Environmental Engineering Reference
In-Depth Information
Table 9.4
Network batches and corresponding generation time (in seconds for 100 configurations)
Batch code
Description of generation settings
Max.4-5 con.
Free con.
R55-65
Random, between 55 and 65 links
5983 (4 con)
389
R65-75
Random, between 65 and 75 links
475 (4 con)
92
R55-75
Random, between 55 and 75 links
385 (4 con)
96
Rfull
Random, no restriction on number of links
380 (4 con)
88
NfullC2
Non-random, no link restriction, complexity 2
135 (4 con)
130
NfullC3
Non-random, no link restriction, complexity 3
150 (4 con)
149
NfullC4
Non-random, no link restriction, complexity 4
58 (5 con)
57
NfullC5
Non-random, no link restriction, complexity 5
71 (5 con)
72
NfullC6
Non-random, no link restriction, complexity 6
64 (5 con)
66
N55-65C2
Non-random, 55 to 65 links, complexity 2
133 (4 con)
142
N55-65C3
Non-random, 55 to 65 links, complexity 3
196 (4 con)
165
N55-75C2
Non-random, 55 to 75 links, complexity 2
142 (4 con)
144
N55-75C3
Non-random, 55 to 75 links, complexity 3
196 (4 con)
162
N55-75C4
Non-random, 55 to 75 links, complexity 4
58 (5 con)
66
N55-75C5
Non-random, 55 to 75 links, complexity 5
69 (5 con)
70
To illustrate the type of configurations, each batch is represented by one network, mostly the
first one or the last one in the sequence of 100, which is shown in Figures 9.5a to 9.5o. The
figures give visual impression that complies with the selected complexity level in the non-
random generation. Moreover, this mode generates pretty similar layouts within the same
complexity level, which suggests that the non-random generation needs much larger number
of simulations to arrive at significantly different layouts if all matrix columns are to be
visited; just a few dozen layouts may differ in one or two connections, only. The random
generation produces more variety within the batch, and is less dependent on the combinations
of selected number of pipes and/or the maximum number of connections. Specific settings of
both of these parameters, which have clear relation to the number of nodes, can quite easily
impact the generation time and the number of available configurations. This is particularly
visible in the significantly higher generation time of the first two random batches (R55-65).
On average, the random generation has been longer than the non-random one.
Figure 9.5a
Sample of generated networks R55-65: 4 connections - left, free connectivity - right
Figure 9.5b
Sample of generated networks R65-75: 4 connections - left, free connectivity - right
Search WWH ::
Custom Search