CA, PCA/FA and
DA to improve the accessibility to the information needed for
effective evaluation of control strategies. As a side effect, there is
also a reduction in the cognitive load on the decision maker,
yielding more knowledge than current evaluation methods and
enhancing understanding of the whole evaluation process.
6. Conclusions
This paper contributed in the eld of wastewater engineering
with an integrated approach that complements the environmental
assessment carried out by LCA during the evaluation of WWTP
alternatives with economic, technical and legal criteria. A preliminary version of the Benchmark Simulation Model No 2 layout was
selected as a case study where 12 control strategies were implemented, simulated and evaluated using multiple criteria. The result
was a huge and complex evaluation matrix that was mined using
different multivariate statistical techniques.
The results showed that the main differences in terms of process
performance were due to the addition of an external carbon source,
the aeration system and the TSS controller. Also, it was possible to
create aggregate indicators correlating some of the criteria used to
quantify the degree of satisfaction of the different objectives
included in the evaluation procedure. Thus, the different control
alternatives were characterized according to different points of
view i.e. environmental, economical, legal and technical. Finally,
the quantity of external carbon source, heating energy, risk of
microbiology-related separation problems and the percentage of
time that the plant is in violation with the BOD5 standards were

identied as the factors with the biggest differences in terms of

environmental, economic, technical and legal objectives.
To summarize, the authors emphasize the need to carry out this
type of analysis in order to obtain more information about how the
different evaluated alternatives may vary. A multi-objective evaluation of the WWTP alternatives includes an additional dimension
within the evaluation procedure increasing the chance of success of
the selected alternative. Thus, besides from keeping in mind the
potential environmental impact of each alternative it is possible to
identify possible problems at an early stage, such as the risk of
having separation problems or high operating costs improving the
whole decision making.
Acknowledgements
The authors fully acknowledge the nancial support of
the Spanish Ministry of Education and Science under the projects
MEC DPI2006-15707-C02-01 and CONSOLIDER-CSD2007-00055.
This work was also funded by Xunta de Galicia
(PGIDIT09MDS010262PR).
References
Basson, L., Petrie, J.G., 2007. An integrated approach for the consideration of
uncertainty in decision making supported by Life Cycle Assessment. Environ.
Modell. Softw 22 (2), 167176.
Bengtsson, M., Lundin, M., Molander, S., 1997. Life Cycle Assessment of Wastewater
Systems: Case Studies of Conventional Treatment, Urine Sorting and Liquid
Composting in Three Swedish Municipalities. Chalmers University of Technology, Gothemburg.
Comas, J., Rodrguez-Roda, I., Gernaey, K.V., Rosen, C., Jeppsson, U., Poch, M., 2008.
Risk assessment modelling of microbiology-related solids separation problems
in activated sludge systems. Environ. Modell. Softw 23 (1011), 12501261.
Copp, J.B., 2002. The COST Simulation Benchmark: Description and Simulator
Manual. Ofce for Ofcial Publications of the European Community,
Luxembourg.
Dennison, F.J., Azapagic, A., Clift, R., Colbourne, J.S., 1998. Assessing management
options for wastewater treatment works in the context of life cycle assessment.
Water Sci. Tech. 38 (11), 2330.
European Commission, 2000. Working documents on sludge 3rd draft. Available in:
http://ec.europa.eu/environment/waste/sludge/pdf/sludge_en.pdf.
Eurostat, 2008. Agricultural statistics. Main results 2006-2007. Available in:
http://epp.eurostat.ec.europa.eu/cache/ITY_OFFPUB/KS-ED-08-001/EN/KS-ED08-001-EN.PDF.
Flores-Alsina, X., Rodriguez-Roda, I., Sin, G., Gernaey, K.V., 2008. Multi-criteria
analysis of wastewater treatment plant control strategies under uncertainty.
Wat. Res. 42 (17), 44854497.
Gallego, A., Hospido, A., Moreira, M.T., Feijoo, G., 2008. Environmental performance
of wastewater treatment plants for small populations. Resour. Conserv. Recy 52
(6), 931940.
Gernaey, K.V., Van Loosdrecht, M.C.M., Henze, M., Lind, M., Jrgensen, S.B., 2004.
Activated sludge modelling and simulation: state of the art. Environ. Modell.
Softw 19 (9), 763783.

X. Flores-Alsina et al. / Journal of Environmental Management 91 (2010) 11931201

Gernaey, K.V., Rosen, C., Jeppsson, U., 2006a. WWTP dynamic disturbance modelling an essential module for long-term benchmarking development. Water
Sci. Tech. 53 (45), 225234.
Gernaey, K., Jeppsson, U., Batstone, D.J., Ingildsen, P., 2006b. Impact of reactive settler
models on simulated WWTP performance. Water Sci. Tech. 53 (1), 159167.
Guinee, J.B., Gorree, M., Heijungs, R., Huppes, G., Kleijn, R., de Koning, A., van
Oers, L., Weneger, A., Suh, S., Udo de Haes, H.A., de Bruijn, H., van Duin, R.,
Huijbregts, M., 2001. Life Cycle Assessment: An Operational Guide to the ISO
Standards. Ministry of Housing, Spatial Planning and Environment, Bilthoven,
the Netherlands.
Hair, J.Fr, Andersen, R.E., Tatham, R.L., Black, W.C., 1998. Multivariate Data Analysis,
fth ed. Prentice-Hall, London, UK.
Heijungs, R., Guinee, J.B., Huppes, G., Lankreijer, R.M., Udo de Haes, H.A.,
Weneger, A., Ansems, P., Eggels, P.G., van Duin, R., de Goede, H., 1992. Environmental Life Cycle Assessment of Products. Centre of Environmental Science
(CML), Leiden University, Leiden, the Netherlands.
Henze, M., Gujer, W., Mino, T., van Loosdrecht, M.C.M., 2000. Activated Sludge
Models ASM1, ASM2, ASM2d and ASM3. IWA Scientic and Technical Report
No. 9 IWA, London, UK.
Hischier, R., Classen, M., Lehmann, M., Scharnhorst, W., 2007. Life Cycle Inventories
of Electric and Electronic Equipment: Production, Use and Disposal. Swiss
Centre for Life Cycle Inventories, Dubendorf, Switzerland.
Hobson, J., 2000. CH4 and N2O Emissions from Waste Water Handling. Good
Practice Guidance and Uncertainty Management in National Greenhouse Gas
Inventories. Intergovernmental Panel on Climate Change. (IPCC) Publications,
Geneve, Switzerland.
Hospido, A., Moreira, M.T., Martn, M., Rigola, M., Feijoo, G., 2005. Environmental
evaluation of different treatment processes for sludge from urban wastewater
treatments: anaerobic digestion versus thermal processes. Int. J. Life Cycle Ass
10 (5), 336345.
Hospido, A., Moreira, M.T., Feijoo, G., 2008. A comparison of municipal wastewater
treatment plants for big centres of population in Galicia (Spain). Int. J. Life Cycle
Ass 13 (1), 5764.
Huijbregts, M.A.J., 1999. Life Cycle Impact Assessment of Acidifying and Eutrophying
Air Pollutants. Calculation of Equivalency Factors with RAINS-LCA. Faculty of
Environmental Science. University of Amsterdam, Amsterdam.
Huijbregts, M.A.J., Guinee, J.B., Reijnders, L., 2001. Priority assessment of toxic
substances in life cycle assessment, III: export of potential impact over time and
space. Chemosphere 44 (1), 5965.
International Energy Agency, 2009. World Energy Outlook 2008. Paris.
ISO 14040, 2006. Environmental management. Life cycle assessment. Principles and
framework (ISO 14040:2006). Geneva.
Jenkin, M.E., Hayman, G.D., 1999. Photochemical ozone creation potentials for
oxygenated volatile organic compounds: sensitivity to variations in kinetic and
mechanistic parameters. Atmos. Environ. 33 (8), 17751793.
Jenkins, D., Richard, M.G., Daigger, G.T., 2003. Manual on the Causes and Control of
Activated Sludge Bulking, Foaming and Other Solids Separation Problems, third
ed. IWA Publishing, London.
Jeppsson, U., Rosen, C., Alex, J., Copp, J., Gernaey, K.V., Pons, M.N., Vanrolleghem, P.A.,
2006. Towards a benchmark simulation model for plant-wide control strategy
performance evaluation of WWTPs. Water Sci. Tech. 53 (1), 287295.
Johnson, R.A., Wichern, D.A., 1992. Applied Multivariate Statistical Analysis. Prentice-Hall International, Englewood Cliffs, USA.
Karrman, E., Jonsson, H., 2001. Normalising impacts in an environmental systems
analysis of wastewater systems. Water Sci. Tech. 43(5) (29), 3300.
Lima, S.R., Park, J.M., 2009. Environmental impact minimization of a total wastewater treatment network system from a life cycle perspective. J. Environ.
Manage. 90 (3), 14541462.

1201

Liu, C.W., Lin, K.H., Kuo, Y.M., 2003. Application of factor analysis in the assessment
of groundwater quality in a blackfoot disease area in Taiwan. Sci. Tot. Environ.
313 (13), 7789.
Lundin, M., Bengtsson, M., Molander, S., 2000. Life Cycle Assessment of wastewater
systems: inuence of system boundaries and scale on calculated environmental
loads. Environ. Sci. Technol. 34 (1), 180186.
Machado, A.P., Urbano, L., Brito, A.G., Janknecht, P., Salas, J.J., Nogueira, R., 2007. Life
cycle assessment of wastewater treatment options for small and decentralized
communities. Water Sci. Tech. 56 (3), 1522.
Mels, A.R., van Nieuwenhuijzen, A.F., van der Graaf, H.H.J.M., Klapwijk, B., de
Koning, J., Rulkens, W.H., 1999. Sustainability criteria as a tool in the
development of new sewage treatment methods. Water Sci. Tech. 39 (5),
243255.
Mossier, A.R., 1993. In: van Amstel, A.R. (Ed.), Nitrous Oxide Emissions from Agricultural Soils. Bilthoven, Netherlands.
oz, I., Peral, J., Ayllocn, J.A., Malato, S., Martin, M.J., Perrot, J.Y., Vincent, M.,
Mun
Dome`nech, X., 2007. Life-cycle assessment of a coupled advanced oxidationbiological process for wastewater treatment: comparison with granular activated carbon adsorption. Environ. Eng. Sci. 24 (5), 638651.
Nemecek, T., Kagi, T., 2007. Life Cycle Inventories of Agricultural Production
Systems. Ecoinvent V. 2.0 Report No. 15. Agroscope FAL Reckenholz and FAT
Taenikon. Swiss Centre for Life Cycle Inventories, Dubendorf, Switzerland.
Padmanabhan, S., 2005. Sludge management shifts to reuse through innovative
thermal solutions. Water Wast. Int. 20 (4), 39.
Pasqualino, J.C., Meneses, M., Abella, M., Castells, F., 2009. LCA as a decision support
tool for the environmental improvement of the operation of a municipal
wastewater treatment plant. Environ. Sci. Technol. 43 (9), 33003307.
Renou, S., Thomas, J.S., Aoustin, E., Pons, M.N., 2008. Inuence of impact assessment
methods in wastewater treatment LCA. J. Clean. Prod. 16 (10), 10981105.
Roeleveld, P.J., Klapwijk, A., Eggels, P.G., Rulkens, W.H., van Starkenburg, W., 1997.
Sustainability of municipal wastewater treatment. Water Sci. Tech. 35 (10),
221228.
Siegrist, H., Brunner, I., Koch, G., Linh, C.P., Van Chiewu, L., 1999. Reduction of
biomass decay under anoxic and anaerobic conditions. Wat. Sci. Tech. 39 (1),
129137.
Solomon, S., Qin, D., Manning, M., Chen, Z., Marquis, M., Averyt, K.B., Tignor M.,
Miller H.L., 2007. Climate Change 2007: The Physical Science Basis. Contribution
of Working Group I to the Fourth Assessment. Report of the Intergovernmental
Panel on Climate Change. Cambridge, UK and New York, USA.
Tan, R.R., Culaba, A.B., Purvis, M.R.I., 2004. POLCAGE 1.0-A possibilistic life-cycle
assessment model for evaluating alternative transportation fuels. Environ.
Modell. Softw 19 (10), 907918.
Tillman, A.M., Ekvall, T., 1994. Choice of system boundaries in life cycle assessment.
J. Clean. Prod. 2 (1), 2129.
Tillman, A.M., Svingby, M., Lundstrom, H., 1998. Life Cycle Assessment of municipal
waste water systems. Int. J. Life Cycle Ass 3 (3), 145157.
Vanrolleghem, P., Gillot, S., 2002. Robustness and economic measures as control
benchmark performance criteria. Wat. Sci. Tech. 45 (45), 117126.
Vidal, N., Poch, M., Martic, E., Rodricguez-Roda, I., 2002. Evaluation of the environmental implications to include structural changes in a wastewater treatment plant. J. Chem. Technol. Biot 77 (11), 12061211.
Vrecko, D., Gernaey, K.V., Rosen, C., Jeppsson, U., 2006. Benchmark simulation
model No 2 in matlab-simulink: towards plant-wide WWTP control strategy
evaluation. Wat. Sci. Tech. 54 (8), 6572.
Wanner, J., 1994. Activated Sludge Bulking and Foaming Control. Basel. Technomic
Publishing Company, Inc, New York.
World Meteorological Organisation, 1999. Scientic assessment of ozone depletion:
1998. Report no 44, Geneva.