Sei sulla pagina 1di 37

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/278296366

Performance analysis in water supply

Chapter · January 2009

CITATIONS READS

7 330

1 author:

Carlo Ciaponi
University of Pavia
30 PUBLICATIONS   242 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Carlo Ciaponi on 15 June 2015.

The user has requested enhancement of the downloaded file.


EDITOR
Copyright © Centro Studi Idraulica Urbana (CSDU) 2009
c/o Politecnico di Milano, D.I.I.A.R. - Sez. CIMI
piazza Leonardo da Vinci, 32 - 20133 Milano
Tel. (02) 2399 6212 - fax (02) 2399 6207

Tutti i diritti sono riservati a norma di legge


e a norma delle convenzioni internazionali.

Qualsiasi riproduzione, parziale o totale, anche a scopo


didattico, priva di autorizzazione scritta da parte del Centro
Studi Idraulica Urbana, sarà perseguita ai sensi di legge.

ISBN 978-88-903223-0-3

Ristampa:

4 3 2 1 0 2009 2010 2011 2012

Stampato da Digital Print, Segrate (MI)


Printed in Italy
Il CENTRO STUDI IDRAULICA URBANA (CSDU) è un’associazione senza scopo di
lucro a carattere nazionale nata nel 1992 per iniziativa di un gruppo di docenti universitari
allo scopo di promuovere ed effettuare ricerche, pubblicazioni ed iniziative attinenti alle
materie di idraulica, costruzioni idrauliche e ingegneria sanitaria-ambientale.
Il CSDU ha sede in Milano presso il Politecnico di Milano - Dipartimento di Ingegneria
Idraulica, Ambientale e del Rilevamento - Sezione CIMI.
Aderiscono, in qualità di soci sostenitori, imprese e ditte impegnate nel settore della
progettazione e costruzione dei manufatti ed impianti attinenti il settore idraulico.

PUBBLICAZIONI DEL CENTRO STUDI IDRAULICA URBANA


F. Calomino, A. Paoletti (a cura di), Le misure di pioggia e di portata nei bacini sperimentali
urbani in Italia, CSDU, Milano, 1994
C. Cao, G. La Loggia, C. Modica (a cura di), I modelli di dimensionamento per le fognature
pluviali. Atti dei Seminari di Carloforte (1-3 ottobre 1992) e di Taormina (2-4 giugno 1993),
CSDU, Milano, 1995, solo su CD ROM
A.A.V.V., Sistemi di Fognatura. Manuale di Progettazione, CSDU – Hoepli, Milano, 1997
G. Calenda, A. Paoletti, G. Rasulo (a cura di), Modelli quali-quantitativi del drenaggio
urbano. Atti dei Seminari di San Cassiano (28-31 marzo 1995) e Sorrento (11-14 giugno
1996), CSDU, Milano, 1998
A. Paoletti (a cura di), Risorse idriche ed impatto ambientale del drenaggio urbano. Atti delle
giornate di studio in memoria di Carlo Cao. Cagliari 18-20 novembre 1998, CSDU, Milano,
2000
G. La Loggia (a cura di), Dalle fognature alla tutela idraulica e ambientale del territorio.
Atti della II Conferenza Nazionale sul Drenaggio Urbano. Palermo 10 - 12 maggio 2000,
CSDU, Milano, 2001
G. Becciu, U. Maione, B. Majone Letho, R. Monti, A. Paoletti, M. Paoletti, U. Sanfilippo (a
cura di), New Trends in Water and Environmental Engineering for Safety and Life: Eco-
compatible solutions for aquatic environments 2nd International Conference. Capri (Italy),
June 24-28 2002, CSDU, Milano, 2002, CD ROM allegato
A. Paoletti, P. Piro, U. Sanfilippo (a cura di), La tutela idraulica e ambientale dei territori
urbanizzati. Atti dei seminari di Parma (5-6 febbraio 2004) e Cosenza (13-15 dicembre
2004), CSDU, Milano, 2005
A.A.V.V., Acqua e città. I Convegno Nazionale di Idraulica Urbana - Sant’Agnello (NA) 28-
30 settembre 2005, CSDU, Milano, 2005, CD ROM allegato
A.A.V.V., Acque di Prima Pioggia: Insediamenti Produttivi ed Infrastrutture. Atti della
giornata di studio. Genova, 26 Novembre 2004, CSDU, Milano, 2005
A.A.V.V., La separazione delle acque nelle reti fognarie urbane. Atti della giornata di studio
Roma 25 giugno 2003, CSDU, Milano, 2007
A.A.V.V., Acqua e Città. II Convegno Nazionale di Idraulica Urbana - Chia (CA) 25-28
settembre 2007, CSDU, Milano, 2007, CD ROM allegato
A.A.V.V., Water Resources Assessment under Water Scarcity Scenarios, CSDU, Milano,
2008
1. PREFACE

The complexity of problems related to water supply systems management and the need for new
approaches which could integrally analyze all involved aspects (environmental, social, economical and
technical) have led, in the last few years, to the development of new procedures for measuring the
performance of water systems and services.
These methodologies mainly use synthetic performance indicators (PIs for short) which have been
ascribed different formulations by various institutions of international relevance.
The sets of indicators proposed till now are mainly based on the industrial aspects of water services
and are not suitable to support most of the decisions that need to be made in the different phases
(planning, design, realization and management) by which the organization of water services can be
developed. Each of these phases, in fact, is characterized by crucial choices requiring specific decision
tools which must be appropriate for the problem scale and congruent with the available (at the same
scale) information.
For example, the planning phase requires indicators and decision support tools which can compare
different solutions not only in terms of technical and economic efficiency, but also in terms of water
saving and environmental sustainability, with particular reference to the vulnerability and the
impoverishment risk of water resources. According to these needs, the PIs systems available in
literature are inadequate and they must be integrated in a significant way.
Passing from the planning stage to the realization stage, design activity plays an important role with
regards to an efficacious use of the financial resources; its development, in accordance with
appropriate levels of investigations, requires indicators which are able to provide more detailed
evaluations than the ones involved in the planning stage. In the water supply field, scientific literature
focuses particular attention on the calculation problems (hydraulic analysis and design) of water
distribution systems. Regarding the resolution of these problems, in recent years, the integration of the
traditional mathematical approaches with new tools which are better measuring and quantifying the
system performances, taking into account different functional aspects usually not considered in the
traditional analysis (such as those concerning reliability) has become an important task. However,
presently, the definition of reliability indicators usable in design procedures is a significant and
ongoing problem, worthy of further in-depth studies.
The management stage presents several different problems many of which are already considered
by the indicators proposed in international literature. However, while most of the existing indicators
are connected to factors concerning the technical and financial organization of the management
(internal factors), relatively easy to control, the variety and complexity of managerial problems often
derive from external factors, such as the variability of available water sources and of users’ demands,
as well as the failures of components in the system subject to wear with age and to accidental damages.
The variability of water sources becomes a critical problem in the system management when water
sources are scarce, as happens in many areas in southern Italy and in many Mediterranean countries. In
these cases, the usual problems of network management and maintenance are inextricably linked to the
need to provide a uniform, balanced and fair water distribution and specific indicators have to be set
up for the evaluation of managerial aspects associated with drought situations.
A further element that can affect the efficacy and efficiency of the managerial organization is the
behavior of the users in terms of water demands. The evaluation of this aspect becomes even more
important in the above-mentioned circumstances of drought.
Finally, an important managerial problem, particularly relevant in the Italian scenario, regards
water losses reduction. The analysis of the generation and evolution processes of water losses has been
internationally faced, finding a structured organization in the activities developed in England since the
1990s and supported by stakeholders through the National Leakage Control Initiatives. Following
researches, promoted since 1996 by the IWA Water Losses Task Force, have led to the formulation of
specific performance indicators. In this context, however, it has to be highlighted that the results
obtained so far, mainly developed in the UK, have to be analyzed in order to detect possible
specificities related to validation environment and to verify the possibility of applying them to different
scenarios, such as those in Italy.
In conclusion, the analysis of scientific literature points out that operators and researchers are
significantly interested in the performance indicators, as they are a useful and strong decision support
tool within the different activities that characterize water distribution system organization. The issues
related to the performance indicators, internationally developed only since the end of the 1990s, still
represent a very new topic requiring further investigations.
In this context, five research groups belonging to five different Italian Universities (Polytechnic of
Milano, Universities of Bologna, Catania, Palermo and Pavia) proposed in 2005, in the framework of
PRIN (Relevant National Interest Project), a research project aimed at developing and validating, also
by application to study cases, a set of performance indicators which can be integrated with those
already proposed in literature, in order to solve typical problems related to technical activities of
planning, design and management in the water supply field.
The research project was financed and developed in the two-year period 2006-2007.
This book collects the reports showing the research activities and results obtained by the different
groups.

Pavia, January 2009

Carlo Ciaponi
(coordinator of Research Program)
2. TABLE OF CONTENTS

Performance analysis in water supply


C. Ciaponi ..................................................................................................................... 1

Performance indicators for the analysis of water supply systems at planning level
G. Becciu, A. Paoletti, U. Sanfilippo ........................................................................... 29

Indicators of environmental compatibility for the planning of water supply systems


G. Becciu, A. Paoletti, U. Sanfilippo ........................................................................... 41

Characterization of urban water consumption by the analysis of historical time series


A. Campisano, P. Cutore, C. Modica .......................................................................... 51

A simplified procedure for water distribution networks reliability assessment


C. Ciaponi, L. Franchioli, S. Papiri ............................................................................. 71

A performance based approach for the analysis of urban water distribution systems
under drought conditions

M. Cannarozzo, C. M. Fontanazza, G. Freni, V. Notaro, E. Oliveri............................ 95

Short term forecasting models for urban water consumption


A. Campisano, P. Cutore, C. Modica ........................................................................ 121

Water losses analysis in water distribution systems by means of performance indicators


T. Liserra, C. Bragalli, M. Maglionico, S. Artina ...................................................... 149

Performance indicators in water distribution systems aggregated by means of bayesian


networks

M. Maglionico, T. Liserra, C. Bragalli ..................................................................... 165


C. CIAPONI – Performance analysis in water supply

PERFORMANCE ANALYSIS IN WATER SUPPLY

Carlo CIAPONI

SUMMARY

This paper deals with performance evaluation in the water supply field from a general point of
view. It distinguishes between approaches based on industrial aspects of water services and aimed
at evaluation of their efficacy, efficiency and cost, and those which are more scientific, proposed
for measuring the hydraulic performance of systems.
With reference to the latter types, the paper goes into the analysis of distribution networks
performance based on the measurement of reliability obtained using indicators which are defined
by the relationship between volumes actually delivered and those required by the users.
In particular, the problems associated with the hydraulic simulation of the distribution
networks using the PDA approach, the probabilistic analysis of mechanical failures and the
characterization of demand are dealt with.

1. INTRODUCTION

The complexity of problems linked to the water supply field and the need for approaches which
could integrally analyze all aspects involved (environmental, social, economical and technical)
have led, in the last few years, to the development of new procedures for measuring the
performance of water systems and services.
In the water supply field, the development of these methods appears to essentially move along
two distinct lines although characterized by some contiguities and intersections.
The first one was conceived and developed in close connection with the industrialization
process of water services with the principal aim of efficacy, efficiency and costs assessment.
Evaluation is based on synthetic performance indicators, generally very simple in structure (often

 Dipartimento di Ingegneria Idraulica e Ambientale, Università degli Studi di Pavia, via Ferrata, 1 – 27100 Pavia

1
PERFORMANCE INDICATORS FOR THE PLANNING, THE DESIGN AND THE MANAGEMENT OF WATER SUPPLY SISTEMS

even banal). Values are obtained by way of elaborations, often elementary, of data collected and
organized in a suitable database. These indicators are mainly individuated and proposed by
international Associations or Organizations, but are not lacking in significant input from the
scientific community.
The second way, however, encompasses the various analysis methods aimed at hydraulic
performance of water systems (especially the networks) evaluation. These techniques, which are far
more complex than those developed in the first approach, can be regarded as the natural evolution
of the calculation models and methods available from the 60s and the 70s, consolidated in
professional practice. The development area is that of scientific research; these methods at present
constitute the object of most published scientific works pertaining to questions regarding the water
supply sector.
Literature covering both approaches is vast and it is not easy to get ones bearings amongst the
different works. The first publications of this type date back to the 80s and, still today, there is no
clear common thread that leads to methods which are completely convincing under a conceptual
profile and applicable on an operative level.
This report highlights various aspects have to be considered with particular attention.

2. PERFORMANCE ANALYSIS OF WATER SERVICES

As already mentioned in the introduction, an important vein of development in water supply


performance analysis regards the methodologies (developed in connection with water services
industrialization) aimed at measuring (in terms of efficacy, efficiency and costs) the levels of water
services which depend upon the water system structure, the physical and social-economic context
of supplied area, the operative practice and the organization and administrative approach of
management.
These methodologies make mainly use of synthetic performance indicators (PIs for short)
which have been ascribed different formulations by various institutions of international relevance,
among which are the World Bank (Yepes e Dianderas, 1996), the International Water Association
(IWA) (Alegre et al., 2000; Alegre et al., 2006), l’Ecole Nationale du Génie Rural des Eaux et des
Forêts (ENGREF) (Guérin-Schneider, 2001) and the American Water Works Association
(AWWA) (AWWA, 2004). In Italy, a set of indicators principally aimed at assessment of efficacy,
efficiency and inexpensiveness with which the water manager company delivers the water service
was adopted by the Comitato per la Vigilanza sull’Uso delle Risorse Idriche (2006).
The indicator system known as IWA-PIs (Alegre et al., 2006), is particularly important in terms
of its acknowledgement and diffusion on an international level too. This system is made up of 170
indicators that deal with numerous aspects of the service concerning, amongst others, water
resources used, the physical components of the system (reservoirs, networks, purification
treatment, etc.), frequency of inspections and breakages, the extent of losses, staff involved, costs,

2
C. CIAPONI – Performance analysis in water supply

revenue, profits, etc. The whole evaluation system is based on data constituting over 200 quantities
deriving from different aspects of the service. All required data are generally fairly easy to collate
if manager is diligent; it can be gathered from information that manager has direct access to
(especially regarding economic and financial aspects, personnel and the users supplied) and from
archives (data regarding the system-historical data) or from measurements (especially of water
volumes passing a certain section of the system in order to meet necessary water volume balances).
At least as far as the original IWA formulation is concerned, the use of simulation models is not
strictly required to obtain any of the data, although in rare cases it may be suggested.
Performance measurement is carried out using very basic mathematical structures (often
requiring a simple division) considering two or more quantities considered in the database. The
new quantities defined in such a way are created to enable the system performance to be quantified
with reference to a particular aspect; these therefore constitute performance indicators and render
the quantitative information collected in the database far more visible.
The performance measurement is then translated into a verdict by comparison of the obtained
values with the reference values such as values-objective (such that the indicator expresses how far
the performance is from the desired objective), values relating to previous evaluation periods
(examining PI values over a period of time gives useful information regarding the efficacy of
actions taken by manager to correct any dysfunctions) or values obtained under other managements
(to highlight and explain any significant differences). Sometimes, especially when the indicator
derives from a dimensional quantity that only an expert can interpret, the judgment is expressed by
a score between 0 and 1, assigned by means of appropriate penalty curves (Coelho, 1997).
Because of evaluation system simplicity and database structure, some authors maintain that this
type of assessment is too general and incapable of taking into account all the specific aspects of the
system. This criticism is, in part, shareable. For example, with reference to inefficiency that may
afflict users, the IWA-PIs are limited to calculating the number and/or the extent of undesired
events that occurred during the evaluation period (minimum duration period being one year)
without providing any support for diagnosing the cause of events or for defining necessary
corrective action to be taken.
On the whole, this approach has to be considered remarkably useful for internal evaluation by
the manager who, using indicators, is able to keep the results of his activity under control, and for
external evaluation by the people who has to invigilate manager’s work. The availability of easily
applicable indicators which enable the measurement of service performance, referring to various
aspects and comprehensible for all those involved, strengthens and renders more efficient the
representation of public interest in the relationship between Public Authority and system manager.
In fact, using appropriately chosen PIs, Public Authority can be more concrete expressing, also by
using formulations with contractual significance if needed, what it expects from a service and can
follow the evolution of quality over time, highlighting improvements achieved and/or any drifts
requiring change in direction or intervention.

3
PERFORMANCE INDICATORS FOR THE PLANNING, THE DESIGN AND THE MANAGEMENT OF WATER SUPPLY SISTEMS

Furthermore, by using the same PIs in different territories, it is possible to compare different
managers and stimulate competition. Lastly, PIs bring to light improvements in the service which
are difficult for users to perceive (such as those relating to the renewal of underground networks
with subsequent reduction of water losses), thus also allowing any subsequent rate increases to be
justified.
There are, however, certain problematic aspects regarding essentially the reliability and
accuracy of data, the quality of which greatly depends on the diligence with which the database is
managed. It must be remembered that the IWA recommends that for each piece of data, not only
must the reliability of the source be evaluated, but also the level of uncertainty must be ascertained
and that propagation is evaluated in the indicator.
Another problem with regards to the interpretation of values obtained is that interpreting can be
difficult due to the fact that the reference values are not always easy to define, especially in the
presence of non homogeneous externalities1. Added to this is the difficulty encountered when
converting the indices, of which there are many, into a global evaluation (Liserra et al., 2007).
Lastly, it is important to underline that assessment procedures, which presently concentrate on
measuring efficacy and efficiency of services, need to also take into consideration other aspects of
general interest such as the environmental sustainability of the system subject to evaluation
(Paoletti et al., 2007).

3. PERFORMANCE ANALYSIS OF WATER SUPPLY SYSTEMS

A second development line in performance analysis in the water supply field involves the
methodologies used to make a technical evaluation of water system performance (no longer that of
services).
The main performance that one expects from a water system is that, in all conditions that can
reasonably occur during its working life, it supplies good quality water, in sufficient quantity to
satisfy demand and with sufficient pressure to guarantee delivery2. Further on, the other
performances that must be guaranteed will be addressed but this, without doubt, is the principle
one.
The aim of performance analysis is to evaluate if, and how much, the performance requirement
is achieved. Since aspects related to water quality are generally treated separately from the others,
reference will be made to the analysis of hydraulic aspects only, the likes of which can be carried
out using two different approaches.

1 The non homogeneous externalities are those referring to factors which cannot be controlled by manager (for example:
altimetry of the area, population density, ground characteristics, etc.).
2 This definition, indeed somewhat generic, can be specified with reference, for example, to the standards on service levels
guaranteed by the “services chart”.

4
C. CIAPONI – Performance analysis in water supply

The first one is deterministic and serves to check if the desired performance is reached by
referring to a limited number of operative working conditions (project conditions). If the system
performs as desired for all the pre-determined project conditions, analysis results give an adequacy
response; in the case of inadequacy, the analysis provides the diagnostic elements to indicate
corrective actions to be taken.
The second one is the probabilistic approach and evaluates the functioning of a system with
reference to many different working conditions which are not usually taken into consideration on a
design level, but do however have a certain probability of occurring. Statistical elaboration of
results enables an evaluation of the extent to which performance requirements are met in
probabilistic terms; the probability value/s obtained provide us with a measure of the system
performance efficacy.

3.1 THE DETERMINISTIC APPROACH

The most common and consolidated way to carry out performance analysis of a water supply
system (or one of its components) is based on the deterministic approach described previously. It
involves the evaluation of the investigated system’s capacity to satisfy the users’ demand under
well-defined conditions.
For example, in the case of a water distribution network, conventional performance analysis
generally consists of its hydraulic analysis, supposing total availability of the components and
assuming peak demand on the day of maximum consumption. Besides this, hydraulic analysis is
often also carried out with the network in fire-fighting mode and, sometimes, in conditions of a
main pipe being out of service.
If all these tests result positive, the network is considered suitable. For networks that result
unsuitable, corrective actions to be taken are chosen according to the deficiencies that analysis
identifies.
This approach, whilst used a great deal, has some limits. First of all, the scenarios considered
are limited in number, and since it is not possible to attribute them with a probability value, they
assume, above all, a conventional significance. Furthermore, analysis can label a system adequate
or inadequate but it is not able to grade to what extent using a scale that permits the different
solutions or systems to be compared.
These limits are particularly emphasized in the scientific works of authors who propose
methods based on the alternative approach (probabilistic). Nonetheless, one cannot ignore the fact
that deterministic analysis is the one which is presently used in all real applications and that the
extensive experimentation means that, if applied in correct way, using suitable safety coefficients,
it gives substantially robust results.

5
PERFORMANCE INDICATORS FOR THE PLANNING, THE DESIGN AND THE MANAGEMENT OF WATER SUPPLY SISTEMS

3.2 THE PROBABILISTIC APPROACH

One of the first contributions to the solution of measuring the performance of a water supply
system was given in the early 80s by Hashimoto et al. (1982). According to these authors, the
efficacy of a system can be quantitatively measured starting from the failures that it experienced
during its working life (failure meaning any inability to perform to the standards guaranteed to the
users). In particular, the efficacy of the system is defined and can be measured using the following
three concepts:
1) Reliability: measures the probability that the system does not fail.
2) Resilience: measures the rapidity with which the system resumes full operation following a
breakdown.
3) Vulnerability: measures the probable damage subsequent to a failure.

Consider, for example, a water distribution system where Qr(t) represents the volume required
and Q(t) is the volume actually delivered, both functions of time.
Consider the system in a state of failure if Q(t) < Qr(t) and over the course of its working life
(duration T) the system experiences M periods of failure.
Figure 1 represents the j-th state of failure of duration dj during which the system is
characterized by a supply volume deficit Wj:

t1  d j
Wj   t1
[Qr (t )  Q(t )] dt (1)

In this situation, the three indicators of Hashimoto et al. (1982) are defined as follows:

1) Reliability: is the probability that state S of the system corresponds to a condition of non failure
NF, or:

Re l  P S  NF  (2)

therefore, in the given case:

d
j 1
j

Re l  1  (3)
T

2) Resilience: is the conditioned probability that at time (t) state S of the system corresponds to a
condition of failure F whereas at time (t+1) it corresponds to a condition of non failure NF or:

6
C. CIAPONI – Performance analysis in water supply

Re s  PS ( t  1 )  NF ; S ( t )  F  (4)

In the case given, it can be defined as the inverse of the average duration of failure (Kjeldsen &
Rosbjerg, 2004):

1
1 M 
Re s  
 M
 j 1
dj

(5)

3) Vulnerability: is the sum of damages hj associated with each state of failure, each one multiplied
by its probability ej :

M
Vul  e
j 1
j hj (6)

In the example given, assuming water deficit Wj as the damage value associated with each state of
failure and assuming, for simplicity, the same probability for each state of failure occurrence (e1 =
e2 …= eM = 1/M), you obtain:

W
1
Vul  j (7)
M j 1

Q
PORTATA

Qr

wj

dj

t1 j t1 j + d j TEMPO

Figure 1 - Representation of j-th failure interval of a distribution system (Kjeldsen &


Rosbjerg, 2004)

7
PERFORMANCE INDICATORS FOR THE PLANNING, THE DESIGN AND THE MANAGEMENT OF WATER SUPPLY SISTEMS

The system subject to evaluation is therefore easily characterized by three indices which
measure its performances in a global sense. In the case of a distribution network, if the functions
Q(t) e Qr(t) were known for each delivery node, the same indices could be calculated for each
node. Therefore, a measure of performance is also obtained at a local level, which is especially
useful to identify zones characterized by most risk.
In reality, it is not so simple as, leaving now aside the difficulty involved in estimating the
volumes Q(t) and Qr(t), the definition of the three indices given above are based on the assumption
that the system has only two states (state F for failure and state NF for non failure) which are
defined by a single threshold value. These indices, therefore, do not account for the partial
functioning of a system; simply because demand is not fully met, does not mean that the level of
service is zero 3.
Furthermore, it is simple to recognize that the three index values depend on the threshold value,
to a varying extent, from case to case (and therefore in an unpredictable way). Slight variations in
the threshold, which are insignificant on a practical scale, can in fact determine very sensitive
variations in the three indices. The last two indices variations can be either positive or negative
depending on the case. This makes interpreting the measurement problematic, and deciphering it is
further complicated by the contemporary use of three indicators.
The Hashimoto et al. (1982) evaluation method, despite being frequently cited by many other
authors4, has been surpassed by other evaluation criteria which have sought to curtail the
measurement of performance using one single concept, which is generally identified as reliability.
Moreover, evermore differing meanings have been ascribed to the term reliability, overshadowing
the original definition given above. The concepts of resilience and vulnerability have been
practically abandoned, even if the measurement of failure repercussions (vulnerability) has often
been englobed by the definitions ascribed to the concept of reliability5.
Lastly, it is important to note that performance analysis must be applied to the entire water
system, including all its components, from captation to distribution. In reality, the study of
performance measurement in most cases focuses attention on the single component and, in

3 According to Tanymboh et al. (2001), each definition that is not able to quantify the partial failures is not suitable for
use in water system performance analysis. It is therefore necessary to carry out an evaluation which pays total respect to
the standard, even an auxiliary evaluation which considers only the significant reductions in service levels to be
unsatisfactory, where the term significant refers to the extent and temporal spatial incidence of the variance (Bertola
Rejtano, 2004). For this reason it was suggested that the concept of reliability, intended as the probability that the system
functions correctly (respecting its mission) in a given period of time, were to be completed with a measurement of spatial
and temporal range of the failures (Walski, 1987) and the consequence of failure in terms of the extent of water volumes
required and not supplied. (Xu, 1990).
4 Various authors (amongst others La Loggia e Mazzola, 1989) have proposed several indices based on the concepts
formulated by Hashimoto et al (1982).
5 It has also been observed that resilience has not to be explicitly taken into account, despite its strict correlation with
vulnerability (Kjeldsen & Rosbjerg, 2004).

8
C. CIAPONI – Performance analysis in water supply

particular, on the distribution network. It is not clear if this is due to the fact that the distribution
network is the weak link in the whole drinking water system, where the functional problems most
perceptible to the users most commonly arise, or because the network with its topological
complexity and the wide variety of working conditions to which it is subjected presents arduous
calculation problems and is therefore more stimulating for researchers.

4. RELIABILITY INDICATORS BASED ON DELIVERED VOLUME DEFICITS

For some time there have been three important questions regarding the concept of reliability
applied to water distribution systems: what is its definition; how must it be measured; what values
can be considered as acceptable. In the not too distant past, various authors held considerably
different positions regarding these issues, but in recent years a significant convergence on certain
criteria that seem to prevail over others is visible.
For example, many authors define system reliability as the capacity to meet the users’ needs
under all operative conditions that can arise during a system’s working life. Furthermore, it seems
that the measurement of reliability is consolidated by way of indices based on the relationship
between volumes actually delivered WE during the evaluation period and the volume WR required
by the users6 (amongst others: Gupta & Bhave, 1994; Gargano & Pianese, 2000; Tanymboh et al.,
2001; Bertola & Nicolini, 2004):

WE
R (8)
WR

This definition, which can be applied on a global scale or on a local scale (at the single delivery
node) has the advantage that other indicators do not; it has a simple physical significance which is
immediately referable to the performance level experienced by the users.
Furthermore, in an industrial context, these types of indices allows manager to evaluate how
much more water could be sold if the network were to work better.
For some authors (Bertola & Nicolini, 2004; Gupta & Bhave, 1994; Tanymboh et al., 2001)
the reliability index coincides with the above mentioned relationship, calculated over a long period
of time. With this formulation, the concept of reliability loses its strict probabilistic significance
and assumes a significance closer to that of performance index.
Other authors (Gargano & Pianese, 2000) continue to try and give a probabilistic significance
to the concept of reliability, defining it as the probability that the supply deficit does not exceed a
pre-fixed threshold value.

6 A closer look at the sustainable use of water resources shows that the water volume required (demand), which in its
current meaning also includes waste and improper use of water, can assume the meaning “sustainable maximum water
endowment”.

9
PERFORMANCE INDICATORS FOR THE PLANNING, THE DESIGN AND THE MANAGEMENT OF WATER SUPPLY SISTEMS

The reliability indices can be formulated differently depending on the adopted approach.
Following are presented some possible formulations which, according to a fairly diffused
procedure, are expressed in terms of flow7.

4.1 STATIONARY DEMAND AND DIFFERENT POSSIBLE WORKING STATES

For an assigned distribution network, subject to different possible working states as far as the
availability of its components is concerned (each working state is defined by index k with k=1, NS
where NS is the number of working conditions analyzed), consider a single demand configuration
(stationary demand) defined by the value of volume Qrj required at each node (each node is
identified by index j with j = 1, NN where NN is the number of nodes).
Qj is the flow actually delivered at the j-th node, evaluated as a function of the head at the
node (Qj = Qrj if the nodal head is greater or equal to that necessary to meet the full demand;
0<Qj<Qrj if the nodal head is less than that necessary to meet full demand).
With these assumptions, the following indices can be formulated:

Q j ,k
R j ,k  local indicator for working condition k-th (9)
Qr j

NN

R
j 1
j ,k Qr j
RRk  NN
global indicator for working condition k-th (10)
 Qr
j 1
j

 R 
NS
Rj  j ,k wk local indicator for all working conditions (11)
k 1

NS
RR   RR
k 1
k wk  global indicator for all working conditions (12)

7 The indices expressed in terms of flow maintain the significance of (8) based on volume deficits, on condition that the
temporal intervals assumed in the calculation are constant.

10
C. CIAPONI – Performance analysis in water supply

In (11) and (12), wk is the weight assigned to the value of the k-th working condition; this can
coincide with the probability values associated with each working condition taken into
consideration.

4.2 TIME-DEPENDENT DEMAND AND FIXED WORKING STATE

If one considers, for a fixed working state of the system, a demand which varies over time in
such a way defined by means of n flows (Qrj) configurations required at the nodes (each defined
for the i-th interval of time (i = 1, n) of duration ∆t (with ∆t constant for the n intervals
considered), the performance indices assume the following formulations:

Q j ,i
R j ,i  local indicator for i-th interval (13)
Qr j ,i

Q
i 1
j ,i
Rj  n
local indicator for the entire period considered (14)
 Qr
i 1
j ,i

NN

R
j 1
j ,i Qr j ,i
RRi  NN
global indicator for i-th interval (15)
 Qr
j 1
j ,i

NN n

 Q
j 1 i 1
j ,i

RR  NN n
global indicator for the entire period considered (16)
 Qr
j 1 i 1
j ,i

11
PERFORMANCE INDICATORS FOR THE PLANNING, THE DESIGN AND THE MANAGEMENT OF WATER SUPPLY SISTEMS

4.3 TIME-DEPENDENT DEMAND AND DIFFERENT WORKING STATES

The third situation examined combines the two previous and proposes the evaluation of the
performance indices of a network subject to NS different possible working states as far as the
availability of its components is concerned (each working state is defined by index k with k=1, NS
and has an associated probability equal to wk) and is characterized by time dependent demand,
defined as in the previous paragraph 4.2.
Repeating calculations in (14) and (16) for each of the NS mechanical working conditions of
the network, the indices assume the following formulations:

Q
i 1
j ,i ,k
R j ,k  n
for each node: local indicators ( number equal to NS) (17)
 Qr
i 1
j ,i

NN n

 Q
j 1 i 1
j ,i ,k

RRk  NN n
for whole network: global indicators (number equal to NS) (18)
 Qr
j 1 i 1
j ,i

Attributing to each working condition a weight wk equal to its associated probability, it is


therefore possible to calculate the following indices:

 R 
NS
Rj  j ,k wk local indicators for all working conditions (19)
k 1

NS
RR   RR
k 1
k wk  global indicator for all working conditions (20)

4.4 SUPPLEMENTARY INDICATORS

It is easy to realize that the indicators presented above cannot fully illustrate the reliability of a
system in that, as highlighted in Table 1, water deficits of different severity in relation to the spatial
and temporal distribution of failures can determine the same index values.

12
C. CIAPONI – Performance analysis in water supply

Table 1- The incapacity of indices to measure the severity of water deficit

Level of Deficit % % Global Local index


severity (Qr-Q)/Qr time nodes index

Low severity 10% 100% 100% 0.9 0.9


Severe 100% 10% 100% 0.9 0.9
Very severe 100% 100% 10% 0.9 0 for 10% nodes
1 for 90% nodes

It was therefore suggested that the following supplementary indices should be introduced (Gupta &
Bhave, 1994):

1) Temporal factor: is the ratio between the duration of acceptable situations and the total
considered duration:

NS NN n

  
k 1 j 1 i 1
j ,i t
Ft  (21)
NS NN T

- =1 if Qj,i / Qrj,i  acceptable value (for example 0.5);


- =0 if Qj,i / Qrj,i < acceptable value
- T = total duration of considered period

2) Nodal factor: corresponds to the geometric mean8 of the nodal indices;

1 / NN
 NN 
Fn   
 j 1
Rj 

(22)
 

8 As is known, one property of the geometric mean is that small values are far more important than big values, and the
presence of a single null value is sufficient to render null the mean.

13
PERFORMANCE INDICATORS FOR THE PLANNING, THE DESIGN AND THE MANAGEMENT OF WATER SUPPLY SISTEMS

Therefore, the global performance index RR can be corrected as follows:

RRc = RR Ft Fn (23)

Applying corrective coefficients to the cases indicated in Table 1 enables the following
performance indices to be found for the three different levels of severity: low severity cases: RRc=
0.81; severe cases: RRc= 0.73; very severe cases: RRc= 0.

5. RELIABILITY CALCULATION

Calculating reliability ex ante according to the definition given above requires to define, by
means of a mathematic model, failure states which the system will be subjected to during its
working life. This essentially involves statistical elaboration and evaluation of the performance
indices (generally defined as a function of supply deficits) which are calculated using hydraulic
simulation of the system starting from the different functioning scenarios, each characterized by
probabilistic point of view. The statistical elaboration (which could also consist of a simple
average) of these indices leads to the calculation of reliability.
To define the different scenarios, are taken into account, considering their probabilistic
distribution, the following factors, because of which system failure may occur:
- Mechanical factors such as break downs, interruptions of electrical supply or some system
component being out of order (pipes, pumps, valves, etc.);
- Hydraulic factors such as random variations (in terms of both time and space) in demand
and/or reduction of hydraulic conductance in the pipes due to age.
It must be noted that, since calculations seek to value, in order to arrive at a final evaluation,
the various performance indices based on the probability of the failure factors which determined
them, only the failure factors which can be expressed in probabilistic terms can be dealt with.
However, uncertainties about which probabilistic characterization cannot formulate reasonable
hypotheses cannot be included in the calculation.
Literature proposes many approaches and they are distinguished by the typology of failure
causes considered: some take into consideration only mechanical causes, others only hydraulic
causes and then there are those which seek to incorporate both types.
The proposed calculation techniques also vary greatly. They are mainly based on simulations
over long time periods (even 50-100 years) and are generally carried out with hourly resolution; in
this approach, the elements affected by aleatority or uncertainty are stochastically simulated using
Monte Carlo techniques. The number of simulations required for convergence of the M.C.
techniques are rarely stated and the application examples given in literature always refer to
networks consisting of a very small number of components.

14
C. CIAPONI – Performance analysis in water supply

In many works, by way of suitable simplifications, the aim is to formulate calculations in such a
way that the number of hydraulic simulations is significantly reduced and attempts are made to lead
a part of the problem to closed-form to be coupled with the numerical simulations (Xu & Goulter,
1998; Salandin & Darvini, 2007).
There are certain crucial aspects concerning the analysis which will be addressed in the
following paragraphs.

5.1 SYSTEM HYDRAULIC SIMULATION

To correctly calculate the reliability indices according to the definition described above, it is
necessary to evaluate the actual flow delivered in the cases that the pressure at one or more nodes
is less than that needed for the delivery of the required flow. It must be remembered that the
calculation models presently used for network hydraulic analysis set for each node (index j), the
delivered flow Qj (problem datum), assuming it equal to the required flow Qrj, with the implied
hypothesis that the head Hj (unknown element) is sufficient to satisfy it. These models, defined in
literature as DDA (Demand-Driven Analysis), give correct results only when the hydraulic analysis
of the network is positive that is when, for each node, Hj is equal to or greater than the required
head Hrj to meet the full demand. If however calculation reveals critical nodes for which Hj < Hrj,
the results given by the model are not correct, as the flows Qj assigned as delivered from the
critical nodes are not compatible with the Hj values which result from the calculation.
To date, this conventional approach has been considered satisfactory because the aim of the
hydraulic analysis has always been to validate the network design and to underline the need for
correction in the case of negative outcome. The need to calculate, in order to evaluate the
reliability indices, the actual delivery also in cases of insufficient pressure involves, however, a
different approach which is referred to in literature as PDA (Pressure-Driven Analysis). PDA
allows to determine the solution which satisfies not only the momentum and continuity equations
but also the Qj=f(Hj) equations which link the flow delivered and the head actually available at
each node.
This poses two rather complex problems.
The first one concerns the definition of the link between delivered flow Qj and pressure hj
(directly linked to the head Hj) at each node. In fact, this link depends on a variety of factors
amongst which, particular important, are the configuration and the size of the secondary network
fed from the node and the spatial distribution of the delivery devices.
Evaluation of these elements, which can vary greatly for different nodes, would require detailed
information which is practically never available in the normal diagnostic activity of networks.
Consequently, the link Qj = f (Hj) can be introduced in hydraulic analysis only using general,
approximated schemes.
In general, the flow actually delivered is expressed as a function of the required flow Qrj, by the
relationship (Gupta & Bhave, 1996):

15
PERFORMANCE INDICATORS FOR THE PLANNING, THE DESIGN AND THE MANAGEMENT OF WATER SUPPLY SISTEMS

Qj = j Qrj (24)

where j is between zero and one, in relation to the pressure value.


The most diffused scheme sets two threshold values for the nodal head:
- Hminj = nodal head value below which delivery is nil; this value can be assumed to be equal
to the zj ground level or equal to 23 m above it;
- Hrj = head value required to meet demand Qrj
and assumes the following relationship:

j = 1 for Hj  Hrj (25)

j = 0 for Hj  Hminj (26)

0 < j < 1 for Hminj < Hj < Hrj (27)

For the definition of j values according to (27), literature proposes different expressions
(Wagner et al., 1988; Fujiwara & Ganesharajah, 1993; Fujiwara & Li, 1998; Tucciarelli et al.,
1999) that may give very different evaluations, without any rational criterium to support the choice.
Another problem which arises regards the numerical solutions involved in the PDA approach.
According to the formulations recently proposed in literature (Todini, 2003; Cheung et al.,
2005; Giustolisi et al. 2007), it seems that the main approach is to solve the entire system of
momentum and continuity equations, integrated with equations representing the relationship
between delivered flow and pressure at node. Nonetheless, at present, the numerical methods for
system solution have not been sufficiently tested, especially for the aspects related their
convergence, which could result problematic because of the special formal structure for the
relationship Qj = f (Hj). Despite the availability and diffusion of efficient software packages based
on DDA analysis, various authors adopted calculation procedures which, whilst formulated
according to PDA logic, can use the conventional DDA hydraulic solvers already available.
Amongst these procedures, worthy of mention are those that simulate the critical nodes as fixed
head and unknown flow reservoir-nodes (Bhave, 1981; Ozger & Mays, 2003; Todini, 2003) and do
not take into account the relationship Q=f(H). However, some example calculations provided by
Ardenti et al. (2007) show that these procedures, due to the fact that they do not consider the bond
between delivered flow and available head at node, give improbable results in some cases.
However, other procedures which even though they avail of already consolidated hydraulic
solvers, include the relationship Qj=f(Hj) at each node, seem more reliable. One of the most
interesting of these procedures is the one based on an option offered by EPANET-2 (Rossman
2000) which allows to model water delivery by particular devices (emitters) the outward flow of
which is linked to pressure by a fixed relationship.

16
C. CIAPONI – Performance analysis in water supply

Ardenti et al. (2007) show that the EPANET-emitters procedure, advantageous for its
simplicity, gives results which are practically superimposable with those arrived at using an
iterative procedure which gives, with small approximations, the exact solution.
Ardenti et al. (2007) have also highlighted, in the case of networks characterized by pressure
deficiency at various nodes, the considerable errors connected with conventional hydraulic analysis
or with simplified procedures which seek to evaluate the flow actually delivered starting from
piezometry calculated with a DDA approach. This casts serious doubt on many of the works
presented in literature, which, in order to calculate network reliability according to pressure and/or
water volume deficits, refer to DDA simulations.

5.2 PROBABILISTIC ANALYSIS FOR MECHANICAL FAILURES

Among the most important aleatory factors that affect system performance is the temporary
unavailability of one or more system components due to mechanical failure (breakdown of
electromechanical equipment, out of order piping awaiting repair, etc.).
Many authors consider only piping related failures in reliability evaluation. Su et al. (1987)
have further underlined that the probability of simultaneous unavailability of two or more pipes is
very low and therefore the results, in terms of reliability, vary only slightly when considering the
unavailability of only one component, or the combined unavailability of more than one component.
So, in many works, mechanical failure is identified as the unavailability of only one pipe a time in
the network. Because of the fact that pipes are repairable components, many authors (Khomsi et
al., 1996; Gargano & Pianese, 2000; Tanymboh et al., 2001) carry out the probabilistic analysis of
mechanical failures on the basis of the availability concept which encompasses both the probability
of breakdown and the time required for repair. Availability Ai, defined for each i-th component as
the probability that it is available when needed, is assessable with the equation:

MTTFi
Ai  (28)
MTTFi  MTTRi

where MTTF is mean time to failure and MTTR is mean time to repair9.
The complement to one of the availability Ai is the unavailability Ui, that is the probability that
the component is not available:

9 The MTTF and MTTR values are easily defined once the failure rate  (annual number of breakages per unit of length
of pipe) and the repair rate i (the inverse of mean repair time) are known: MTTFi = 365/(i Li) and MTTRi = 1/i

17
PERFORMANCE INDICATORS FOR THE PLANNING, THE DESIGN AND THE MANAGEMENT OF WATER SUPPLY SISTEMS

MTTRi
U i  1  Ai  (29)
MTTRi  MTTFi

Once Ai and Ui, have been established, calculation of the probabilities associated with the
different working states of the network is immediate.
The probability that no pipe of the network is unavailable and therefore the network runs with
all its pipes active, is given by the following expression:

NT
p0    Ai (30)
i 1

The probability that pipe f, and only pipe f, is unavailable is:

Uf
p  f   p 0  (31)
Af

The probability values given by (30) and (31) can be used as weight coefficients wk to suitably
mediate the performance indices obtained for the different working conditions of the system in
equations (11), (12), (19) and (20).
An alternative approach, used by some authors (Salandin & Bertola, 1996; Bertola & Nicolini,
2004; Salandin & Darvini, 2007) involves establishing the probability density function (PDF),
failure and repair times and applying Monte Carlo techniques in order to define some possible
sequences (sample cases) of different working states (regular or with mechanical failure) and
continually simulate (continuous simulation) the associated hydraulic behaviour. Nodal and global
reliability is then calculated using the statistical analysis of the results obtained from a suitable
number of sample cases.
For both approaches, the failure rate  and repair rate  are required in order to define MTTF
and MTTR or for the parametrization of the failure10 and repair times11 probability distribution.
Usually, the failure rate  is assumed constant in time as described by the well-known “bath-
tub” curve which illustrates how, after a brief initial period in which the component can experience
construction defects that did not appear during inspection, the failure rate drops to values that
remain more or less constant for most of its life (during which breakdowns are essentially
aleatory). The rate then increases rapidly during the old age period due to breakages caused mainly
by wear and tear.

10 Usually assumed exponential


11 Usually assumed exponential or uniform

18
C. CIAPONI – Performance analysis in water supply

As is easily intuitable, the breakage rate depends on numerous factors which are also often
associated with local and very particular situations. Values published in literature range from a
minimum of 0.05 breakages/(km year) to a maximum of 1 breakage/(km year) 12 with a large
dispersion (especially big for the smaller diameters). Nonetheless,  generally decreases with
diameter.
To calculate reliability, the breakage rate  is often estimated using regression formulae which
link it to diameter D of the pipe. Among these, the most frequently used are those proposed by Su
et al. (1987) deriving from the analysis of data concerning the city of St. Louis:

16192 118 183558


  0,0261  3, 26
 1,3131
 (32)
D D D 3,5792

( is the annual number of breakages per kilometer and D is expressed in millimeters) and that of
Mays (1989) which directly gives a value for MTTF:

MTTF  0,21218 D1.462131 (33)

(MTTF is expressed in years and D in inches).


Studies regarding the Italian situation are fairly scarce. Among them is the analysis of the repair
interventions carried out on Verona’s water network (Bertola & Pavia, 2002), which can be
interpreted by:

  2,08546  0,854936 log D (34)

Another example is the analysis for the Tricarico network (Ermini et al., 1998), the data of which,
elaborated together with that of St. Louis, gave rise to the following relationship:

  116 D 1,05 (35)

In both expressions,  is the number of annual breakages per kilometer and D is expressed in
millimeters.
In other works, data regarding breakages have been used to analyze and experiment
probabilistic methods for their modeling (Alvisi et al., 2007).
Other than the uncertainty associated with the fact that  depends on factors linked to local
situations, it must be highlighted that to evaluate reliability of a network with respect to mechanical

12 According to Pelletier et al. (2003), it can be considered that a network is in good condition when   0.2, in
acceptable condition when 0.2 <  < 0.4 and in bad condition when   0.4. According to Sundal (1997), the breakage
rate is low when   0.08, normal when 0.08 <  < 0.16, high when 0.16 ≤  < 0.28, very high when   0.28.

19
PERFORMANCE INDICATORS FOR THE PLANNING, THE DESIGN AND THE MANAGEMENT OF WATER SUPPLY SISTEMS

failures, only repair interventions which require a pipe to be out of service should be taken into
consideration. The analysis (carried out by the author) of quite 2500 data referring to repairs
executed between 1998 and 2005 on a water supply system of about 700 km in the Pavia
Apennines (Italy) showed that the majority of repair interventions carried out (68% on steel pipes;
66% on PE pipes and 39% on PVC pipes) did not require isolation of the pipe in question. The use
of the expressions given above, derived with reference to all repairs, is therefore not correct to
calculate reliability.
As far as the average repair time MTTR is concerned, the examination of (29) and (31) shows
that the probability of pipe failure is directly proportional to its unavailability U which, in turn,
depends almost linearly in its average repair time MTTR. Therefore, the estimate of MTTR is
notably important when evaluating reliability in connection with mechanical failures, as Walters &
Knezevic (1988) have already point out. However, data referring to repair times is scarce and the
MTTR value is generally assessed by assigning indicative values. The most frequently accepted
value is equal to 1 day (Khomsi et. al, 1996; Gargano & Pianese, 2000; Shinstine & Lansley,
2002). Other authors (amongst others: Bertola & Nicolini, 2004) consider the repair rate ( =
1/MTTR) a random variable described by a uniform or exponential probability distribution. In
reality, it would seem reasonable to admit that a repair time which requires a pipe to be out of
service, largely depends on its diameter. A relationship obtained by the regression of some
experimental data was proposed by Walski e Pelliccia (1982):

MTTR = 6,5 D0,285 (36)

where MTTR is expressed in hours and diameter D of the pipe is expressed in inches.

5.3 WATER DEMAND SIMULATION

The characterization of demand and the evaluation of associated uncertainties is of considerable


importance for water distribution system reliability assessment.
As indicated by numerous studies including those covering Italian cases (Molino et al., 1986;
Lamberti et al., 1994), water demand is influenced by a multiplicity of dynamic factors over the
course of time. The effect of the factors can be detected in a long period trend (mainly depending
on the social-economic development of the users), in seasonal cyclicity (linked to climatic aspects
which affect water usage, economic factors such as tourism and seasonal production, etc.) and
daily cyclicity (mainly associated with the users’ daily rhythms and habits). Besides the structural
components of the consumers’ diagram, there is a considerable aleatory component intrinsic to the
nature of a service on demand.
Generally, only daily cyclicity, defined using a set of average hourly values, is taken into
account when running simulations to calculate reliability; sometimes, the aleatory component is

20
C. CIAPONI – Performance analysis in water supply

also considered13 by introducing a stochastic forcing with an assigned probability distribution,


defined simply by means of exemplifying hypotheses without the support of any experimental
investigation.
The demand calculated for a whole system is generally divided among the various delivery
nodes proportionally to the extent of the respective supplied users. Therefore, demand at each node
has a constant impact as regards the entire network demand at the different evaluation time
intervals.
In the rare cases in which aleatority is considered at the single nodes independently from the
other nodes (to take into account the demand’s spatial variation), a correct, not just exemplifying,
statistical characterization is required as the local aleatorities have a greater impact than those
which characterize global demand. Here arises difficulty in defining the independent local aleatory
demands; whilst independent, their sum must however be consistent with the statistical
characterization of the global demand.
Rarely, because of the probability is difficult to evaluate, demand model considers cases in
which the network has to satisfy the needs of the fire-fighting service.

5.4 THE MEASUREMENT OF OTHER FUNCTIONAL ASPECTS

The evaluation of system reliability described in the previous paragraphs is oriented towards
measuring the systems ability to supply the users with enough water to satisfy demand and provide
sufficient pressure to guarantee delivery in all conditions that could arise during its working life.
In the case of evaluating a project or existing system, it would also be useful to measure
performances that are not directly perceived by the users but are, however, important for the
efficient functioning of a system.
For example, Pianese & Villani (1994 a, 1994 b) proposed indices aimed at measuring system
adequacy from a hydraulic point of view, making particular reference to aspects related to excess
speed in pipe (inducing vibrations which are damaging to the joints) and excess oscillation of the
nodal head (which, when frequently repeated, stresses the joints).
Another important performance indicator, never mentioned in literature, should measure the
probability of pressure dropping below the minimum safety values indicated in order to avoid
depression in pipe. The indicator should also measure the temporal and spatial range of this
situation.
Other indicators which are especially useful for comparing different solutions should measure
energy efficiency in terms of dissipated power compared to the power delivered.
Finally, for water systems working in conditions of scarce resources, there are the fair
satisfaction of water demand and fair access to the resource indicators (La Loggia et. al, 2005)

13 This obviously involves generating a high number of demand scenarios (which, in some formulations, are verified for
each mechanical failure situation examined) with a considerable calculation increase.

21
PERFORMANCE INDICATORS FOR THE PLANNING, THE DESIGN AND THE MANAGEMENT OF WATER SUPPLY SISTEMS

which aim to measure the fairness with which a distribution network distributes the resource when
fed with flows inferior to demand.

6. WATER DISTRIBUTION SYSTEM PERFORMANCE WITH REFERENCE TO


QUALITATIVE ASPECTS

The aspects related to the links between the quality of water distributed and the network
hydraulic behaviour are extremely complex and have not been studied so much (as is often the case
for topics with strong interdisciplinary features), even though, since a few yars, they are receiving
increasing attention.
In general, there should be some changes of water quality in the distribution network due to
phenomena such as: precipitation and flocculation of some substances, decay of disinfectant agents
with subsequent bacterial growth, corrosion of pipe walls, infiltration of water from outside in
cases of depression, all phenomena whose negative effect increases the longer the water remains in
the network.
Particular important amongst these phenomena is the development of biofilm on the pipe walls
Due to the presence and persistence of different microbial species, it is a crucial problem when
controlling the quality of drinking water. As recently highlighted by researchers at the Italian
Istituto Superiore di Sanità (Bonadonna & Della Libera, 2005), the extent of biofilm development
is conditioned not just by the availability of nutrients but also by the time spent in the network and
the water temperature.
It is therefore important that water spends as little time as possible in the network. In terms of
network hydraulic functionality, this means that the speed with which water travels through the
network must be sufficiently high, and the passage of the water from the feeding points to the
delivery nodes must be reduced to the minimum.
Moreover, it is worth remembering that this last criterion, suggested in various classic Italian
texts (Marchetti, 1949; Ippolito, 1960), shaped the design of water distribution networks for many
years, at least in Italy. At present, however, this criterion does not seem to have space in the new
design methods which pay more attention to aspects of reliability.
It has recently been highlighted that, in networks built according to the conventional criterion
of minimizing construction costs, the circulating water configuration for the assigned project
demand respects, with good approximation, the minimum distance covered by water within the
network criterion (Ciaponi & Papiri, 2007). This would seem to revalue, at least in part, design
methods based on purely economic optimality criteria which, bringing to such a diameters size
which favour minimum time that water spends in the network, imply respecting a reliability
criterion as far as water quality is concerned.

22
C. CIAPONI – Performance analysis in water supply

7. CONCLUSIONS

The evaluation of water system performance is a topic of great interest, both from a technical
and scientific point of view, which has important repercussions on integrated water service
management. Given the significance of repercussion to the quality of citizens’ life and the
economy, when examining the different methodologies proposed, attention must be focused on
three important aspects: conceptual quality of the performance evaluation in terms of
methodological correctness and completeness of analysis; the applicability of the measurement
method; the capacity to translate the measurement into a correct and objective judgment.
The considerations developed in this report highlighted that the present situation is
characterized by the availability of different evaluation instruments which, however, are not always
completely satisfying and present much room for improvement.
The performance analysis methodologies such as the IWA-PI types are characterized by a
suitable conceptual quality and good applicability. Furthermore, they are extremely useful both for
evaluations internal to the management organization and for external assessment by supervising
public authority.
Currently focused on measuring efficacy and efficiency of water services, these evaluation
systems can be developed as planning activity support by introducing aspects of general interest
such as energy efficiency and environmental sustainability. The limits are linked mainly to the need
for a completely reliable database and to the interpretation of results, which is not always easy and
objective.
Water system performance study by way of conventional hydraulic analyses based on a
deterministic approach gives good conceptual quality and optimal applicability in the context of
real cases. However it is difficult, with this methods, to translate the results into measurements (and
hence judgments) regarding the performance efficacy of the systems evaluated.
For some years now, the usefulness of performance measurement has promoted the
development of methodologies based on probabilistic approaches. Presently, these methods
generally provide excellent conceptual quality, but they are difficult to apply to cases that are not
simply exemplifying. Methods proposed in literature which are conceptually very interesting but
applied to a network of around ten pipes requires a few days of calculations, are certainly useful for
acquiring more knowledge, but they do not contribute to real progress in the water supply sector.
The idea of quantifying water system performance in probabilistic terms needs to be translated into
methodologies which can be applied to real cases.
Research in water system performance analysis therefore has a long way to go considering also
that paying more attention to the real needs of the water supply sector should also involve going
beyond “centric-network” vision, which to date has characterized research, and lead to a global
approach which looks at important aspects, such as water quality, that have so far been ignored.

23
PERFORMANCE INDICATORS FOR THE PLANNING, THE DESIGN AND THE MANAGEMENT OF WATER SUPPLY SISTEMS

8. REFERENCES

Alegre H., Hirner W., Baptista J.M. & Parena R., (2000) - Performance Indicators for Water
Supply Services - IWA Publishing, London, ISBN 1900222272, pp.160
Alegre H, Baptista J.M., Cabrera E., Cubillo F., Duarte P., Hirner W., Merkel W. & Parena R.,
(2006) – Performance Indicators for Water Supply Services – Second Edition - IWA
Publishing, London, ISBN 1843390515, pp. 289
Alvisi S., Grata S. & Franchini M., (2007) – Analisi e confronto di due modelli probabilistici
per la rappresentazione delle rotture in una rete acquedottistica - Atti del Convegno
“Approvvigionamento e Distribuzione Idrica: Esperienza, Ricerca ed Innovazione”, Ferrara,
28-29 giugno.
Ardenti P, Ciaponi C., Franchioli L. &Papiri S., (2007) – Simulazione idraulica delle reti di
distribuzione idrica in condizioni di pressione insufficiente in uno o più nodi – Atti del 2°
Convegno Nazionale di Idraulica Urbana, Chia (Ca), 25-29 sett.
AWWA, (2004) - Selection and Definition of Performance Indicators for Water and
Wastewater Utilities - ISBN 158321304x, pp.125
Bertola P. & Nicolini M., (2004) – Valutazione dell’affidabilità complessiva di una rete di
distribuzione idrica con simulazioni quasi stazionarie di lungo periodo - Atti del 29°
Convegno di Idraulica e Costruzioni Idrauliche, Trento, pp. 141-148
Bertola P. & Pavia S., (2002) - Reti di distribuzione dell’acqua potabile e tasso di fallanza
delle condotte - L’Acqua, Vol. 1-2, pp 124-130
Bertola P. & Rejtano B., (2004) – Definizioni e misure dell’affidabilità dei sistemi
acquedottistici – La gestione e l’affidabilità dei sistemi acquedottistici (a cura di P. Bertola
e M. Franchini), Ed. Bios, ISBN 88-7740-384-5, pp. 97-108
Bhave P.R., (1981) - Node flow analysis of water distribution systems, Journal of
Transportation Engineering, Vol. 107, No. TE4, pp. 457-467
Bonadonna L. & Della Libera S., (2005) – Problemi di alterazione microbiologica: biofilm e
biofouling (fenomeno, cause, effetti)- Atti del Convegno “Influenza dei sistemi di
distribuzione sulla qualità dell’acqua potabile”, Genova, 18 novembre
Cheung P.B., Van Zyl J.E. & Reis L.F.R., (2005) - Extension of EPANET for pressure driven
demand modelling in water distribution system, International Conference on Computing
and Control in the Water Industry “ Water Management for the 21 st century, Exeter, UK, 5-
7 sept.
Ciaponi C. & Papiri S., (2007) – Dimensionamento ottimale delle reti di distribuzione idrica:
validità ed implicazioni del criterio del minimo percorso dell’acqua - Atti del Convegno

24
C. CIAPONI – Performance analysis in water supply

“Approvvigionamento e Distribuzione Idrica: Esperienza, Ricerca ed Innovazione”, Ferrara,


28-29 giugno.
Coelho S. T., (1997) – Performance in water distribution: a system approach – Research
Studies Press LTD, Taunton, Somerset, England, ISBN 0863802192, pp. 225
Comitato per la Vigilanza sull’uso delle Risorse Idriche, (2006) - Indicatori di prestazione dei
servizi di acquedotto, fognatura e depurazione - Delibera n° 7 del 27/04/06
Ermini R., Viparelli R. & Fiorentino M., (1998) – Una metodologia per la valutazione
dell’incidenza della vulnerabilità meccanica sul disservizio nelle reti acquedottistiche –Atti
del XXVI Convegno di Idraulica e Costruzioni Idrauliche, Catania.
Fujiwara O. & Ganesharajah T., (1993) - Reliability assessment of water supply systems with
storage and distribution networks, Water Resources Research, Vol. 29, No. 8, pp. 2917-
2924
Fujiwara O. & Li J., (1998) - Reliability analysis of water distribution networks in
consideration of equity, redistribution and pressure-dependent demand, Water Resources
Research, Vol. 34, No. 7, pp. 1843-1850
Gargano R. & Pianese D., (2000) – Reliability as Tool for Hydraulic Network Planning -
Journal of Hydraulic Engineering, Vol 126, n° 5, pp. 354-364
Giustolisi O., Savic D.A. & Kapelan Z., (2007) -Un modello di simulazione idraulica delle reti
di distribuzione per la gestione delle perdite, atti del Convegno “Approvvigionamento e
Distribuzione Idrica: Esperienza, Ricerca e Innovazione”, Ferrara, 28-29 giugno
Guérin-Schneider L., (2001) – Introduire la mesure de performance dans la régulation des
services d’eau et d’assainissement en France – Instrumentation et organisation - Thése de
gestion, ENGREF, 447 p.
Gupta R. & Bhave P. R., (1994) – Reliability Analysis of Water Distribution Sistems – Journal
of Environmental Engineering, Vol. 120, n° 2, pp. 447-460
Gupta R. & Bhave P.R., (1996) - Comparison of methods for predicting deficient-network
performance, Journal of Water Resources Planning and Management, Vol. 122, No. 3, pp.
214-217
Hashimoto T., Loucks D. P. & Stedinger J., (1982) – Reliability, resilience and vulnerability
for water resources system performance evaluation – Water Resources Research, 18(1), pp
14-20
Iannelli G., (1967) - Elementi e dati per il proporzionamento delle reti di distribuzione,
Ingegneria Sanitaria, n° 4.
Ippolito G., (1960) - Appunti di Costruzioni Idrauliche - Liguori Editore, Napoli.

25
PERFORMANCE INDICATORS FOR THE PLANNING, THE DESIGN AND THE MANAGEMENT OF WATER SUPPLY SISTEMS

Khomsi D., Walters G.A., Thorley A.R.D. & Ouazar D., (1996) - Reliability Tester for water
distribution networks, J. Computing in civil engineering, ASCE, 10 (1), pp. 10-19.
Kjeldsen T. R. & Rosbjerg D., (2004) – Choice of reliability, resilience and vulnerability:
estimators for risk assessments of water resources systems - Hydrological Science Journal
49(5), 755-767.
La Loggia G., Fontanazza C., Freni G., Notaro V. & Oliveri E., (2005) – Programmazione e
gestione dei sistemi acquedottistici in condizioni di scarsità di risorsa – Relazione generale
presentata al 1° Convegno Nazionale di Idraulica Urbana, Sorrento, 28-30 settembre.
La Loggia G.. & Mazzola M.R., (1989) – Syntetic performance indices in a real decision
process, Closing the gap between theory and practise, Proceedings of the Baltimore
Symposium, IAHS Publ. N°180.
Lamberti P, Maiolo M., Mazzola M.R. & Veltri P., (1994) – Rapporto sui consumi idropotabili
in Italia sul finire degli anni ’80 – a cura di P. Bertola e L. Taglialatela, CUEN, Napoli.
Liserra T., Bragalli C., Maglionico M., Casadio A., Artina S. & Pilati S., (2007) – Indicatori di
prestazione in sistemi acquedottistici aggregati mediante rete bayesiana - Atti del 2°
Convegno Nazionale di Idraulica Urbana, Chia (Ca), 25-29 sett.
Marchetti M., (1949) – Acquedotti – Ed. Tamburini, Milano
Mays L.W., (1989) – Reliability analysis of water distribution system – Ed. ASCE, New York,
ISBN 0-87262-712-8
Molino B., Rasulo G. & Taglialatela L., (1986) – Struttura dei consumi idrici dell’area
napoletana – Idrotecnica, n° 6, pp. 371-381.
Ozger S.S. & Mays L. W., (2003) - A semi-pressure-driven approach to reliability assessment
of water distribution networks, Proceedings of the Thirtieth Congress, International
Association of Hydraulic Research, Thessaloniki, Greece, 2003
Paoletti A., Becciu G., Sanfilippo U., Carnevali S. & Giambruno S., (2007) – Indicatori di
prestazione per il supporto alle decisioni nell’ambito della pianificazione dei sistemi
acquedottistici - Atti del 2° Convegno Nazionale di Idraulica Urbana, Chia (Ca), 25-29 sett.
Pelletier G., Mailhot A. & Villeneuve J., (2003) – Modeling water pipe breaks – Three case
studies - Journal of Water Resources Planning and Management, Vol. 129, No. 2, pp. 115-
123
Pianese D. & Villani P., (1994 a) – Verifica di affidabilità delle reti idriche in pressione. Parte
prima: valutazione mediante indici locali. - Atti del XXIV Convegno di Idraulica e
Costruzioni Idrauliche, Napoli, pp. T3b67-T3b80
Pianese D. & Villani P., (1994 b) – Verifica di affidabilità delle reti idriche in pressione. Parte
seconda: valutazione mediante indici globali. Esempio di applicazione ad un caso di

26
C. CIAPONI – Performance analysis in water supply

studio. - Atti del XXIV Convegno di Idraulica e Costruzioni Idrauliche, Napoli, pp. T3b81-
T3b94
Rossman L.A., (2000) - EPANET 2 users manual, U.S. Environmental Protection Agency,
Cincinnati, Ohio
Salandin P. & Bertola P. (1996) - Analisi dell’affidabilità delle reti di distribuzione soggette a
richiesta aleatoria – Atti del XXI Convegno di Idraulica e Costruzioni Idrauliche, Torino,
pp. 217-228
Salandin P. & Darvini G., (2007), - Sistemi di distribuzione con rotture delle condotte e
richiesta della portata aleatorie: una tecnica di valutazione dell’affidabilità – Atti del
Convegno “Approvvigionamento e Distribuzione Idrica: Esperienza, Ricerca ed
Innovazione”, Ferrara, 28-29 giugno.
Shinstine D. S.. & Lansley K.E., (2002), Reliability/Availability analysis of municipal
distribution networks: case studies - Journal of Waterway, Port, Coastal and Ocean
Engineering, ASCE 128(2), 140-151
Su Y., Mays L.W & Lansey K.E., (1987) – Reliability-based optimisation model for water
distribution system – Journal of Hydraulic Engineering, Vol 113, n° 12, pp. 1539-1556
Tanymboh T.T., Tabesh M. & Burrow R., (2001) – Appraisal of Source Head Methods for
Calculating Reliability of Water Distribution Networks – Journal of Water Resources
Planning and Management, Vol. 127, n° 4, pp. 206-213
Todini E., (2003) - A more realistic approach to the “Extended Period Simulation” of Water
Distribution Networks - Proc., 2003, Advances in Water Supply Management CCWI
Conference, London, 2003.
Tucciarelli T., Criminisi A. & Termini D., (1999) - Leak analysis in pipeline system by means
of optimal valve regulation, Journal of Hydraulic Engineering, Vol. 125, No. 3, pp. 277-285
Wagner J.M., Shamir U. & Marks D.H., (1988) - Water distribution reliability: simulation
methods, Journal of Water Resources Planning and Management, Vol. 114, No. 3, pp. 276-
294
Walski T. M. & Pelliccia A., (1982) – Economic Analysis for Water Main – Journal of
AWWA, Vol 79, n° 3, pp. 140-147
Walski T.M., (1987) – Discussion of “Multi-objective optimisation of water distribution
networks”- Civil Engineering System, Vol 4, n° 1, pp. 215-217
Walters G. A. & Knezevic J., (1988) – The influence of reliability on the layout and design of
water distribution networks – Proceeding of Symposium on Reliability-Based Design in
Civil Engineering, Ecole Polytechnique Fédérale de Lausanne, Switzerland, 1, pp. 237-243

27
PERFORMANCE INDICATORS FOR THE PLANNING, THE DESIGN AND THE MANAGEMENT OF WATER SUPPLY SISTEMS

Xu C. & Goulter I. C., (1998), - Probabilistic Model for Water Distribution Reliability –
Journal of Water Resources Planning and Management, Vol. 124, n° 4, pp. 218-228
Yepes G. & Dianderas A., (1996) – Water and wastewater utilities indicators – Water
Sanitation Division, The World Bank. Washington, Usa (downloadable from the website:
http://www.worldbank.org/html/fpd/water/pdf/indicators.pdf)

Acknowledgements
This work was carried out in the PRIN 2005 research program “Performance indicators for the
planning, the design and the management of water supply systems” financed by the Italian Ministry
of Education, University and Research (MIUR).

28

View publication stats

Potrebbero piacerti anche