SAWIS – Juin 2015

SPRINGER
OUTIL D’AIDE À LA DÉCISION POUR L’OPTIMISATION DE LA GESTION DES
RÉSEAUX D’ALIMENTATION EN EAU POTABLE
Meryem AGADI (1), Mohammed Karim BENHACHMI(1) ,El FILALI SANAA(2),, Aniss Moumen(3)

(1)Laboratoire de Génie des Procédés et Environnement, Université Hassan II , FSTM Faculté des Sciences et
Techniques de Mohammedia.
(2)Laboratoire : Traitement de l’Information et Modélisation, Faculté des sciences Ben M’SIK , Casablanca.
(3) 1Geosciences Laboratoire des Ressources naturelles (GeoNaRes), Faculté des Sciences, Université
IbnTofial, Kenitra.

E-mail : meryemagadi@gmail.com

Abstract :

TIn this article ,we propose an Assistance prototype for an interactive decision to manage the
rehabilitation of potable water and to reduce the difficulties and complexities of managing
intervention’s operations taking into account several criteria that are distributed according to axes
in relation with the objectives of sustainable development manager . We aims to develop a decision
support tool to provide solutions to the problems of sewer networks management/maintenance in
order to assist the manager to sort sections upon priority of intervention by taking account of the
technical, economic, social and environmental standards.The model uses the Analytic Network
Process (ANP)developed by Thomas Saaty, coupled with a set of tools for modelling and collecting
integrated data from a geographic information system (GIS) as well as modelling tools operation to
provide a lot of information that enables decision makers to have a strategic vision on the shares of
maintenance and to advanced expertise.

Keywords: Decision making process, GIS, Multicriteria analysis (ANP), rehabilitation, sustainable
development.

Introduction

The management of drinking water supply systems evolves, in the same way as other technical
networks, on the basis of environmental and socio-economic factors due to which they act and inter
operate with other networks. Main management of such networks is main aim to improve the
performance of the facilities while ensuring consumer water meets the standards of quality at a
reasonable price with a continuity of service without defect. Such facilities represent a limited life
expectancy. Indeed, the drinking water supply network constitutes a heritage that is aging and it is
necessary to renew when it reached a threshold of obsolescence limited thus implying a reduction
in yield and an increase in expenditures for the maintenance of the service. This threshold depends
on many factors whether environmental, technical and socio-economic. Where the importance of a
good understanding of the interaction.

Building infrastructures is not the only important issue that matters, but their functioning,
maintenance and control that must be assured, in addition to the implementation of an effective
strategy to guarantee their long-term function in a way that meets the objectives of a sustainable
development (Bouamrane, 2012).

1
 SAWIS – Juin 2015
SPRINGER
Nowadays, the major concern for the public sewer services is to measure the levels of the
objectives of services rendered to the users, by seeking a cost control of investment, functioning
and maintenance of the system under given conditions of operating safety to accomplish a required
function (AFNOR, 2001; Granger, 2009). To achieve this objective, the question that arises is:
What strategy should be implemented to ensure an effective management of our networks? To cope
with this situation and optimize rescissory actions, sewer networks maintenance managers need a
comprehensive methodology to assist in decision-making within a given period (Aflak, 1994).

In Morocco, the Department of water and the environment says the State of degradation of the
networks in the cities led to a loss of 35% of the delivered water «Rabouli 2014». This explains the
development of multiple methods to estimate the current and future performance and degradation
of the lines in a perspective of renewal decision making optimization. In this paper, The objective
is to offer an interactive tool for decision support that will allow managers to optimize the
programming of renewal in order to ensure a level of service and a satisfactory cost to the final
consumer. This decision support tool is proposed to reduce the difficulties and complexities of
maintenance managements. To provide general information, this includes a set of tools for data
collection, analysis, modelling, prioritizing and planning on a particular schedule.It aims to sort the
priority sections for intervention taking into account several criteria (social, institutional,
environmental, legal, techno-economic, etc.).

This decision tool based on the Analytical Hierarchy Process AHP Multicriteria method enables the
calculation of emergency replacement levels applied to drinking water supply systems. All these
elements will be integrated into a system for data collecting and managing; in other words, to
rehabilitate the appropriate section at the right time using a suitable rehabilitation technique at a
low cost (Sægrov, 2006).

ORGANISATION OF THE MANAGEMENT OF THE OPERATION OF DRINKING

WATERSYSTEMS: APPROACHES DECISION-MAKING

Currently, management of drinking water systems is no doubt one of the main urban problems
which is based on several issues more complicated on different aspects of these networks. Many
research projects have been developed in recent years in order to facilitate the decisions of
managers in the planning of intervention operations (FINA, 2006; Engelhardt, 2000; Blindu, 2004;
Haidar, 2006. Medhi, 2007). These include Patil David et al. (2011) who used the Markov model
to predict the needs of renewal, taking into account climatic factors as well as renewal and history.
Others are are supported on optimization as the genetic algorithm approach to planning the
priorities and program renewal (FINA, 2006). And some offer a model that optimizes the dates of
renewal of licence on Semiparametric models to calculate the probability of each failure and to
estimate the cost of damage due to failures.Alexandre (1996), Blindu (2006), the waffle and al.
(2002) proposes to use methods of multi-criteria analysis, for the prioritization of maintenance
actions and the choice of the technical rehabilitation. It may be noted also the European project
CARE-W (Computer Aided Rehabilitation of Water networks) which includes five modules
connected to a database of tools and management including subject-specific modules is to build a
model for the rehabilitation of drinking water (haddock, 2006).

ADOPTED METHODOLOGY
We propose an optimization tool based on an innovative approach structured in multiple successive
and separate steps to improve the strategy of the current management of drinking water supply
networks, to help the Manager to make his decision successfully.

2
 SAWIS – Juin 2015
SPRINGER
This approach will be integrated in the system tools for decision support to meet the demand for
managers and which policy should move towards designing a sustainable and economic system.
The proposed methodology is consisted by several levels of modeling and treatment (Figure 1):
 Data structured collection
 Implementation of a conceptual model based on modelling:
 Mapping and cartography results on geographic information software.

Figure 1: The methodology proposed to optimize the management of drinking water supply
system networks

APPLICATION OF THE METHODOLOGY TO THE SYSTEM

OF THE CITY OF CASABLANCA: CASE OF DISTRICT TAQADDOUM OF
SIDI BERNOUSSI
Located northeast of the Morocco, it covers an area of 50 hectares. It has a population of more than
4 million inhabitants. The distribution of drinking water in Casablanca system consists of a
network with a total length of more than 4.541 km its management is ensured by the LYDEC
(Figure 2).

Figure 2: Drinking water - pressure-storey distribution network (Source: BCEOM/G2C, 2008).

3
 SAWIS – Juin 2015
SPRINGER
PROGRAMS CONSTRUCTION SUPPORT FOR
INVESTIGATION
AND REHABILITATION

DATA COLLECTION STRUCTURATION

Data collection conducted as part of our work aimed to gather all the raw data and previous or
ongoing studies available on the functioning of sewer networks. And like all database management
systems, our system is powered by a set of data and information about the network, its environment
and the hydraulic and structural performance defined by managers or by the modelling. For this, a
decisional database is intended to collect all data and results concerning the decisional modelling.
The data collected about the objective of our work are:
 Various technical features of the system (description plans, functional data, networks and
structures, materials, age, etc ...).
 Natural environment of the system (physical environment, geology, climate, local
resources, etc...).
 Social and economic environment of the system (concerned human activities and urban
areas, municipal boundaries, population, traffic, economic activities etc...).

EVALUATION OF THE HYDRAULIC PERFORMANCE OF SEWER NETWORKS

Model the behaviour of the network along its distribution using a simulation model is essential to
understand and interpret its functioning in different States. This modeling was conducted under
software EPANET, and necessitated the creation of a scheme reflecting the physical characteristics
of the components of the system: reservoirs, pipes, junctions, etc. Namely:
 pour a section of line: length, diameter, surface roughness, presence or absence of a closed
valve, a valve;
 pour a junction (or node): altitude, application, type of modulation
 pour curve a tank: the invert elevation, diameter, low level, high level, starting level.

The compendium of physical data on the network, as described above, is complemented by

representative data for the operation of the network. In particular:
-the floor of pressure supplying the area is upstairs 96 (96 being served side);
-the supply tank has a capacity of 52000 m 3;
-valve separation exists upstream from the distribution network of the district;
-the measuring (pressure and flow) PM 13 point is located upstream from the network

The design of the model has been conducted step by step from the network mapping software
allows to install in background. The results of the modelling used therein to describe and make an
abstraction over the current and future situation of different parameters such as pressure, speed and
flow. The validation/calibration of the latter requires numerous information and measurements in
the field such as flow, pressure each node and stretch. To implement this, we conducted this
operation on the district Taqadoum is one of the ancient districts of the prefecture of Sidi Bernoussi
in Casablanca, the district date over 60 years. We conducted three types of measures. One was
made on the tank (output) and the other two on the lower parts of the network to measure pressure.
All this with the goal to validate the calibration of the model (Figure 3).

The following graph presents the measures which have been carried out. The Pikes of the night rate
(Q min) is an indicator of abnormal consumption, fraud, or most often physical loss due to network
failures.
4
 SAWIS – Juin 2015
SPRINGER

Figure3 : Night flow Q min Taqadoum February 2012 ; Figure4 : Import of network 2

This district drinking water distribution network is composed principally of asbestos cement, as old
as the District lines, and which are the subject of several problems of degradation and aging of the
network (leaks, breaks, pressure drop, etc.) This work focuses on the blocks 35, 36, 37, 38 and 39
of the district (Figure 4) .

MULTICRITERIA DECISIONAL MODELLING OF MAINTENANCE PROGRAM:

We used ANP model steps in our case study as a decision-making technique with the ability to
involve criteria interactions and dependencies, in order to provide a more accurate approach to
model the decision’s complex environment. The model is developed by Thomas Saaty as an
extension of the AHP model to solve the problem of its strictly hierarchical design between the
decision levels and with no ability to handle all the complexities of the real world decisions. ANP
offered synthetically a logic that presents the problem as a network of criteria and alternatives
(sections) and takes in consideration the dependencies and interdependencies among various
criteria and sub-criteria. To use this technique, four main steps are required (Saaty, 1996; Saaty,
2004) as follows:

First step: Define the problem and determine its aim.

The structure is defined on the basis of interviews with various interviewers in decision-making
and relevant studies for the sewer networks management.
With regard to the construction of the model, we determined the elements that are interdependent.
Indeed, a (test) Group (sub-criteria) elements can influence on other elements in the same group,
that is, those other groups.
So in order to achieve our research method that treats the problem of the choice of priority sections
to intervene, we have developed three evaluation groups (criteria).
Regarding the model construction, it is necessary to determine the interdependent elements. Indeed,
the elements (sub-criteria) of a group (criterion) can influence whether other elements in the same
group or in other groups to realize our study problem and looks at the problem of selecting the
sections with the priority of intervention.
For this, we have built two evaluation groups (criteria). (Figure 5):
Technical and Environmental group (criteria) with elements (sub-criteria) that are:
Network type, hydraulic function, external factor, structural sub-criteria, hydrogeological
risk, degree of pollution and the site vulnerability.
Economic and Social group (criteria) with elements (sub-criteria) that constitutes it:
Disturbing the economic activities, Traffic commoners and the lifecycle of the pipe,
population density, importance of the place and impact of the maintenance.

5
 SAWIS – Juin 2015
SPRINGER

Figure 5: Decompose the problem complex in a hierarchical structure (levels).

Second step: Pairwise comparisons and priority vectors.

In ANP, like AHP, pairs of decision elements at each cluster are compared with respect to their
importance towards their control criteria. The matrix is the most effective framework for such
comparisons in pairs (blindu 2004).
Comparison between all criteria is given by the following matrix: A = [a ij] of order n (1)
Where: aij = wi/wj, wi and wj are the relative weights of the criteria gi and gj respectively.
The establishment of measures for the criteria is a need for the comparisons to specify the
importance degree of one criterion over another. Table 1 gathers the scales used to make the
pairwise comparisons. (Table 1)

Table 1 : Determination of weights associated with each criterion.

To calculate the relative importance (weight) of each criterion with regard to its contribution to the
objective, the procedure is as follows:
The values in each column are summed.
Each element in the matrix is divided by the sum of its column (normalization).
The average for each element in a row of the matrix is calculated.

The averages represent the weight vector (eigenvector). The weight associated
with the evaluation criteria i is given by the following relationship:

With the sum of wi that must be equal to one.

Consistency of judgment

6
 SAWIS – Juin 2015
SPRINGER
The great advantage of the method is that it allows calculating a consistency index, which evaluates
the calculations done. Thus, we can know to what extent our judgments are consistent, since we
want to avoid that our decision would be based on little coherent assessments that might seem
random. The coherence index (CI) is determined by the following formula:

Table 2: Random Index

Where: max is the matrix biggest eigenvalue.

n: is the number of compared elements.

The larger the consistency index becomes, the more the judgments of the user are inconsistent

Coherence ratio (CR) is given by the following formula:

With:

CR: the consistency ratio.

RI: random index (Table 2).
The assignment of weights is acceptable if CR is less than 10%. In case it exceeds 10%, the
assessments may require some revisions.

Third step: Creating the supermatrix

The concept of supermatrix is similar to the
Markov chain process (Saaty, 1996). It aims to
obtain the overall priority of every criteria and
sub-criteria with the interdependent influences. it
is necessary to introduce local eigenvectors
estimated in step 2 in the homologous columns.
Each column in the matrix represents the relative
priorities of all elements, compared to a given
element. This matrix is called unweighted supermatrix.

W: unweighted supermatrix.
eij: j th element of the ith group.
Wij: Matrix of relative priorities among the elements of group C and elements of the group Cj.
Ni: elements number in group i.

The supermatrix must be stochastic in column, i.e., the sum of a column is equal to 1. To obtain
limited priorities, it needs to multiply each block of the nonweighted supermatrix by the weight of
the group (corresponding to the block) in the group matrix, which generates a weighted
supermatrix.
For this, the limited supermatrix calculation is present as: the weighted super matrix will reach a
steady state until all columns of the weighted supermatrix converge to the same values and each
row i of them goes to a consistency αi to complete their convergence. The weighted supermatrix is
raised in power 2k + 1; where k is an arbitrary number. Finally, the obtained final results give us
the final weight (Final Priorities) of each criteria.

Fourth step: emergency level Evaluation for each alternative (section):

In case where a limited supermatrix is developed, the last thing is to assess the emergency level
(total weight) of each section. The largest among them compared to the emergency level should be
one with intervention priority.
7
 SAWIS – Juin 2015
SPRINGER

APPLICATION AND CASE STUDY

Presentation of the study area
This methodology is applied on an area of a district SIDI BERNOUSSI of the city of Casablanca,
which is located in Northouestern Morocco. It covers an area of 38 km 2. The district of Sidi
Bernoussi (165 000 inhabitants) is the more industrial Casablanca (there are between 6,000 and 10
000 Mills). However, there is still a populaire.and a population area. The city has a sewer network
with a connection rate covering currently 94 % of the 390 kml.

Prioritization of sections (alternatives)

After obtaining the evaluation results of the overall relative weights to each criterion or sub-
criterion, it is necessary hereafter to proceed with sections hierarchization operation (alternatives)
for the maintenance action. The
calculation method used is simple: it
starts by evaluating the level of urgency
for each section that will be calculated
by the sum of the relative weight
assigned to all the criteria of our grid and
present by the following relationship : j ij
Pcj l (5)
Where:
Pcj: the overall value of the urgency
level for each section
lji: the rate of the overall relative
weights for each criterion Cj of section Ii
After having obtained the urgency levels of each section (figure5), the resulting weighted sums
sorted in a descending order, which allows carrying out a sections ranking to highlight the position
of each one (Figure 6).

Figure 6: Example of application analysis multi-criteria under Arcgis

RESULTS ANALYSIS AND EXPLOITATION

The hierarchy of the sections
The final result is agenda of renewal of pipelines and subsequently, having a tiered plan for the
renewal of lines in the Taqadoum district, in the form of map that will be easily exploited by
managers. In implementing this step has been facilitated by the use of the Super Decisions software
is a graphical user interface and will execute under Windows environment. It offers the possibility
to do a sensitivity analysis of the different parameters of the method to test the impact of their
variation on the result.
8
 SAWIS – Juin 2015
SPRINGER

Figure 7: Value of urgency levels

Figure 8: Value hierarchization of the urgency levels and sections

Sensitivity/robustness analysis
The sensitivity analysis of the results is a very important step because it helps supporting the
validity of the results obtained by specifying the limits in which they allow conclusions that remain
robust and stable towards the variation of one or more parameters used (Oumhani, 2006).

The results of the sensitivity analysis showed that the hydraulic criterion should be considered in
the same priority as the structural one in the implementation of maintenance decision operations in
view of the graphical results that show a small displacement compared to the reference solution
(Assignment1) (Figures 9 and 10). This result is supported by the work of Mazouzi(2012).

First, we conducted a sensitivity analysis for some criteria individually (hydraulic and structural)
to show their influence on the obtained results. (Assignment 1).
In a second step, we looked for the capacity of the proposed solution to resist tolerable change in
the weight assignment . (Assignment 2).

This decrease causes a slight change in the priority ranking with a simple permutation of the
hierarchical order of sections (Figures 9 and 11).

Figure 9: Graphical representations of sensitivity and robustness analysis results.

9
 SAWIS – Juin 2015
SPRINGER

Figure 10: Comparison of the values of the basic solution (assignment 1) and sensitivity test
(assignment 2).

Figure 11: Comparison of the values of the basic solution (assignment 1) and robustness test
(assignment 3).

CLASSIFICATION OF PIPES ACCORDING TO THEIR UERGENCY CATEGORY

The method ABC called the law 20-80 was developed by the Italian Wilfredo Pareto in the
economic field, adopted in many sectors, particularly in the field of maintenance (Zwingelstein,
1996 ).
The method is the most effective framework for our case to represent the urgency categories. It
allows the manager to identify the priority actions targets, but also to determine the negligible
elements to alleviate the study (and Monchy and al., 2010).
The results obtained by the ABC method in our case study are as follows (Table 3, Figure 12):
A. Low priority pipes: The 70% of the network pipes cost 05% of all annual values of the
maintenance interventions fees.
B. Average priority pipes: The 10% of the network pipes cost approximately 15% of all
annual values of the maintenance interventions fees.
C. High priority lines: The 20% of the network pipes of high emergency level explain 80%
of the interventions. For that, these pipes should be classified as strategic and require
increasing number of inspection operations and preventive maintenance interventions to
ensure their conditional optimal mission cost and damage.

Table 3 Emergency Categories Results

Class Urgency category weight
A Low priority A < 0,2101
B Average priority 0,3001 > B > 0,2101
C High priority C > 0,3001

10
 SAWIS – Juin 2015
SPRINGER

Figure 12: Emergency categories

INTEGRATION OF GEOGRAPHIC INFORMATION SYSTEM GIS IN THE DECISION

PROCESS

To better see the results of the analytic hierarchy process , it is necessary to define the level of
urgency of each segment for the maintenance action.
It is therefore necessary to execute the step of integration of IMIS in order to generate a real
progress in the management of the network. The contribution of GIS allowed us to characterize the
combination of all these last with the urban existing database which aims to facilitate the tasks of
decision-makers to navigation on the network data, and produce also a new family of cards called
thematic maps priorities for action (figure 12) and it includes three classes : low, medium and high
priority to intervention.
The results provided by the previous methods are in the form of notes and emergency categories of
maintenance actions. They are not flexible to be exploited, so these results will appear in the form
of images to facilitate the manager’s tasks.
Concerning the visualization that is supposed to be one of the strengths of GIS, the results are
displayed in terms of priority for each line so that the lines appear on the screen associated with
colours that reflect their degree of priority set by the ABC method. Other relevant spatial data such
as hydraulic operation of networks, roads, frames, etc... could easily be viewed.

Figure 12: Cartographic Visualization of priority pipes in intervention maintenance.

11
 SAWIS – Juin 2015
SPRINGER

CONCLUSION

The objective of this work is to implement a tool to help a decision for the management of drinking
water systems. And thus facilitate the complexity of the mission manager maintenance while taking
into account a set of criteria and sub-criteria for decision-making on various criteria (technical,
economic, social and environmental) associated with the objectives of sustainable development,
and the strategy of the Manager.
The core of this tool is an algorithm of processing and analysis steps exploiting simultaneously the
advantages that the integration of GIS, the Analytic Network Process (ANP) method as well as the
network modelling tool (SWMM).
The integration of these latest with GIS and hydraulic modeling (EPANET) has strengthened the
potential of decision through the storage capabilities, management, analysis and display of IMIS
and profit conducted by modelling that allows to evaluate the current hydraulic operation and this
after the application for intervention, rehabilitation or replacement. They offer a robust tool capable
to process and analyze network data quickly and to evaluate its performance.
The application of the proposed methodology allows all stakeholders to make informed decisions
and to intelligently share information but also to save, process and finally refresh the urban base
that exist. In addition, the tool provides a daily inspection and maintenance program and annual
multi depending on the urgency to deal with.
In perspective we intend to address in the future taking into account statistical tools for focus on the
development of a probabilistic model to forecast the pipes degradation in order to optimize the
pipes renewal date considering all direct and indirect maintenance costs as well as evaluating the
economic and social failures consequences of the network and works implemented to repair them.

REFERENCES BIBLIOGRAPHIQUES

AFLAK, A., (1994). Elaboration d’un cadre methodologique pour l’aide a la décision en matière de
gestion de la maintenance du réseau technique urbain d’assainissement, These Doctorat de
l’I.N.S.A. de Lyon, 300 p.
ALEXANDRE O., ELNABOULSI J. (1996). Le Renouvellement des réseaux d’eau potable,
Canadian Water Resources Journal.
BENGASSEM J. (2001). Élaboration d'un système d'aide au diagnostic hydraulique et structural
des reseaux d’assainissement urbains. These doctorat, école de technologie supérieure université du
Québec, Canada, 149p.
BENZERRA A., CHERRARED M., CHOCAT B., CHERQUI F, ZEKIOUK T. (2012). Decision
support for sustainable urban drainage system management: A case study of Jijel, Algeria, Journal
of Environmental Management Vol.101, 46-53
BLINDU I. (2004). Outil d'aide au diagnostic du réseau d'eau potable pour la ville de Chisinau par
analyse spatiale et temporelle de dysfonctionnements hydraulique, Thèse doctorat, école nationale
supérieur des mines Saint Etienne, France, 263p.
BOUAMRANE, A. and al. (2012). Outil d’aide a la gestion et la maintenance des reseaux
d’assainissement, troisieme forum de l’eau Djerba (TUINISE) ,26 au 28 mars.
CHERRARED M., CHOCAT B., BENZERRA A. (2007). Problematic and feasibility of
sustainable development of urban sewerage in Algeria, Novatech, 295-302
CHARRAT M. (1995). Conception d’un outil de supervision de la production et de la distribution
d’eau potable à Lyon en période courante et en période de crise, Thèse de doctorat, INSA Lyon,
278p.
DIRKSEN J., CLEMENS F.H.L.R. (2008). Probabilistic modelling of sewer deterioration using
inspection data, Water Science and Technology, Vol. 57, n o 10, 1635-1641.
ELNABOULSI J., ALEXANDRE O. (1998). Le renouvellement des réseaux urbains d’eau
potable : Une approche economique d’optimisation, Ingénieries – EAT, n° 15.
ENGELHARDT M.O., SKIPWORTH P.J., SAVIC D.A., SAUL A.J., WALTERS G.A. (2000).
Rehabilitation strategies for water distribution networks: a literature review with a UK perspective ,
Urban Water 2, 153- 170
12
 SAWIS – Juin 2015
SPRINGER
GAUFFRE P., LAFFRECHINE K., BAUR R., POINARD D., HAIDAR H., SCHIATTI M. (2004).
Aide multicritere aux decisions de rehabilitation d’un reseau d’eau potable, XXIIe Rencontres
Universitaires de génie civil, Marne-La-Vallée, France.
AGADI M, BENHACHEMI K. , (2014). Outil d’optimisation du diagnostic et de la
programmation des interventions pour maintenir le fonctionnement du réseau d’Alimentation en
Eau Potable, WISMed DAYS, 20 Mars- 22 Mars, Rabat, Maroc.
AGADI M. , BENHACHEMI K. , (2014). Outil d’aide à la décision pour l’optimisation de la
gestion des réseaux d’alimentation en eau potable, Workshop sur le Système Régional
d’Information d’Eau– SRIE 2015, Errachidia, Maroc
HAIDAR H. (2006). Rehabilitation des reseaux d’eau potable: méthodologie d’analyse multicritere
des patrimoines et des programmes de rehabilitation, Thèse doctorat, Institut National des Sciences
Appliquées de Lyon, France, 227p.
LAARIBI A. (2000). SIG et analyse multicritère, Paris, Éd. Hermes, Science Publications, ISBN 2-
7462-0122-4.
LAFFRECHINE K. (2010). Aide à la décision pour la gestion patrimoniale des réseaux d'eau.
LE GAUFFRE P., BAUR R, LAFFRECHINE K., MIRAMOND M. (2002b).
Multicriteria decision support for the annual rehabilitation programmes of water networks,
Proceedings Int. Conf. Decision Making in Urban and Civil Eng., DMUCE’2002 London (UK), 6-
8 November 2002, ISBN 0904354474, 655-660.
MEHDI A., FRÉDÉRIC C., JEAN-CHRISTOPHE M., PASCAL L.G. (2014). From sewer
inspection programmes to rehabilitation needs: research and results related to data quality and
availability with the support of numerical experiment, European Journal of Environmental and
Civil Engineering.
MICHAEL M., SITZENFREI R., KLEIDORFER M., RAUCH W. (2014). Performance
improvement with parallel numerical model simulations in the field of urban water management.
NAFI A. (2006).: La programmation pluriannuelle du renouvellement des reseaux d’eau potable,
Thèse doctorat Université Louis Pasteur, Strasbourg I, France, 232p.
POINARD D., LE GAUFFRE P., LAFFRECHINE K., HAIDAR H. (2005). Modélisation du
vieillissement des réseaux d'eau potable, Revue Européenne de Génie Civil, Vol. 9, n°3, 415-428.
SAATY T. (1990). How to make a decision: the analytic hierachy process, European Journal of
Operational Research, Vol.48, 9-26.
SAATY T. (1996). Decision making with dependence and feedback: The Analytic Network
Process, WS publications, 4922 Ellsworth Avenue, Pittsburg, PA 15213.
SAATY T. (2004). Fundamentals of the Analytic Network Process-Dependence and feedback in
decision making with a single network, Journal of systems science and systems engineering, 129-
157.

13