Decision Support Tool for Design and Reliability Assessment of Water Distribution Networks - Pattern Recognition for Reliability Assessment of Water Distribution Networks - page 254

Table 9.8 Results of network filtering for random batch of 5000 nets

Batch
code

Time (hrs) Pipe range Avg. Net match MisM range Avg. miss

R55-75 (U) Approx. 8 57-75 70.80 1 0-10 5.66

Two representative layouts of the batches from Table 9.7 have been shown in Figures 9.7a to
9.7e, the first one and the last one in the series of 1000. Although slightly better, the results
for 1000 layouts do not show much different patterns than in case of 100 layouts per batch.
Remarkably (or actually not), the random network generator has kept the average number of
pipes not matching the template, at 5.66 in all three cases: 100, 1000 and 5000 layouts.
Hence, the overall impression is that the layout of template can play essential role in the
matching of the generated files. To document this, Table 9.9 presents the increase in number
of matching layouts per maximum number of links in each of these, allowed not to exist in
the template.

Table 9.9 Compliance with the template, for non-matching links

Net match for the max. number of non-matching pipes

Batch Code

0 1 2 3

R55-75(U)1000 3 7 21 93

R55-75 (U)5000 1 19 118 409

NfullC2(4)1000 477 1000 1000 1000

NfullC4(U)1000 0 252 1000 1000

N55-75C2(4)1000 375 1000 1000 1000

N55-75C4(U)1000 0 304 1000 1000

Figure 9.7a Sample of generated networks R55-75(U): layout nr.1 - left, layout nr.1000 - right

Figure 9.7b Sample of generated networks NfullC2(4): layout nr.1 - left, layout nr.1000 - right