smart technologies for safety engineering

68 Smart Technologies for Safety Engineering
having lots of data in dynamics is a much more time-consuming numerical analysis. Good
qualitative identification of two delamination zones (including the edge) has been achieved
(see Figure 3.35). Experimental verification of delamination identification in dynamics has
also been successful (see Figure 3.36).
A proposition for on-line identification has been put forward. The problem has just been
recognized at the numerical level. An experimental verification will be the subject of future
research.
3.4 Leakage Identification in Water Networks
3.4.1 Introduction
The problem of management of water sources is more and more important in the world scale.
In particular, the problem of detection and identification of leakages (mostly due to corrosion)
in water networks is an important engineering challenge, especially in tropical countries often
suffering from the lack of water. On the other hand, the consequences of unpredicted largescale
leakage in the operating water network may be very serious. Therefore, there is a need for
an automatic monitoring system able to detect and localize leakages in the early stage of their
development. A number of papers addressing the issue have been published in the hydraulicsrelated
journals of the American society of civil engineers (ASCE). The use of a genetic
algorithm for solving the inverse transient problem has been proposed in References [45]
and [46]. Frequency response, provoked in open-loop piping systems by periodic opening and
closing of the valves, has been investigated in Reference [47]. The damping of transient events
has been discussed in [48]. The papers [49] and [50] demonstrate that leakages can be not only
detected but also reduced by optimal valve control.
The proposed approach is based on continuous observation of the water heads at the nodes
of the water network. By having a reliable (verified versus field tests) numerical model of
the network and its responses for determined inlet and outlet conditions, any perturbations to
the original network response (water head distribution) can be detected. Then, applying the
proposed numerical procedure, the inverse problem of the water flow distribution can be solved.
The possibility of simultaneous identification of several leakages with different locations and
intensities is included in the presented approach.
The presented methodology for leakage identification [51] is an extension of the VDM applications
in structural mechanics (cf. Section 3.2). Analogies between truss structures and water
networks, both modeled as closed-loop systems, the system theory and graph representation
of the truss-like system, have been effectively used.
3.4.2 Modeling of Water Networks and Analogies to Truss Structures
Advantage will be taken of the general system theory (cf. Reference [52]) in order to build a
numerical tool for modeling and analysis of water networks (cf. Reference [53]). To this end,
a water network needs to be visualized as an oriented graph (a two-loop example is depicted
in Figure 3.39) with the direction of water flow indicated by arrows.
The network shown in Figure 3.39 consists of five real branches, connecting four nodes.
The water inlet is located at node 1 and the outlet at node 4. There is also a reference node
W0 serving as an arbitrary level H0 for measuring water heads at the network’s nodes. Four
fictitious branches connecting the reference node W0 with the corresponding four network
nodes are provided in the analysis. It can be seen that the topology of the water network is