Multicriteria procedure for the design and the

Water Science & Technology Vol 55 No 4 pp 145–153 Q IWA Publishing 2007

management of infiltration systems
P. Moura*, S. Barraud* and M. Baptista**
*URGC - INSA Lyon - Bâtiment J.C.A. Coulomb, 34 avenue des Arts - 69621 Villeurbanne Cedex – France
(E-mail: [email protected]; [email protected])
**Departamento de Engenharia Hidráulica e Recursos Hı́dricos - Escola de Engenharia da UFMG Av.
Contorno, 842-80 Andar - 30110.060 - Belo Horizonte - MG – Brazil (E-mail: [email protected])

Abstract Infiltration systems are frequently used as an option to manage urban storm drainage. By
reducing flows and volumes in downstream sewers or in surface waters, they decrease the overflows
and make it possible to recharge groundwater. They come in various forms with different uses; therefore,
their performance is diverse and integrates multiple aspects. Consequently, a multicriteria approach was
developed in order to quantify the performance of these systems and to help in decision making
problems. For that purpose, a list of performance indicators integrating technical, economical,
environmental and social aspects was developed. The performances were defined with the help of a
working group composed of engineers from different technical or strategic departments from Greater
Lyon and researchers from different fields. The paper presents the last version of the performance
indicators tested according to a set of quality requirements: availability of data, relevance, fidelity,
precision, sensitivity/robustness. This critical review of the set of indicators has led us to redefine a
certain number of indicators, identify numerous biases and allow putting forward general instructions for
criterion or indicator construction. The last phase is to propose multicriteria decision aid methods; a
procedure using ELECTRE methods should be used.
Keywords Infiltration systems; multicriteria analysis; performance indicators

Introduction
Infiltration systems are frequently used as an option to manage urban storm drainage. By
reducing flows and volumes in downstream sewers or in surface waters, they decrease the
overflows and make it possible to recharge groundwater. Finally, they come in various
forms with different uses such as floodable sport grounds or porous roads. Despite these
advantages, long-term sustainability of infiltration systems is not certain (Dechesne et al.,
2004). The performance of such systems is diverse and does not only integrate technical
aspects but also environmental, economic and social ones, regarding sustainable develop-
ment. Consequently, a multicriteria approach was developed in order to quantify the per-
formance of these infiltration systems and to help in decision making problems. More
particularly the procedure presented in the paper aims at evaluating and comparing: i)
several alternatives or different projects at the design stage (choice of a good project
among a set of alternatives), ii) strategies to be applied for an existing infiltration
system (choice of maintenance strategies, selection of technical/social/environmental
improvements, choice of rehabilitation solutions).
The paper presents the last version of the performance indicators tested according to a
set of quality requirements: availability of data, relevance, fidelity, precision, sensitivi-
ty/robustness.

doi: 10.2166/wst.2007.104 145

 General considerations about the construction of indicators
The construction of a set of indicators has to be initiated by defining objectives: what are
the indicators supposed to represent? Then criteria have to be identified, defined precisely
with their evaluation mode and at last, they must be verified in terms of consistency.
The sustainable development indicators can have different functions: they can support
the definition of sustainable development objectives, they can aid in the program and pro-
ject management or can be useful in public communication and participation (Pastille,
P. Moura et al.

2002). The sets of indicators have therefore to be elaborated according to their uses. An
indicator can be useful to one objective but not to another. Concerning the existent litera-
ture recommendations about the construction of sets of indicators, most of them present
two forms: a priori check-list of quality criteria or a posteriori analysis by means of a list
of common errors made when the set of indicators is applied to real problems (Kastner,
2003).
According to Labouze and Labouze (1995), Personne (1998), Bellagio (1996), Pastille
(2002), Perrin (1998) and Riley (2001), the following questions must be considered if
one wants to evaluate the quality of a set of indicators: (1) Are they easy to calculate and
the data available? (2) Do they meet the requirements? (3) Aren’t they ambiguous? (4)
Are their evaluations acceptable regarding uncertainties? (5) Do they present good sensi-
tivity and high robustness? (6) Are the spatial and temporal scales relevant? (7) Isn’t the
number of indicators too large? In certain works, it is advisable to use quantitative indi-
cators as much as possible, but in certain cases the quantitative indicators present large
uncertainties and the use of qualitative indicators can be a better solution.

The set of indicators

A first set of indicators (PIs) regarding 17 performances of infiltration systems (Barraud
et al., 2001) was proposed. The set was supposed to quantify infiltration system perform-
ances and compare various strategies. However, problems appeared while using them.
This analysis led us to consider PIs at the design stage (infiltration devices are just
planned but do not exist) and at operation phase (technical systems are existent, they can
be observed and monitored). In this paper the last version of the set of PIs is presented.
The performances were defined with the help of a working group composed of engin-
eers from different technical and strategic departments from the Greater Lyon, research-
ers from different fields (economy and social science, soil science, hydrology,
hydrobiology, chemistry). Considering each performance, one or more complementary
indicators or several options of the same indicator were proposed in accordance with
different levels of data availability. No evaluation method was dismissed. The use of
expertise, modelling, on site measurements… were thus integrated in the evaluation pro-
cess when necessary. On the contrary, for some performances, the lack of data or infor-
mation was so important that no indicator was explicitly proposed.
The performances with the associated PIs are presented in Table 1. The PIs’ definition
and their assessment method are also presented. This critical review of the previous set
of indicators has led us to redefine a certain number of indicators, identify numerous
biases and allow putting forward of general instructions for criterion or indicator con-
struction (Barraud et al., 2004).

Performance P1 – Low flooding (frequency and quantity)

As a part of a global drainage system, the hydraulic performance of an infiltration facility
has to be related to direct (e.g. infiltration basin flooding) or indirect risks (e.g. sewer
flooding due to the dysfunction of an infiltration device). At the design stage, two indi-
146 cators were proposed. These indicators met the principal quality requirements defined in
Table 1 Set of performances

P1- Low flooding (frequency and quantity) P10- Aptitude to be secure and
safe for users
P2- Low water resources pollution P11- Aptitude to be secure and
safe for the staff
P3- Positive contribution to groundwater recharge P12- Low cost systems
P4- Little use of raw material P13- Ability to preserve / encourage
economic activity

P. Moura et al.
P5- Low air pollution (odors, pollution…) P14- High system adaptability
P6- Low soil pollution P15- Ability to provide other functions
P7- Efficiency in terms of pollution P16- Good social acceptance
retention
P8- Ability to be managed and P17- Low waste production
maintained easily
P9- Low disturbance of other technical
systems
Performance having an indicator proposition performance without indicator proposition

the second paragraph.

IP121 ¼ Fdim ð1Þ

Fdim: design flooding frequency and

IP122 ¼ MaxðVdebi Þk ð2Þ

Vdebi: flooding volume for a storm event i (m3), in a reference period, obtained by model-
ing, k: vulnerability coefficient depending on the flooded area (ranging from 0 to 1, 1
being the more vulnerable)
At the operation stage three indicators are defined:

IP121 ¼ Fdeb =Fdim ð3Þ

X
IP1221 ¼ Vdebi ð4Þ
i

and
If there is flooding,

IP1222 ¼ Tdeb ð5Þ

If there isn’t flooding,

IP1222 ¼ %T ¼ NT , 24h=NT ð6Þ

and

IP123 ¼ Dslp ð7Þ

Fdeb: flooding frequency evaluated by modeling, Fdim: design flooding frequency, Vdebi:
flooding volume for a storm event i (m3), in the lifetime or in a reference period, Tdeb:
Flood duration, in the life time or in a reference period, %T: percentage of storm events
with an emptying period ,24 hours, NT: total number of storm events in the lifetime or
in a reference period, Dslp: Functioning period during which no permanent problems are
observed (persistent water levels in the infiltration device for example). 147
 Performance P2 – Low water resources pollution
This indicator was modified many times. You can find in Barraud et al. (2004) their pre-
vious versions and their methodological drawbacks. At the design stage, the actual prop-
osition considers the thickness of the unsaturated zone as an indication of groundwater
pollution risk.
8
< If the unsaturated zone is less than 1 metre; IP2 ¼ 21
P
P. Moura et al.

Sinf2ZNS.5m ð8Þ
: If not; IP2 ¼ Sinf

In the second case, IP2 represents the percentage of infiltration areas where the unsatu-
P
rated zone is greater than 5 m; Sinf2ZNS.5m being the sum of infiltration surfaces with
an unsaturated zone greater than 5 m, Sinf the total sum of infiltration surfaces.
For the operation phase a binary indicator was defined in order to indicate whether the
system behaves normally or not. It is based on continuous monitoring of the electrical
conductivity and dissolved oxygen (DO) in groundwater downstream infiltration systems
(Malard et al., 2004).
IP2 ¼ 0

If, after a rain event, there is not a decrease of the electrical conductivity in groundwater
or, if there is no rise in dissolved oxygen (when groundwater is unsaturated with DO) or if
there is a decrease of DO (when groundwater is saturated with DO), the system does not
behave normally and the origin of the problems has to be investigated more precisely.
IP2 ¼ 1 in other cases.

Performance P3 – Positive contribution to groundwater recharge

This performance indicator is calculated by equation (Eq. 9) at the design or at the oper-
ation stage. After testing it, this indicator turned out to be of good quality.
IP3 ¼ Vinf =Vplu ð9Þ

Vinf: total infiltrated volume by the infiltration device, Vplu: total rainfall volume for the
considered catchment

Performance P4 – Little use of raw material

This performance deals with the consumption of materials and energy that are necessary
for construction and operation of an infiltration system during its whole lifetime. It is still
in progress.

Performance P5 – Low air pollution (odours, pollution…)

This performance was finally taken into account in P10 and P11.

Performance P6 – Low soil pollution

This indicator measures the aptitude of the infiltration device to keep the pollution on the
top. In the design phase, we have two indicators. One is an indicator of pressure on soil
(Eq. 10) and the second one is related to the soil ability to trap pollution (Table 2).


Sinf i
IP621 ¼ Mini £ Fi ð10Þ
Sa i

Sinf i: infiltration surface, Sa i: effective surface of the catchment, Fi: weighting factor ran-
ging from 1 to 3. Fi ¼ 1 for roads, parkings and industrial zones, Fi ¼ 2 for roofs with
148 metal parts, Fi ¼ 3 for cycle tracks, pedestrian ways and other roofs.
Table 2 Soil ability to trap pollution – IP622

pH Soil grain size

Fine material Medium to coarse material Fissured rock

5 , pH , 8 High Medium Infiltration not allowed

pH , 5 or pH . 8 Medium Low Infiltration not allowed

P. Moura et al.
As for the operation phase, two indicators were proposed. They compare the pollutant
concentrations in the soil to Dutch Standards target and intervention values (NMHSPE,
2000). Target values indicate a sustainable soil quality whilst intervention values indicate
that the soil functional properties for human, plant and animal life are seriously threa-
tened. This performance is based on two components K1 and K2 (Dechesne et al., 2004).
P P
wi wi
Ci ,CDSt Ci .CDSi
K1 ¼ P and K2 ¼ P ð11Þ
wi wi

wi: toxic weight of pollutant i, Ci: pollutant i concentration in the basin soil, CDSt: pollu-
tant i target value given by Dutch Standards (NMHSPE, 2000), CDSi: pollutant i interven-
tion value given by Dutch Standards (NMHSPE, 2000)
This definition implies that if a single pollutant exceeds Dutch Standards intervention
values, the entire soil layer is considered as being polluted. The two indicators of con-
tamination are then given as follows:
IP6 – 1 ¼ depth where pollution becomes low or nil. A low pollution is characterized
by a high K1 (K1 [ ½1 2 x; 1
, where x is an exigency threshold that generally ranges
from 0 to 0.3)
IP6 – 2 ¼ percentage of highly polluted soil samples
A different way to define the thresholds in K1 and K2 is open to question and will be
studied. However, at the present time, this definition gives satisfactory results, sensitive
and robust enough as shown in Table 3. Applied to four infiltration basins, the indicators
differentiate various situations properly. For example, ZAC des Pivolles is very different
from Droits de l’Homme and Centre Routier. The pollution is deeper (IP621 . 90 cm)
but more localised (IP622 ¼ 22%). This is actually confirmed by observation on site.

Performance P7 – Efficiency in terms of pollution retention

This performance is represented by the same indicator for the design and for the oper-
ation phase. The indicator represents the global trapping efficiency of the pollutants by
the infiltration system. The main problem is to estimate the different mass properly. The
procedure proposed by Le Coustumer et al. (2005) is used.

IP7 ¼ Mpk =Mapk ð12Þ

Mpk: mass of pollutant k trapped by the system on a reference period, Mapk: mass of pol-
lutant k brought to the system on the reference period.

Table 3 Values of IP621 and IP622 (x ¼ 0.25) – weights equal to 1

Centre Routier Droits de l’Homme ZAC du Chêne ZAC des Pivolles

60– 70 cm 45 cm 35 cm .90 cm
26% 33% 44% 22%
149
 Performance P8 – Ability to be managed and maintained easily
The aptitude to be easily maintained depends on many factors: accessibility to the differ-
ent parts to be maintained, specificity of the maintenance equipment, ability of the sys-
tems to be easily controlled, the system scale, its geographic situation. Management of
infiltration systems is often made more complex by their diverse legal status (public or
private) or by their different uses. A first indicator was proposed and based on a score
ranging from 0 to 18 evaluated by three sub-indicators (accessibility to the different
P. Moura et al.

devices to be maintained, specificity of the maintenance equipment and ability of the sys-
tems to be easily controlled). Applied to real case studies the indicator turned out to be
entirely non-objective (very different evaluations according to different appraisers). It has
been abandoned. At the present, we intend to make an evaluation by means of interviews
with the staff during the maintenance campaigns. It is still in progress.

Performance P9 – Low disturbance of other technical systems

Depending on the proximity of constructions and on the site vulnerability, the risks
induced by infiltration facilities are potentially significant. This performance needs more
investigation.

Performances P10 and P11 – Aptitude to be secure and safe for users and for the staff in charge of the
maintenance
The user safety is often put forward by the users themselves and by the decision makers.
It is thus necessary to estimate the performance of the equipment with respect to security
and to the health of users. The exposure to pollutants has to be considered. The indicators
proposed can be applied at the design and at the operation stage.
Sanitary aspect related to the soil:
The indicator is based on the average concentration of pollutants measured in the 30
first centimetres of soil. It is determined according to two components K3 and K4. K3 is
the percentage of pollutants (in concentration) lower than the recommended values
(VCIUS indicating low pollution). The K4 represents the percentage of pollutants (in con-
centration) greater than the limits (VCIUNS qualifying a high degree of pollution).
K3 ¼ %½Cmesi # VCIUSi
and K4 ¼ %½Cmesi $ VCIUNSi
ð13Þ

Cmesi: measured concentration of the pollutant i, VCIUNSi, VCIUSi: recommended values

for pollutant i according to French standards (BRGM, 2000)
if K3 ¼ 100% ) IP1021 ¼ 1

if K3 , 100% ) IP1021 ¼ K1 but; If K4 . 0% ) IP1021 ¼ 0

Sanitary aspect related to air:

 
Cmesi
K5 ¼ % .a ð14Þ
Clim i

Cmesi: measured concentration to the pollutant i, Clim i: limit concentration of pollutant i,

a: alarm value.
For users Clim i is fixed according to WHO guidelines for air quality (WHO, 2000).
If
ðCmes Þi
. 1 ) IP1022 ¼ high k . 50% ) IP1022 ¼ medium
ðClim Þi

150
k # 50% ) IP1022 ¼ low
For the staff, we use the same formulation of IP1022 but considering the soil and air
exposures together. The indicator considers the limit values for the professional exposure
to the chemical agents proposed by the INRS (National Institute of Research and Secur-
ity) (INRS, 2005).

Performance P12 – Low cost systems

P. Moura et al.
The cost is an important factor during the design phase. Two indicators have been
defined. The second can also be used for existing systems all along its life span.
IP1221 ¼ ðCinv þ Cstudy Þ=Vs ð15Þ

IP1222 ¼ Cmain =Vs ð16Þ

Cinv: investment cost, Cstudy: study cost, Vs: useful storage capacity, Cmaint: updated
maintenance cost per year

Performance P13 – Ability to preserve/encourage economic activity

The costs or site constraints related to traditional drainage pipe networks often handicap
urban and economic development of certain areas. Infiltration is then useful and makes
development possible. The considered alternative to the indicator definition is based on
information of the catchment area but no indicator is still defined.

Performance P14 – High system adaptability

This performance considers the fact that certain strategies are more “robust” with urban
development than others. The present performance is extremely complex; the actual
research is based on the urbanizing capacity related to the infiltration system.

Performance P15 – Ability to provide other functions

This performance relates to “the possible optimization” of the facility by making it multi-
functional. The multi-functionality can either be social, technical, economic or landscape
enhancing. A binary indicator integrating the proprietorship of the system (public or pri-
vate) and their different possibilities (mono-functional or multi-purpose) was proposed
(Table 4). This indicator is rather rough but at the present no better indicator has been
proposed.

Performance P16 – Good social acceptance

Even if a multipurpose system is always preferable, it remains more or less accepted by
users. Some indicators were developed but in fact the major problem is the basic knowl-
edge of social acceptance of alternative techniques. At the present, this is carried out by
social science researchers of the group but no indicator is yet proposed.

Table 4 Definition of the PI15 values

STATUS Collective and multipurpose system 1

Collective system with only one hydraulic function 0
Individual and multipurpose system with information and/or education of the users 1
Individual and multipurpose system without information given to the users 0
Individual system with only one hydraulic function 0
151
 Performance P17 – Low waste production
This indicator has to consider the quality and quantity of the produced waste. It will be
monitored in the Lyon’s infiltration basins in the near future and in that way an indicator
could be proposed.

Decision aid methodology

P. Moura et al.

Concerning the evaluation and the comparison of several alternatives at the design stage
(choice of a good project among a set of alternatives), and at the operation stage (choice
of good strategies to be applied for an existing system like the choice of a maintenance
policy, the selection of technical/social/environmental improvements, the choice of reha-
bilitation solutions), multicriteria decision aid methods are quite useful because they
allow the optimization of multiple attributes including different natures of information.
The choice of the multicriteria method has to consider the number of criteria, the nature
of each criterion and the type of problem (choosing a solution, sorting good solutions,
ranking, describing).
To use the proposed indicators the multicriteria methods which are more adequate are
those based on partial aggregation, because of the large number of different types of indi-
cators. Moreover, the method has to consider uncertainties and allow action classification.
ELECTRE III method (Roy, 1978) seems to be a good one. It is based on the integration
of the fuzzy (imprecise and uncertain) nature of the indicators. This method takes expli-
citly into account indifference, preference and veto thresholds. The ELECTRE III starts
with a finite set of actions evaluated on a consistent family of pseudo-criteria and aggre-
gates these partial preferences into a fuzzy outranking relation. Two complete pre-orders
are then obtained through a “distillation” procedure built from fuzzy outranking relation-
ships. The intersection of these pre-orders indicates the most reliable part of the global
preference.

Conclusion
In this paper the last version of a set of performance indicators related to infiltration sys-
tems was presented. The performance evaluation is diverse and uses multiple types of
indicator involving the sustainable development aspects. The proposition of a set of per-
formances is an arduous and an interactive task and depends on a multi-disciplinary
assessment. The indicators presented in the present paper are not accomplished; this work
is a step and will continue.

Acknowledgements
The authors thank the Greater Lyon for financing this project and the OTHU (Field
Observatory in Urban Hydrology) for scientific support.

References
Barraud, S., Bardin, J.P., Alfakih, E. and Abdelmalki, L. (2001). Construction d’indicateurs de performances
techniques, économiques et environnementales des ouvrages d’infiltration des eaux pluviales en milieu
urbain. Programme de recherche: Milieu physique, ambiances urbaines et technologies, Ministère de la
Recherche – Action Concertée Incitative Ville, Rapport final, p. 128.
Barraud, S., Miramond, M. and Alfakih, E. (2004). Critical analysis of the quality of a set of performance
indicators used to qualify sustainability of urban infiltration storm water drainage systems. 4th
International Conference on Decision Making in Urban & Civil Engineering. Porto, Portugal, October
28 – 30, 10 p in [CD-ROM] ed. by A. Antunes (University of Coimbra & Decision Making in Urban and
152 Civil Engineering Society).
Bellagio (1996). Bellagio Principles: Guidelines for the Practical Assessment of Progress Towards
Sustainable Development, International Institute for Sustainable Development, Winnipeg, Manitoba,
Canada. Available at khttp://iisd1.iisd.ca/measure/bellagio1.html.
BRGM (2000). Gestion des sites (potentiellement) pollués. Version 2. 314 p.
Dechesne, M., Barraud, S. and Bardin, J.P. (2004). Indicators for hydraulic and pollution retention assessment
of stormwater infiltration basins. Journal of Environmental Management, 71(4).
INRS (2005). Valeurs limites d’exposition professionnelle aux agents chimiques en France. Note
Documentaire. Institut National de Recherche et Securité. Available at kwww.inrs.frl.

P. Moura et al.
Kastner, A. (2003). Etude critique d’un jeu d’indicateurs pour l’évaluation des techniques alternatives
d’infiltration des eaux pluviales, INSA Lyon, Lyon, pp. 116, DEA Génie Civil - URGC.
Labouze, E. and Labouze, R. (1995). La comptabilité de l’environnement. Revue Française de Comptabilité,
(272), 92.
Le Coustumer, S., Barraud, S. and Béranger, Y. (2005). Procédure de mesure et proposition de modélisation
des flux d’eau et des polluants au sein des systèmes d’infiltration des eaux de ruissellement en milieu
urbain. Les XXIIIèmes Rencontres de l’AUGC “Génie-Civil: Risque et Environnement”. Grenoble: 26 &
27 mai 2005.
Malard, F., Datry, T., Bouger, G. and Gibert, J. (2004). The use of multilevel wells and multiparameter
loggers for monitoring groundwater quality below stormwater infiltration basins. 5th International
Conference on Sustainable Techniques and Strategies in Urban Water Management, Novatech 2004,
Lyon, France, June 6 – 10, pp. 713 –720.
NMHSPE (2000). Circular on Target Values and Intervention Values for Soil Remediation, The Netherlands
Ministry of Housing, Spatial Planning and the Environment. Available at khttp://www.vrom.nl/l.
Pastille (2002). Indicators into action: A practitioner’s guide for improving their use at the local level.
Pastille. Available at khttp://www.lse.ac.uk/collections/PASTILLE/images/PGUIDE.pdfl.
Perrin, B. (1998). Effective use and misuse of performance measurement. American Journal of Program
Evaluation, 19, 367 – 379.
Personne, M. (1998). Contribution à la méthodologie d’intégration de l’environnement dans les PME-PMI:
Évaluation des performances environnementales. PhD, Institut National des Sciences Appliquées de Lyon
et École Nationale Supérieure des Mines de Saint Etienne, 295.
Riley, J. (2001). Indicator quality: for assessment of impact multidisciplinary systems. Agriculture,
Ecosystems and Environment, 87(2).
Roy, B. (1978). ELECTRE III: Un algorithme de rangement fondé sur une représentation floue des
préférences en présence de critères multiples. Cahiers du Centre d’Études de recherche opérationnelle,
20, 3 –24.
WHO (2000). Air quality guidelines for Europe. Available at khttp://www.euro.who.int/airl.

153