Decision Support Tool for Design and Reliability Assessment of Water Distribution Networks - Pattern Recognition for Reliability Assessment of Water Distribution Networks - page 267

an additional pipe of 600mm has been connected between the nodes 3 and 34. In all these
cases, the pipe lengths have been recalculated according to the coordinates of the newly
connected nodes. Eventually, the hydraulic simulations have given the pressures shown in
Figures 9.14 and 9.15 and two HRDs as in Figure 9.16.

1.00 1.00

Underconnected Underconnected Underconnected Underconnected

0.90 0.90

0.80 0.80

0.70 0.70

0.60 0.60

0.50 0.50

0.40 0.40

0.30 0.30

0.20 0.20

0.10 0.10

Overconnected Overconnected Overconnected Overconnected

0.00 0.00

0.00 0.20 0.40 0.60 0.80 1.00 0.00 0.20 0.40 0.60 0.80 1.00

1-ADF 1-ADF

Figure 9.16 HRD: layout nr. 3307 - left, layout nr. 962 - right

The summary and comparison of the main results with the original layouts are shown in
Table 9.20.

Table 9.20 Comparison of originally generated and modified networks

Net nr. Property Cost (10 6 ) p min -p max (msl) ADF avg NBI PBI I n

3307 Original 4.00 20.01 - 38.28 0.983 0.785 0.962 0.363

M3307 Modified 4.11 20.55 - 39.37 0.993 0.891 0.965 0.426

962 Original 3.78 19.47 - 34.90 0.976 0.725 0.969 0.272

M962 Modified 3.98 19.60 - 37.92 0.986 0.818 0.974 0.372

Both the figure and the table show the improvement of reliability at relatively minor cost
increase. Yet, the addition of the pipe 3-34 was more useful in case of the layout nr.3307 and
to lesser extent in case of nr.962; further improvement of that layout would require change of
more pipe routes/diameters. Interesting to observe is that the modified layout nr.962 costs
approximately the same as the original layout nr.3307 and with relatively similar values of
the parameters in the table, which can be considered as a coincidence.

9.6 CONCLUSIONS

This chapter has dealt with the decision support tool for network design and reliability
assessment (NEDRA), compiled from the software applications specifically developed for the
research presented in Chapters 5 to 8. The tool has been composed of the modules for
network generation and filtering, initialisation, GA-optimisation on the least-cost diameter,
diagnostics of network geometric-, hydraulic-, and economic parameters and finally the
assessment of its resilience. Computations can be done for a single network or a sequence of
networks of virtually unlimited size and number, mutually similar or different in properties.
The programme has been illustrated on a design example of a gravity network of one source
and 50 nodes. 13,000 network layouts have been generated and further filtered to arrive at