1993 © IWA Publishing 2011 Water Science & Technology | 64.

10 | 2011

Multicriteria decision-aid method to evaluate

the performance of stormwater infiltration
systems over the time
P. Moura, S. Barraud, M. B. Baptista and F. Malard

ABSTRACT
P. Moura (corresponding author)
Nowadays, stormwater infiltration systems are frequently used because of their ability to reduce
M. B. Baptista
flows and volumes in downstream sewers, decrease overflows in surface waters and make it Departamento de Engenharia Hidráulica
e Recursos Hídricos da UFMG,
possible to recharge groundwater. Moreover, they come in various forms with different uses. Despite Escola de Engenharia - Av. Antônio Carlos,
6627- 31 270-901 Belo Horizonte,
these advantages the long term sustainability of these systems is questionable and their real Brazil
E-mail: [email protected]
performances have to be assessed taking into account various and sometimes conflicting aspects.
S. Barraud
To address this problem a decision support system is proposed. It is based on a multicriteria method
Université de Lyon,
built to help managers to evaluate the performance of an existing infiltration system at different F-69003, Lyon, France
Université Lyon 1,
stages of its lifespan and identify whether it performs correctly or not, according to environmental, LGCIE, F-69622, Villeurbanne, France
INSA-Lyon,
socio-economic, technical and sanitary aspects. The paper presents successively: the performance LGCIE, F-69621, Villeurbanne, France
34 avenue des Arts, Bât.J.-C.-A. Coulomb,
indicators and the way they were built, the multicriteria method to identify if the system works
69621 Villeurbanne CEDEX, France
properly and a case study.
F. Malard
Key words | indicators, infiltration systems, multicriteria method, stormwater Université de Lyon,
F-69003, Lyon, France
Université Lyon 1,
Lyon, F-69622, France
CNRS, UMR 5023,
Laboratoire d’écologie des hydrosystèmes naturels
et anthropisés - LEHNA
Villeurbanne, F-69622, France

INTRODUCTION

Infiltration systems are frequently used as an option to evolution over time with clogging phenomena), their effec-
manage urban storm drainage. These techniques present tiveness in environmental and public health protection,
important advantages: the groundwater pollution risk, the management of sedi-
ment trapped, social acceptance, costs, etc.
• they reduce flows and volumes in downstream sewers or
For these reasons, the evaluation of infiltration tech-
surface waters;
niques is indispensable and the research of a compromise
• they limit wash-off phenomena in urban areas that lead to
between their multiple facets is needed.
a reduction of the pollutant load;
To address this problem a decision support system is
• they favor quantitative groundwater recharge;
proposed. It is based on a multicriteria method built to
• they present an important treatment potential;
help designers and managers of such systems to: (i) evaluate
• they allow urban development in areas far from surface
and compare the performance of alternatives or different pro-
outlets;
jects at a design stage (choice of a good project among a set of
• they come in various forms with different uses (e.g. flood-
alternatives) (Moura et al. a), (ii) evaluate the quality of
able sport grounds).
an existing infiltration system and the strategies to be
Despite these advantages, their long term sustainability is applied to improve their performance (choice of maintenance
not certain. Numerous questions rise from their utilization. strategies, selection of technical/social/environmental
It involves flooding protection efficiency (considering its improvements, choice of rehabilitation solutions, etc.).

doi: 10.2166/wst.2011.154
 1994 P. Moura et al. | Multicriteria decision-aid method Water Science & Technology | 64.10 | 2011

The present article only deals with the evaluation if an Table 1 | Indicators quality and way of testing

existing infiltration system still works properly according

Quality required for each PI Way of testing
to technical, environmental and socio-economic aspects.
Relevance (Does it correspond This quality is checked by the
to a relevant aspect to multidisciplinary working
verify?) group which involves different
METHODOLOGY standpoints.
Accessibility (Is the indicator This quality is set a priori and
The method integrates the evaluation of an existing system easy to calculate and are the then verified by applying the
according to different indicators and the way to identify if data for its calculation PI to different contexts and
a system has to be examined more particularly because of available at an acceptable case studies given by the
cost?) professionals of the group.
its suspected low performance at a particular stage of its
Objectivity (Is it ambiguous? The results of the indicator
lifespan.
Can it be interpreted in the evaluation coming from case
The development of the method needs: (i) the construc- same way by different studies are submitted to
tion of performance indicators (PI), (ii) the test of the quality potential appraisers?) different people who must
of each indicator, (iii) the test of the quality of the global set interpret them in the same
of indicators, (iv) the development of a multicriteria method way.

to identify if the system works properly. Robustness (Does it give stable The source of uncertainties was
results according to the evaluated on real case studies.
variation of uncertain Taking into account the range
Construction of indicators parameters?) of uncertainties, we
considered the indicator
In order to evaluate existing systems, a set of PIs integrating robust when the results tended
towards same tendencies.
technical, economical, environmental, social and sanitary
Sensitivity (Does it The indicator was found
aspects was developed. The work was carried out with the
discriminate different sensitive when it was able to
help of a multidisciplinary group and the support of the strategies properly?) identify differences between
research federation OTHU (http://www.othu.org) built to systems known to be different.
develop a Field Observatory in Urban Hydrology situated Fidelity (Can the indicator be Different appraisers evaluate an
in Lyon, France. The working group was composed of estimated with a constant indicator on real case studies,
researchers from different fields (hydrology, biology, chem- bias? Can it be evaluated in the tendency of the evaluation
the same way by different must be the same.
istry, environment, groundwater domain, soil science,
appraisers?)
social science) and professionals from public and private
Univocity (aptitude to give a This quality was checked at the
companies (designers, people in charge of maintenance or univocal evaluation) moment of the construction
control of existing systems). The Field Observatory OTHU by discussions within the
was used to get reliable information and knowledge from working group.
intensive on-site measurements. This observatory was quite
useful to construct indicators and to test the assessment
models of the PIs. involving numerous features, the set of indicators used
is supposed to form a consistent family, meeting conditions
Test of the quality of each PI and test of the whole set of of exhaustiveness, cohesiveness and non-redundancy
indicators (Vincke ).
This has been verified in a very empirical way.
To test the quality of each indicator, criteria developed by Cohesiveness was checked by simple analyses. Qualitative
Labouze & Labouze () were used. Table 1 shows the correlations between indicators were explored to
different criteria and the way each one was tested. When detect potential redundancy. For exhaustiveness, a first
an indicator did not meet the requirements, it was re-defined list of PIs were built and submitted to each member of
until fulfilling an acceptable level of quality. All the tests can the working group. When an aspect was missing, a per-
be found in Moura (). formance was added and so on until stabilization of the
For the evaluation of the quality of the whole set of list. This process does not insure the exhaustiveness of
indicators, three aspects were considered. In multicriteria the set, the set being as exhaustive as the members’
support system and especially in partial aggregation method preoccupations.
 1995 P. Moura et al. | Multicriteria decision-aid method Water Science & Technology | 64.10 | 2011

Choice and principle of the multicriteria method a S b1 (σða; b1 Þ) calculated by using an overall concordance
index and discordance indices, is built and compared with a
After defining the set of PIs, a method had to be determined ‘cutting level’ (λ) in order to determine the global preference
to sort the systems that work correctly over time and those situation between a and b1.
that do not perform well. To assign an existing system to Situation 1: If σða; b1 Þ ≥ λ, σða; b1 Þ ≥ λ ) ðaSb1 ; b1 SaÞ ) a
one of the two categories: ‘performs correctly’ or ‘does not and b1 are indifferent
perform correctly’, the ELECTRE TRI method developed
Situation 2: If σða; b1 Þ ≥ λ, σða; b1 Þ < λ ) ðaSb1 ; not
by Yu () was chosen for three reasons.
b1 SaÞ ) a is preferred to b1
(i) It is a multicriteria method based on partial aggregation Situation 3: If σða; b1 Þ < λ, σða; b1 Þ ≥ λ ) ðnot aSb1 ;
where each criterion is expressed in its own value system and b1 SaÞ ) b1 is preferred to a
scale. Therefore, a large number of different types of indicators
Situation 4: If σða; b1 Þ < λ, σða; b1 Þ < λ ) ðnot aSb1 ;
can be used. (ii) The method allows the user to account for not b1 SaÞ ) a and b1 are incomparable
uncertainties in the evaluation of the indicators. (iii) It is a sort-
ing method, fits to our decision problem which aims at Finally, an optimistic and pessimistic procedures are
assigning actions or strategies (in our case one infiltration applied to assign the system a to one category. In the opti-
system) to predefined ordered categories (in our case: two cat- mistic procedure a is compared to b1, if b1 is preferred to
egories ‘performs correctly’ and ‘does not perform correctly’). a, a is assigned to the category ‘does not perform correctly’.
The principle of ELECTRE TRI is simple. First, each cri- In the pessimistic procedure a is compared to b1, if a is pre-
terion (PI) must be weighted according to its relative ferred to b1, a is assigned to the category ‘performs
importance. Secondly, the limit between two categories is for- correctly’. The results of these two assignments differ
malized by a reference profile (in our case: just one profile (let when a and b1 are incomparable. In this case, the optimistic
us call it ‘b1’) discriminating good and bad performances). procedure would lead to assign a to the category ‘performs
This reference profile is in fact a fictitious user-defined correctly’ whereas the pessimistic procedure would assign
alternative providing references values on each criterion. a to the category ‘does not perform correctly’.
The assignment of an infiltration system (let us call it ‘a’) The reader will find more details in Yu () or
to a category results from the comparison of a and the refer- Rogers et al. (). A sensitivity analysis of the different
ence profile b1. parameters must be done to test their influence on the
For the comparison, ELECTRE TRI validates or invali- final assignment.
dates the statement (a S b1) which means ‘a is at least as
good as the profile b1’.
For that purpose, three thresholds per criterion are
PERFORMANCE INDICATORS
defined (pseudo-criterion): (i) an indifference threshold
Flood protection
under which the decision maker shows clear indifference
(uncertainties of indicator evaluation can be used to fix
Two indicators are used to qualify this aspect.
this threshold); (ii) a preference threshold above which the
decision maker is certain of a strict preference, in between,
there will be situations of weak preference indicating Flooding frequency indicator
decision maker’s hesitation between indifference and strict
RP flood
preference; (iii) a veto threshold (not compulsory) which is ISHYD1 ¼ ðDimensionless; the lower the better,
RPdesign
the maximal difference which turns out to be unacceptable
between a and b1 on a given criterion regardless the per- range : ½0; þ∞½Þ
formances on other indicators.
The validation of the assertion (a S b1) is made according On a determined period, RP flood is the observed return
to two conditions: (i) a concordance condition: the system a is period of flood (in years) caused by events whose recurrence
at least as good as b1 if a sufficient number of criteria (PIs) are interval is lower than the recurrence interval used for sizing
in favour of the statement; (ii) a non-discordance condition: (RPdesign ).
when the concordance condition holds, none of the criteria The determined period of time has to be specifically
should oppose to the statement in a too strong way. Then, defined for each situation (e.g. from the implementation of
an index called the degree of credibility of the statement the system to now…).
 1996 P. Moura et al. | Multicriteria decision-aid method Water Science & Technology | 64.10 | 2011

Global hydraulic performance measuring the potential raining, infiltrated water normally induces a decrease in
for clogging specific conductance and an increase in oxygen concen-
tration of groundwater. If these ‘normal conditions’ are not
For extensive systems, like infiltration basins: observed a potential dysfunction may occur and further inves-
tigations have to be done. The indicator must just specify
ISHYD2 ¼ Ma xðRi Þ ðin h; the lower the better; whether the system works normally or not. For that purpose,
i
monitoring of dissolved oxygen concentration and specific
range : 0; þ∞½Þ
conductance has to be carried out periodically in ground-
water downstream the system during rain events. The
Ri : global hydraulic resistance of a sub-system i compos-
indicator is then defined as follows.
ing the whole infiltration system according to Bouwer’s
If the system works as expected (i.e. in ‘normal con-
model as described in Le Coustumer & Barraud ().
ditions’ as defined before) during rain periods ISGRW ¼ 0
An infiltration system whose hydraulic resistance is
higher than 24 h is supposed to work incorrectly (Gautier
else ISGRW ¼ 1 ðDimensionless, the higher the better,
et al. ; Dechesne et al. ).
range : f0; 1gÞ
For source control systems, like pits, trenches or swales:

For example in Figure 1: during the period, most of the

ISHYD2 ¼ Ma xðKsi Þ ðin m/s; the higher the better;
i events induces an increase in the specific conductance and a
range : 0; þ∞½Þ decrease in dissolved oxygen, the basin does not work
normally, ISGRW ¼ 1.
Ksi : measured hydraulic conductivity of a sub-system i
According to the literature, a system with an hydraulic con- Low soil pollution but high efficiency in pollution
ductivity (measured at different points) lower than 106 m/s is retention
supposed to work incorrectly (CIRIA ; Ellis ).
This performance is based on two indicators: pre-
For mixed areas having extensive and source control treatment efficiency (with regard to TSS trapped) and soil
systems contamination.

ISHYD2 ¼ Ma xðIndi Þ ðDimensionless; the higher the better,

i
Indicator of pre-treatment efficiency
range : f0; 1gÞ
MpTSS
ISPTE ¼ ðDimensionless; the higher the better,
Indi : Partial indicator specifying if clogging of a sub- MapTSS
system i is attested range : ½0; 1Þ
If clogging is not attested Indi ¼ 1, else Indi ¼ 0. For
extensive systems clogging will be attested if the hydraulic
resistance is higher than 24 h. For the source control sys-
tems clogging is attested if the hydraulic conductivity is
lower than 106 m/s.

Low degradation of groundwater quality

For extensive systems, this indicator has been built accord-

ing to the work conducted by F. Malard in Perrodin et al.
() and lies in the following assumptions.
Runoff water presents generally low specific conductance
(<100 μS/cm) and is saturated in dissolved oxygen. Despite of
regional physical-chemical variation, the groundwater has
Figure 1 | Time series of specific conductance and dissolved oxygen concentration (1 h
usually higher level of specific conductance than runoff time step) in groundwater under ZAC du Chêne infiltration basin – Malard in
water and is under saturated in dissolved oxygen. When Perrodin et al. (2005).
 1997 P. Moura et al. | Multicriteria decision-aid method Water Science & Technology | 64.10 | 2011

MpTSS : Mass of TSS trapped by the pre-treatment system #dysi : Number of malfunctions observed in the system
in a fixed period; MapTSS : mass of TSS brought to the pre- i according to a pre-defined check-list established in the
treatment system in the fixed period. working group (Moura ), #dystotal : Total number of
dysfunctions on such system according to a pre-defined
Soil contamination indicator check-list.

After numerous definitions (Moura et al. b), the indi- Protection of users and workers health and safety
cator was determined by two sub-indicators: ISSOIL1 : depth
where pollution becomes low or nil. (in m, the lower the Indicator of potential risk due to soil contamination for
better, range: [0, þ∞]) users or workers

ISSOIL2 : percentage of highly polluted points ðin %; the The definition is given by:
lower the better; range : ½0; 100Þ: If K4 ≠ 0% ISSAN1 ¼ 0 else ISSAN1 ¼ K3 =100

with K4 ¼ %case ½CmesSk ≥ S5k  and K3 ¼ %case

To judge if the pollution is high or low a K1 index is
½CmesSk ≤ S0;2k  (Dimensionless, the higher the better;
used: K1 ¼ #case [Cmesk ≤ Cinitk ]=#totpol
range: [0, 1])
#totpol : Total number of pollutants considered; Cmesk :
maximum concentration of the pollutant k (mg/kg) %case½CmesSk ≥ S5k  : Percentage of cases where
measured; Cinitk : initial pollutant concentration in ½CmesSk ≥ S5k ;
the soil for pollutant k (before construction of the
system) (mg/kg); #case½Cmesk ≤ Cinitk : number of CmesSk : Measured concentration of pollutant k in the
pollutants for which the maximum measured concentration 30 first centimeters of the topsoil; when more than one
is lower than the initial concentration or in the same sample is done, the higher concentration is considered,
range. (mg/kg of dry matter);
A low pollution is characterized by K1 close to 1 S0;2k : Concentration threshold of the pollutant k for
(K1 ∈ ½1  α; 1, α being an exigency threshold) whereas K1 which the soil is considered as compatible with the use
close to 0 indicates a high level of pollution. (mg/kg DM); S5k : concentration threshold of the pollutant
k for which the soil is considered as non compatible with
Aptitude to be well and easily maintained the use (mg/kg DM).
The thresholds S0;2k and S5k were calculated on the basis
A first indicator specifies if the system is regularly and prop- of French standards for the management of polluted sites
erly maintained by the service in charge of operation: and soil (MEDD ).

IEMAINT ¼ #total  #UNDN ðDimensionless; the lower Indicator of potential risk due to air pollution for users
the better; range : ½0; #total Þ and local residents
 
CmesAk
#total : Total number of maintenance tasks to be done in ISSAN2 ¼ %case > α ðin%; the lower the better,
Clim Ak
the system; #UNDN : Number of maintenance tasks undone
range : ½0; 100Þ
in the system
To number the tasks to be done, the evaluator uses a %case½CmesAk =Clim Ak > α :
check-list built for each kind of system (basin, trench, …). Percentage of cases when½CmesAk =Clim Ak > α;
This list was established with the help of the working
group and can be found in (Moura ). CmesAk : Measured air concentration of the pollutant k
A second indicator is based on the dysfunctions (μg/m3); Clim Ak : limit air concentration value of
observed in situ: the pollutant k according to WHO (WHO ), (μg/m3);
α: ratio of the real exposure duration in hours to
ISMAIN ¼ #dys i ðDimensionless; the lower the better; the eight hours used to determine the limit concentration
range : ½0; #dystotal Þ value.
 1998 P. Moura et al. | Multicriteria decision-aid method Water Science & Technology | 64.10 | 2011

Indicator of potential risk due to pollution of air and Good social acceptance
soil particles (for workers)
As it is very difficult to estimate this aspect, the proposed
Concerning the staff we consider the air and soil exposure at indicator only considers the existence of complaints from
the same time. The indicator uses the exposure thresholds the local residents concerning the system since its implan-
from the INRS which is the French institute competent in tation or its more recent restoration.
the area of health and safety at work (INRS ). ISSA ¼ 0 if there are complaints about the system since
its implantation or its more recent rehabilitation; otherwise
 
CmesTk ISSA ¼ 1 (Dimensionless, the higher the better, range: f0; 1g)
If > 1 ) ISSAN3 ¼ 100 else
Clim Tk
 
CmesTk
ISSAN3 ¼ %case > β ð%; the lower the better
Clim Tk CASE STUDY
range : ½0; 100Þ
To evaluate the methodology, case studies were selected.
One of them, the Django Reinhardt basin system, is pre-
%case½CmesTk =Clim Tk > β :
sented here. It concerns an end of pipe system located in
Percentage of cases when½CmesTk =Clim Tk > β;
an urban industrial area, located in Lyon, France. It is com-
posed of two compartments (one retention basin connected
CmesTk : Measured concentration of the pollutant k, in the to one infiltration basin). It drains rainwater from a catch-
air and particles of soil (mg/m3); Clim Tk : Threshold of con- ment of 185 ha with 70% imperviousness. The system is
centration of the pollutant k according to French INRS old (30 years) but it has been rehabilitated in 2002 and com-
standards, (mg/m3); β: Ratio of the real exposure duration pletely cleaned in 2004. This system is not open to the
in hours to the eight hours used to determine the limit public.
concentration value.
Performances and parameters of the methodology
Waste production

The performance indicators and the main parameters of the

The indicator considers the part of the sediments cleaned
case study are given in Table 2. The weights presented here
out from the systems that can be reused.
are the mean of three decision makers’ opinion obtained by
interviews. The indifference thresholds are based on indi-
Msreus
ISWP ¼ ðDimensionless; the higher the better, cator uncertainties. We can notice that no veto thresholds
Msextr
were used. Some indicators were not calculated because of
range : ½0; 1Þ
the lack of available data, like air quality and maintenance
costs.
Msreus : Sediment mass which can be reused, on a Figure 2 presents a graphical comparison between
defined period; Msextr : Sediment mass cleaned out of the Django Reinhardt performance and the Reference Profile.
system on the same period.

Low maintenance costs

RESULTS AND DISCUSSION
The indicator is based on the variation in the maintenance
First, we can notice the lack of information to evaluate some
costs over time. An abnormal increase or decrease in cost
PIs. It is the case for sanitary indicators related to air pol-
must draw managers’ attention.
lution. Actually, no measurement of air pollution is done
on such systems so that we can’t control the potential tox-
ISCOST ¼ CMAc  CM Pr ðDimensionless; the lower the
icity for the workers in charge of maintenance or for
better, range : ½∞; þ∞Þ people living around the system. However, dust coming
from the system is often observed and should be controlled
CMAc : Maintenance cost in year j; CMPr : mean of main- at least once. More surprisingly, it is also the case of the indi-
tenance costs in previous years. cator of maintenance cost. The municipality does not know
 1999 P. Moura et al. | Multicriteria decision-aid method Water Science & Technology | 64.10 | 2011

Table 2 | PI, PI value, data used for the PI calculation, weight, indifference, preference thresholds and value taken by the reference profile

PI Indif. Pref. Ref.

PI value Data used Weight threshold threshold profile

ISHYD1 0 Observation 11.1 0 0 1

ISHYD2 9h Hydraulic resistance calibrated according to inflow, water depth, & 10.3 6 12 24 h
temperature time series
ISGRW 1 Groundwater monitoring (dissolved oxygen concentrations & specific 14.7 0 0 1
conductance)
ISPTE 0.29 TSS load derived from continuous turbidity measurements 5.3 0.03 0.03 0.5
ISSOIL1 1m Soil sampling at different depths and chemical analysis 5.3 0.2 0.5 1m
ISSOIL2 100% 4.1 0 0 33%
ISMAINT 3 In situ observation 13.3 1 2 1
ISSAN1 67% Soil sampling and chemical analysis 6.4 0 0 0%
ISSAN2 – – – – – –
ISSAN3 – – – – – –
ISWP 0 None (No sediments were reused in the study) 16.1 0 0 1
ISCOST – – – – – –
ISSA 1 Inventory of complaints about the system from local residents 13.4 0 0 1

Secondly, regarding the performance of Django Rein-

hardt, incomparability between its performance and the
Reference Profile was found. The basin performance is out-
ranked by the Reference Profile in the pessimistic procedure
and outranks the Reference Profile in the optimistic procedure.
Therefore, the basin presents performances close to the mini-
mal thresholds. The managers have to study more precisely
the origins of this assignment. We can assert that this result is
correct; the basin has actually insufficient performance par-
ticularly in terms of maintenance and waste reuse.
Sensitivity and robustness analysis have been carried out
to test the influence of the variation of the different parameters
within a coherent decision strategy. The methodology was
found robust (Moura ).

Figure 2 | Relative position of indicator values (dashed line) compared to the Reference
values delimiting an area of good (in grey) and worse performance (in white). CONCLUSIONS
Each Y-axis of the different indicators is oriented from the worst to the best
evaluation.
The present article showed the performance indicators to be
how much the regular maintenance of a particular system used in the evaluation of infiltration systems, during its life-
effectively costs. The cost of maintenance is totaled up so span in order to quantify whether it performs correctly or
that a global maintenance cost for wastewater services is not, according to a wide range of criteria. The whole set of
known but not the cost of one system. In order to improve indicators presented satisfactory qualities.
this aspect, the construction of a cost database was decided. Case studies were carried out; one is shown in the
The first conclusion that can be drawn is that, even if paper. The method turns out to be efficient and adapted
all the indicators can’t be evaluated, the method points out to test the quality of an existing system. It points out the differ-
the need of improvements in terms of data acquisition ent aspects that have to be improved and indicates the
for the municipality. necessary shift in the design of future systems. It
 2000 P. Moura et al. | Multicriteria decision-aid method Water Science & Technology | 64.10 | 2011

also highlights the lack of information about some perform- MEDD – Ministère de l’Ecologie et du Développement Durable
ance evaluations which may draw managers’ attention and  La Démarche de l’Interprétation de l’Etat des milieux
(An Approach to Evaluate Ecosystems Conditions). MEDD,
give tracks of improvement of their practice and organization.
Paris. Available from: http://www.sites-pollues.ecologie.
gouv.fr (accessed 10 December 2007).
Moura, P.  Méthode d’évaluation des performances des
ACKNOWLEDGEMENTS systèmes d’infiltration des eaux de ruissellement en milieu
urbain (Performance Evaluation Method of Urban Runoff
Infiltration Systems). PhD Thesis, Institut National des
The authors thank the Région Rhône Alpes for PhD scholar- Sciences Appliquées de Lyon. Available from: http://these.
ship, the Greater Lyon, OTHU (Field Observatory in Urban insa-lyon.fr/publication/2008isal0048/these.pdf.
Hydrology), ANR-Ecopluies and FAPEMIG for scientific Moura, P. M., Barraud, S. & Baptista, M. B. a Multicriteria
support. procedure for the design and the management of
infiltration systems. Water Science & Technology 55 (4),
145–153.
Moura, P., Barraud, S. & Varnier, J.-C. b Comparison between
REFERENCES different approaches for the definition of soil contamination
indicators of stormwater infiltration systems. In Preprint of
CIRIA  Infiltration Drainage: Manual Of Good Practice. 6th International Conference on Sustainable Techniques and
Report 156, Construction Industry Research and Information Strategies in Urban Water Management, NOVATECH 2007,
Association, London, 107 pp. June 25–28, 2007, Lyon, France, pp. 843–850.
Dechesne, M., Barraud, S. & Bardin, J.-P.  Indicators for Perrodin, Y., Delolme, C., Winiarski, T., Bedell, J.-P., Barraud, S.,
hydraulic and pollution retention assessment of stormwater Bardin, J.-P., Le Coustumer, S., Gibert, J., Malard, F., Mermillod
infiltration basins. Journal of Environmental Management 71, Blondin, F., Gourdon, R., Desjardin, V., Brelot, E. & Bacot, L.
371–380.  MGD Infiltration : Maîtrise et gestion durable des ouvrages
Ellis, J. B.  Infiltration systems: a sustainable source-control d’infiltration des eaux pluviales en milieu urbain (Control and
option for urban stormwater quality management. Water and Management of Sustainable Stormwater Infiltration Systems in
Environment Journal 14 (1), 27–34. Urban Areas). Final Repport Programme Réseau Génie Civil &
Gautier, A., Barraud, S. & Bardin, J.-P.  An approach to Urbain, Lyon, France.
the characterisation and modelling of clogging in storm water Rogers, M., Bruen, M. & Maystre, L.-Y.  ELECTRE and
infiltration facilities. In Preprint of Eighth International Decision Support: Methods and Applications in Engineering
Conference on Urban Storm Drainage, August 30–September and Infrastructure Investment. Kluwer Academic Publishing,
3, 1999, Sydney, Australia, Vol. 2, pp. 1007–1015. Dordrecht, The Netherlands, 208 pp.
INRS  Valeurs limites d’exposition professionnelle aux agents Vincke, P.  Muticriteria Decision-Aid. John Wiley & Sons,
chimiques en France (Limit Values for Occupational Chichester, England, 154 pp.
Exposure to Chemicals in France). Repport, Institut National WHO  Air Quality Guidelines: Particulate Matter, Ozone,
de Recherche et Sécurité. Available from: www.inrs.fr Nitrogen Dioxide and Sulfur Dioxide. WHO Regional Office
(accessed 15 December 2007). for Europe, Copenhagen, Denmark.
Labouze, E. & Labouze, R.  La comptabilité de Yu, W.  Aide multicritère à la décision dans le
l’Environnement (Environmental accounting). Revue cadre de la problématique du tri : Concepts,
Française de Comptabilité 272, 92 pp. Méthodes et Application (Multicriteria Decision Aid
Le Coustumer, S. & Barraud, S.  Long-term hydraulic and in Sorting Problems : Concepts, Methods and
pollution retention performance of infiltration systems. Water Application). PhD Thesis, Université de Paris-Dauphine,
Science & Technology 55 (4), 235–243. Paris, France.

First received 1 July 2010; accepted in revised form 3 September 2010