Theoretical and Conceptual Framework - Pattern Recognition for Reliability Assessment of Water Distribution Networks - page 8

2.1 BACKGROUND

Even in developing countries, where reliable distribution network is still not a high priority,
the customer awareness regarding the service levels has risen significantly in present days.
Although many of the systems there suffer from water shortage at the source, poor state of
assets and lack of funds caused by low tariffs and/or poor revenue collection, a reliability
analysis makes sense in order to explore optimal operation under given constraints. As
mentioned by Vairavamoorthy et al. (2001), many water companies operating such systems
will by default be unable to offer 24-7 supply but could, by applying a kind of 'constrained
optimisation', achieve a 'regular irregularity' i.e. a fair share of limited water quantities for all
consumers and at predictable intervals.

Nowadays, water experts feel overall more confident to address the reliability concerns
having at their disposal powerful network modelling software. Some of these programmes are
distributed through the Internet at nominal costs or even free of charge; EPANET software
developed by the US Environmental Protection Agency is a well-known representative of this
group (Rossman, 2000). Other, commercially available software include direct interaction
with data available in geographical information systems and are able to process the results
and calibrate the model in more sophisticated manner. On top of it, the latest generation of
computer programmes, used in water distribution, includes optimisation modules based on
methods applying genetic algorithms. Walski et al. (2003) offer a comprehensive overview of
the state-of-the-art water distribution network modelling practices.

2.2 RELIABILITY ASSESSMENT OF WATER DISTRIBUTION NETWORKS

Already in the last two decades, the reliability of water distribution networks has been
recognised as one of the top water supply challenges. Three questions have been particularly
relevant, namely: (1) what parameters are the most accurate descriptors of the reliability, (2)
what can be considered as an acceptable reliability level, and (3) what is the most appropriate
reliability assessment method? Despite lots of efforts, it is still not easy to answer these
questions because the reliability assessment relates to numerous factors, such as uncertain
nature of demand variables, overall condition of the network, the pressure-flow relationship,
different standards with respect to water consumption, etc. In fact, the literature reveals that
there is no universally acceptable definition and measure for the reliability of water
distribution systems.

Several researchers have tried to define the network reliability; Table 2.1 contains a few
frequently cited quotes in the literature. In most of these, the reliability is related to the
(nodal) pressures and flows. As such, the reliability reflects the ability of distribution network
to satisfy consumer demands i.e. a certain level of service subject to the pressure and demand
requirements under both normal and abnormal conditions. This conclusion brings the
definition of reliability back to its roots, stating that quantified reliability, in general, is a
measure of system performance. Having said that in case of water distribution network
implies its hydraulic performance, which is in the literature defined as the hydraulic
reliability.