Theoretical and Conceptual Framework - Pattern Recognition for Reliability Assessment of Water Distribution Networks - page 25

the reliability where the quantity comes clearly before the quality. In theory, there are two
ways in which the water quality parameters could be included in the reliability assessment:
1. by analysing a probability that once it has been tapped, the water will (not) satisfy
prescribed quality standards; the reasons for this can be various: poor water treatment,
pipe sediments and corrosion, intrusion of pollution through pipe bursts, etc.;
2. by analysing effects of particular water quality on ageing of pipes and consequently
having an idea about the link between the measures taken to maintain a certain water
quality and frequency of bursts, all with implications for the technical/economic life time
of the pipe.

Practically, both of these are very difficult problems to solve; complex mechanisms that
explain interactions between various water composition and pipe materials are the main
reason for it.

(A.4) What are the most appropriate reliability measures/indices?
The network reliability is in the first place about the probability of getting 'a glass of potable
water' any time the tap is open, and less about the probability of having a component failure
in the system. Although these two events are closely related, possible shortage of supply is
not always a result of mechanical failures in the system. Therefore, the concept of hydraulic
availability describes the picture better than the mechanical availability, which describes one
part of the problem, only.

As demonstrated by the two examples in the previous section, many reliability calculations
often end-up with a factor, which takes values between 0 and 1. The following, more or less
general conclusions, can be drawn with respect to such result:
- Expressing the network reliability with single index gives an advantage of easy check of

its sensitivity towards possible reconstructions attempted in the system or changes of
various network parameters. It is possible to assess the improvement i.e. the increase of
the index resulting from the investment and operational costs implying from certain
scenario.

- Nevertheless, while they give fairly straightforward conclusions in terms of 'higher/lower

i.e. better/worse than….' when comparing several alternatives, the index still offers a bit
unclear description of the magnitude of the problem when evaluated independently. It is
therefore not clear in advance which values/ranges can be considered as
high/medium/low reliability, in other words acceptable or unacceptable. This suggests the
need for a research that would deal more in-depth with sensitivity analysis of the
reliability indexes in absolute terms.

- Once the hydraulic reliability has been improved to a certain degree, the question is if and

how this translates into the improvement of mechanical reliability, for instance the pipe
failure frequency, and are the costs involved really justified.

Ideally, any hydraulic reliability index should be convertible into a probability of getting a
sufficient quantity of potable water any time the tap is open. The surrogate indexes described
in this chapter are missing this conversion, next to the fact that they seem to be less
applicable in practice than the concept of availability indices.