Introduction - Pattern Recognition for Reliability Assessment of Water Distribution Networks - page 4

Obviously, good knowledge of the system is prerequisite while defining its limits. A certain
degree of calamities is inevitable in this learning process. Inflicting a bearable damage can be
helpful while assessing and adapting the limits; the equilibrium will be restored after the
calamity and conclusions can be drawn accordingly. Moreover, when properly defined
system objectives have matched the risks, the effect of calamities will be reduced. Such risks
can then be better controlled. Only by accepting controlled risks, we are able to learn more
about the limits and how to expand them. By doing this, the growth of the system will be
sustainable and its lifetime prolonged.

1.3 RELIABILITY

Systems are deemed reliable if they can withstand predicted level of calamity . Each calamity
creates an impact. For instance, in the field of water distribution a pipe burst or electricity
failure creates a drop of pressure and consequently a loss of demand. This calamity/impact is
more or less probable but can also be more or less intense, and can cover a larger- or smaller
area.

Talking about the low- or high reliability, one would have to define it in situations that are
sometimes hardly comparable. For instance, a burst of pipe can affect relatively small area
severely, or a larger area moderately. Which of these two scenarios depict more/less reliable
network? Equally, would a system with new pipes, whose failure can create severe impact for
consumers, be considered as more reliable than the system of old pipes, whose failure results
in moderate impact?

In general, failures inflict consequences based on wide- or limited spatial coverage of the
impact, and its high- or low intensity. Combined with high/low probability of the failure, this
leads to four typical cases with growing concerns for consumers:
- Case 0 (no concern): low spatial coverage with low intensity of the impact.

- Case 1 (low concern): wide spatial coverage but low intensity of the impact.

- Case 2 (moderate i.e. limited concern): limited spatial coverage but high intensity of the

impact.

- Case 3 (potentially high concern): wide spatial coverage with high intensity but low

probability of impact.

Table 1.1 Reliability bandwidth

Reliability Low > > > > > > > > > > > > > > > > > > > > > > > High

Impact probability High Low High High High Low Low Low

Impact intensity High High Low High Low High Low Low

Impact coverage High High High Low Low Low High Low

In any calculation of water distribution network reliability, the meaning of low- or high index
usually taking a value between 0 and 1 could possibly be associated with a bandwidth as
shown in Table 1.1. The final verdict will however depend on the definition of the 'High' and
the 'Low', which can be quite different in different countries. According to the studies done in
The Netherlands in the late eighties, the average frequency of interruptions affecting the
consumers was remarkably low; the chance that no water would run after turning on the tap
was once in fourteen years. In the similar period, in 1993, the frequency of interruptions in