Introduction - Pattern Recognition for Reliability Assessment of Water Distribution Networks - page 3

- To establish clear relation between these processes and appropriate maintenance methods.

- To have a system of monitoring providing information that can allow reasonable

calibration of the model.

- To have good information about economic consequences of certain decisions.

- To develop and/or acquire powerful hardware and programming tools.

Back in the reality, the vast majority of problematic water distribution networks are located in
developing countries, whereas the vast majority of good networks exist in developed
countries. The problem is in the fact that modern computer software alone cannot help to
erase this difference; it is just a tool to work with. Applying it on a network case where good
quality information is missing is just like attempting to drive an expensive limousine on a
road where maximum possible speed is 30 km/h. In spite of futuristic predictions, the world
is also to achieve the Millennium Development Goals, much of them dealing with safe
drinking water.

1.2 LIMITS AND RISKS

Any system is composed of components that interact in an equilibrium established within
certain limits. Once this equilibrium becomes disturbed, the system tries to restore it. If this is
impossible, a calamity is going to occur. Sometime after the calamity, the equilibrium will be
restored. Without external intervention, this will be achieved at lower level i.e. at lower
limits. In extreme cases, this cannot be achieved, even with external intervention, because the
system has collapsed.

Systems operate at certain level of risks. It may be valid to say that risks partly originate from
the lack of awareness about the limits . For example, a pretty inexperienced driver can drive
his/her very bad car, passing by an accident caused by very experienced driver in his/her
brand new car. What initially appears to be more reliable transportation system will suffer
calamity if the limits of its components: the car, the driver and/or the road in this case, have
been underestimated; the latter person was simply driving 'too risky'.

How to quantify risks? These will be normally associated with probability that something bad
is going to happen. The less aware of the limits we are, the higher the risk will be. However,
we are not necessarily concerned with high risks of catching cold, as we are afraid of low
risks to contract some fatal disease. 'High' and 'low' quite often coincide with 'very bad' and
'not that bad'. Hence, the risk is high when something 'unacceptably bad' might happen; no
matter if the calculated chance is 10% or 0.0001%. This means that the level of acceptable
risk is proportional to the magnitude of calamity.

Proper assessment of limits helps to reduce the level of risks. This is easy to say, but where
the real limits are is often a difficult question. Science has equipped us with methods and
tools that allow fairly accurate estimate of limits in many cases. Nevertheless, some systems
are quite complex and it is not always possible to judge interactions between numerous
components. Not only that we do not posses sufficient knowledge about their limits but those
also change in time. Being unsure about it, engineers will usually try to protect the system
from calamities by introducing higher safety factors in their designs. Effectively, the result of
this is a system that most of the time operates below its limits, i.e. an inefficient system.
Contrary to this, what one would expect from any system is an optimal performance i.e. a
performance close to, but not beyond its limits.