Environmental Engineering Reference
In-Depth Information
The reliability is disproportional to the increase of the system resistance i.e. hydraulic loss.
This has proven to be mostly true, but here as well, the correlation is not fully consistent.
The most economic design shall always be cheaper than the most reliable design. There is a
threshold velocity and/or network resistance that can be taken as a border between the
increase of investment- or operational costs, resulting in the most effective reliability
improvement. This hypothesis has also not been proven. The cheapest, in this case the GA-
optimised design, showed in quite some cases the higher resilience than the more expensive
design comprising high operational- and low investment costs. Moreover, a unique threshold
velocity or network resistance does not exists, but the diagrams comparing the investment
and operational costs clearly showed a trend of reliability saturation, making further
investment into the network renovation or operation from the perspective of reliability
improvement, useless.
10.3
RECOMMENDATIONS FOR FURTHER RESEARCH
The consistency of some of the results in this study makes further research on reliability
patterns justified. Nevertheless, this should be done with more substantial integration of
hydraulic and geometrical parameters that influence network resilience. The statistical
analyses done hint some patterns but mostly visible in extreme cases of optimised diameters
or increased demand. Despite large number of simulations, the analysed samples could be
considered too homogeneous and still small i.e. statistically insignificant. To continue with
this approach, many more case networks of different characteristics will have to be analysed.
In parallel, an attempt should be made to build more solid mathematical foundation into the
concept, than was the case in this research. Secondly, the economic considerations of
reliability stayed short of sensitivity analysis of the economic parameters. More time should
be spent to verify the related conclusions, also adding the results of EPS hydraulic analysis to
the considerations.
As for the tool, NEDRA package showed robustness that can with some further adaptations
make it suitable for generation and processing of much larger number of network layouts,
say, 100,000. Even in this version, an option of generating more versatile layouts within the
same run is possible, by skipping some of the matrix columns in the process of non-random
generation. Nevertheless, this feature has not been sufficiently tested in this research. Some
degree of simplification of network generation settings is certainly possible and preferable as
the current setup of menus is not quite user friendly. Nevertheless, this is just the first version
of the software that has showed potential to be further improved and used in practice.
Regarding the network generation tool for the design purposes i.e. using the network
template, a further improvement of the algorithm should consider an alternative of the
template decomposition or instead, an option of filtering the layouts based on the nodal
connectivity in the template, during the process of generation and not after it has been
completed. Last but not the least, the further testing of GA-optimisation module is
recommended, also by considering an options of multi-criteria optimisation. That aspect was
not considered as an objective of this research, but some of the results ask for more
explanation.
Search WWH ::




Custom Search