Environmental Engineering Reference
In-Depth Information
Table 9.5
Results of network filtering for 30 batches of 100 nets
Batch
code
Limited nodal connectivity (4 or 5)
Unrestricted nodal connectivity
Pipe
range
Avg.
Net
match
MisM
range
Avg.
miss
Pipe
range
Avg.
Net
match
MisM
range
Avg.
miss
R55-65
55-65
63.65
0
1-9
5.04
55-65
63.63
0
2-9
4.94
R65-75
65-75
68.82
0
2-8
4.97 65-75
71.04
0
2-9
5.75
R55-75
55-75
68.10
0
1-8
5.11 65-75
71.33
0
3-9
5.85
Rfull
62-74
67.99
0
1-9
4.99 63-81
71.78
0
1-9
5.72
NfullC2
50-54
51.46
54
0-1
0.46
50-54
51.40
36
0-1
0.64
NfullC3
50-53
51.76
0
7-8
7.66
50-54
51.67
0
7-8
7.81
NfullC4
74-77
75.13
0
1-2
1.73
74-77
75.13
0
1-2
1.73
NfullC5
74-77
74.94
0
9
9.00 74-78
75.21
0
9
9.00
NfullC6
74-80
77.16
0
9
9.00 77.16
74-80
0
9
9.00
N55-65C2
55-56
55.10
40
0-1
0.60
55-57
55.16
14
0-1
0.86
N55-65C3
55-56
55.04
0
7-8
7.59
55-57
55.16
0
7-8
7.87
N55-75C2
55-56
55.10
40
0-1
0.60
55-57
55.16
14
0-1
0.86
N55-75C3
55-56
55.04
0
7-8
7.59
55-57
55.16
0
7-8
7.87
N55-75C4
74-75
74.61
0
1-2
1.68 74-75
74.61
0
1-2
1.68
N55-75C5
74-75
74.68
0
9
9.00 74-75
74.68
0
9
9.00
The results of the filtering process are presented in Table 9.5. The average values in the table
show the average number of pipes produced by the generator and the average number of
pipes that do not exist in the template network, which is also reflected by the respective pipe
ranges. The randomly generated layouts have not produced any matching layout, while the
non-random generation had the matching layouts in six batches, displayed in the shaded cells
of the table. Four additional batches had no layouts complying with the template, but the
filtering tool indicated the difference in only 1 or 2 pipes. These are mostly the same pipes as
can be seen in Table 9.6 showing the output layout of the network filtering module
The filtering results prove that the batches of 100 layouts are apparently insufficiently
variable and too small for the limitations imposed by the number of links in the template.
Consequently, the networks that fit to it can be generated only with the low level of
complexity. Finally, the results in Table 9.5 show similarities for some categories, regardless
the selected range of number of pipes, and/or maximum number of nodal connections, which
again can be attributed to the restrictions caused by the interference between these parameters
and the total number of nodes.
To get more insight into the pattern of network configurations, one random- and four of the
non-random batches highlighted in Table 9.5 have been regenerated in the size of 1000 nets;
these results are shown in Table 9.7. The codes in the brackets stand for unrestricted nodal
connectivity (U), and the one maximised at 4 connections per node. Finally, the random batch
R55-75(U) has been regenerated in the size of 5000 layouts, with the results shown in Table
9.8.
Search WWH ::
Custom Search