Decision Support Tool for Design and Reliability Assessment of Water Distribution Networks - Pattern Recognition for Reliability Assessment of Water Distribution Networks - page 240

It is visible on the graphs that the matching of design velocities and hydraulic gradients can
be achieved over the wider range of diameters, for lower S -values and/or higher k -values. The
first condition could be considered favourably from the perspective of reliability, while the
second one has its limitations. It is normal design practice to include the effect of minor
losses though somewhat increased k- values, but if those are much higher than a few
millimetres, that would signal corroded- rather than properly designed pipes.

9.3 NETWORK DESIGN AND RELIABILITY ASSESSMENT TOOL

The ne twork d esign and r eliability a nalysis tool (NEDRA) has been developed to facilitate
design process of water distribution networks by adding more of reliability assessment
perspective to it. The tool has been compiled from the software applications specifically
developed for the research presented in Chapters 5 to 8. By adding the filtering and
initialisation feature to the network generation tool elaborated in Chapter 4, and using the
hydraulic solver for pressure-driven demand calculations, a coherent set of routines has been
created which can analyse numerous design scenarios considering different connectivity, pipe
diameters, demand distribution (both spatial and temporal) and assess these against several
reliability measures and economic scenarios. Although not necessarily relevant for the design
process, the features related to random generation of some input data, namely the
connectivity and nodal elevations, have been preserved keeping NEDRA still applicable for
research purposes.

The programme has been coded in Microsoft Visual C++ 2010 Express, using EPANET
toolkit library to communicate with EPANET software of US EPA (Rossman, 2000). The
GA-optimiser that has been integrated into NEDRA is the Evolving Objects (EO) , used for
network optimisations in Chapters 4 to 8.
NEDRA consists of five main modules dealing with network (1) generation, (2) filtering, (3)
initialisation, (4) optimisation and (5) diagnostics. A brief description of each module is
given in the following sections.

9.3.1 Network Generation Module

The network generation module has been developed applying the concepts of graph theory,
which has been elaborated in Chapter 4. The module, presented there as NGT (network
generation tool) generates different network layouts, based on selected location of nodes and
sources, which is initially prepared in EPANET in INP-file format. The programme reads the
nodal information prepared in EPANET, which includes the coordinates, the elevations and
baseline demands. The latter two can be modified, also by specifying a range of values that
will be randomly assigned in both cases.

The network generation takes place in two ways: (1) non-randomly and (2) randomly. The
graph theory principles applied in the non-random generation mode result in an algorithm
that explores all possible elements of a matrix formed on the limitation that each node in the
process of generation can be connected to maximum three additional nodes. Next to this
condition, the algorithm is developed to connect first the closest available nodes, avoiding the
crossings with the existing pipes, and duplications that could occur as a result of the reversed
order of the pipe nodes .The degree of complexity of generated networks can be influenced