Spatial Network Generation Tool for Performance Analysis of Water Distribution Networks - Pattern Recognition for Reliability Assessment of Water Distribution Networks - page 78

of codes between M1 and M1200 , and Rep2.txt the codes from N1 to N600 , as shown in Table
4.11.

Table 4.11 Contents of Rep1.txt and Rep2.txt support files for networks up to 200 nodes

M1
M2
M3
M4
M5
.....
M1199
M1200

N1
N2
N3
N4
N5
.....
N599
N600

The output of NGT is the set of generated *.inp files with different nodal connectivity, which
are ready for hydraulic simulation in EPANET. The pipe properties in those files are defined
either randomly within specified range, or by using coordinates to calculate the lengths.
Redefinition of nodal elevations and demands is also possible. NGT can also prepare
additional files that are necessary if diameters of generated networks are to be GA-optimised,
which is done in separate programme discussed in Chapter 9.

The flow chart of NGT is given in Figure 4.13. The algorithm has ten distinct steps:

Step 1 : Reading of initially prepared EPANET *.inp file containing information about
reservoirs/tanks/(demand) nodes; reading of supporting files Rep1.txt and Rep2.txt .

Step 2 : The distances between the nodes are calculated using their coordinates and the values
sorted in ascending order.

Step 3 : Three nearest nodes to the source reservoir are used to start up the list of
combinations by occupying the first three columns of the first row; these are having the
vertex degree 1. The combinations (1,2), (1,3) and (2,3) will have vertex degree 2 and the
combination (1,2,3) in the set has vertex degree 3.

Step 4 : The following nearest node to the reservoir fills the next row of combinations made
from its own three closest nodes. Each of those nodes is assigned a unique serial number and
its edge is tested for planarity with already created edges in the adjacency lists. If an
intersection exists, the numbered edge is put into a library of intersections maintained by two
supporting matrices.

The steps 2 to 4 are repeated until the list of nodes has been fully exhausted.

Step 5 : One set of edges creates a planar sub-graph, which is a combination from elements
taken from each row. In non-random generation process, the algorithm generates recursively
all possible combinations for rows and their subsets. In random generation process, each
element is randomly chosen to make a set of edges.

Step 6 : The edges of each created sub-graph are compared with those stored in the library of
intersections. In cases where the IDs match, the other edge of the intersection shall replace
the current one to create alternative sub-graph.