Theoretical and Conceptual Framework - Pattern Recognition for Reliability Assessment of Water Distribution Networks - page 27

optimised at certain minimum pressure and minimum investment costs will result in a layout
that performs well under regular conditions but is not necessarily sufficiently reliable.

(B.3) Is it possible to incorporate the impact of water quality into reliability assessment?
In the wider context of reliability, a distribution network would not be reliable if the full
demand has been satisfied but with water of deteriorated quality, carrying pathogens or
unacceptable colour, taste or odour, even if not hazardous. This aspect is particularly relevant
in view of the fact that the improvement of hydraulic reliability by enlarging the pipes
enhances water stagnation, which is a source of a range of water quality problems.
Correlating the threshold values of velocities and chlorine residuals to the system pressures,
in regular and irregular supply scenarios, would likely add value to the overall reliability
analysis.

(B.4) Are the pipe failure models sufficiently describing the reality?
Water quality and its impact on corrosion is the factor affecting the pipe failure rates.
Corrosion growth results from very complex mechanism and is a long-lasting event that is
difficult to model with confidence. Nevertheless, verifying more complex hypothesis and
models seems increasingly possible, owing to improved monitoring and recording of the field
data. A field study that correlates service conditions (component materials, installation,
operation, maintenance, environment, etc.) to the failure rates could make very useful
contribution to the modelling of the frequency of failure events.

(B.5) Is a snapshot method for preliminary assessment of reliability possible?
The literature review did not result in any quick method for preliminary assessment of
reliability that could call for deeper analyses of reliability, similarly as it is done in medical
investigations. Such approach could correlate the basic system parameters: the network
layout, topography and supplying scheme and would develop an index hinting, already during
regular operation, if the system is sufficiently reliable. Two examples elaborate this concept:
- From the perspective of demand reduction, a network with higher pressures is likely to be

more hydraulically reliable than the network with low pressures, despite the negative
effects on leakage. A snapshot assessment of the hydraulic performance during regular
operation could indicate threshold pressures that shall not drop below the value that could
significantly affect the demand, once the pipe failure has occurred.

- Similarly, a network with larger pipe diameters, more (connected) pipes, and newer i.e.

smoother pipes is potentially more reliable than an insufficiently developed network of
old pipes. One way to survey the reliability in both of these cases could be to compare
their resistances for various demand scenarios and suggest a critical resistance that will
almost surely lead to a drop of pressure below the same threshold value as in the first
example.

A feeling is that there is more room to investigate the correlations between reliability
measures and major performance parameters than is presented in the literature.

2.6.3 Tools (C)

The simulation approaches for reliability assessment rely predominantly on the steady-state
network hydraulic calculations. The pressure-driven models conceptually offer more accurate
representation of the low-pressure conditions but most of the commercial software still
operates based on the demand-driven principles as a default calculation mode, offering faster