Theoretical and Conceptual Framework - Pattern Recognition for Reliability Assessment of Water Distribution Networks - page 26

2.6.2 Methods (B)

The literature review reveals large number of reliability approaches, the simulation
approaches being the most frequently developed and tested. These approaches also look the
most transparent and applicable in reality, although they rely heavily on the quality of data
used for input and calibration of the hydraulic- and failure models. The approaches presented
in the other two groups, analytical and heuristic, are either too simplified, or too complex, i.e.
contain some parameters that are difficult to survey, monitor or interpret in reality.
Nevertheless, the analytical concepts of connectivity, reachability and redundancy look
interesting and would require further research. It is very clear that in cases of pipe failures,
the water changes its path in the system and the network resistance will play essential role in
redistribution of pressures and flows. The questions to address the gaps in analysing the
network reliability are formulated in the following five paragraphs.

(B.1) Are the simulation approaches offering complete picture about network reliability?
The conclusions from the literature review and two applications illustrated in the previous
section can be summarised as follows:
- The simulation approaches take the major aspects of reliability into consideration and as

such are sufficiently practical, provided that reliable data for the model building and
calibration are available.

-

In some aspects, namely the determination of nodal reliability, and consequently the
overall reliability as an average value, the methods simplify the reality. Averaged figures
potentially hide failures that can by no means be tolerated, due to their bad consequences
for the system operation. If these are just a few, but very weak points in otherwise reliable
system, the picture about its overall reliability may be distorted. Introducing some kind of
weighting of nodal reliabilities could be considered to rectify this problem.

- Categorising the pipe availability per diameter may not be entirely accurate as it neglects

implications of the pipe location in the system. The same diameter pipe can have more
chance to burst if located in a high pressure area (e.g. closer to the source), or is exposed
to more aggressive soil conditions, excessive surface loading, etc. than if it is located in
the area where all these risks are smaller or can be avoided. Moreover, the pipe age and
material will also have different impact on the frequency of bursts of the same pipe
diameters. Including at least these two variables should therefore help to diversify this
information.

- Most of the methods focus to pipe failures exclusively, neglecting operations of other

components: pumps, storage and valves. Typically for many water distribution systems in
developing countries, electricity failures are often the major cause of water shortage, and
the hydraulic reliability in such cases will depend largely on available buffer volume in
the reservoirs. Depending on the moment of the electricity failure, the chance that these
reservoirs will not function properly, if normally refilled by pumping, also exists. More
research would therefore be needed on these issues.

(B.2) Is it possible to incorporate economic aspects into reliability assessment?
The presented simulation approaches predominantly deal with hydraulic aspects and much
less with economic aspects of reliability. From the perspective of water companies, a crucial
decision is where their money should be invested: into cleaning, repair, replacements,
additional pipes, pumps, storage volume, etc. What that means for the improvement of the
service level, and reduction of failure rates, losses of water/revenue/consumer confidence,
etc. resulting from the calamities, is important question. It is often a case that the network