Optimal operating rules for joint system of water supply reservoir and seawater desalination plant using genetic programming

Created by W.Langdon from gp-bibliography.bib Revision:1.3963

@MastersThesis{oai:repository.ust.hk:1783.1-81383,
  author =       "Yi Yang",
  title =        "Optimal operating rules for joint system of water
                 supply reservoir and seawater desalination plant using
                 genetic programming",
  year =         "2015",
  school =       "HKUST",
  address =      "Hongkong",
  keywords =     "genetic algorithms, genetic programming, water-supply,
                 saline water conversion, water resources development,
                 water reuse",
  bibsource =    "OAI-PMH server at repository.ust.hk",
  language =     "English",
  oai =          "oai:repository.ust.hk:1783.1-81383",
  URL =          "https://doi.org/10.14711/thesis-b1514746",
  URL =          "http://repository.ust.hk/ir/bitstream/1783.1-81383/1/th_redirect.html",
  size =         "99 pages",
  abstract =     "Due to climate change, population growth and
                 industrial development, there is increasing scarcity of
                 freshwater resources amidst rising demands. In view of
                 this, many coastal places are resorting to seawater
                 desalination as a means of supplementing existing
                 supplies from reservoirs. However, doing so introduces
                 a tradeoff between water supply reliability and cost,
                 as seawater desalination is relatively expensive
                 because of its high energy consumption. Although some
                 studies have been done to combine seawater desalination
                 with other options like reservoir and wastewater reuse
                 for supplying high water demand, they either emphasize
                 too much on economic cost rather than system operation,
                 or lack quantitative investigation into the operation
                 of an integrated or joint system. Thus, a comprehensive
                 model for the operation of a joint system to
                 systematically optimise both water supply reliability
                 and economic efficiency is required. In this study, an
                 optimisation model for the operation of a joint system
                 of a single reservoir and seawater desalination plant
                 was developed for urban water supply. The model aimed
                 to maximise water supply reliability while constraining
                 cost. Taking into account the existing storage of water
                 in the reservoir, the demand for water by various
                 sectors, and current and forecast future inflows to the
                 reservoir, two operating rules that interact with each
                 other were optimised for guiding the operation of the
                 reservoir and seawater desalination plant. Upon
                 attaining the optimal functions, both operational cost
                 and capital cost were calculated on an annual basis for
                 analysis. To solve the above operation model, a genetic
                 programming (GP) iterative tool was designed for the
                 joint system. Using the GPLAB toolbox in MATLAB,
                 genetic programming was applied in an iterative fashion
                 to generate optimal operational rules to govern the
                 releases from the reservoir and water production rates
                 of the desalination plant. In this manner, GP was
                 empowered to optimise the two rules simultaneously,
                 which would not be possible if using GP in a
                 conventional way. Results were obtained for a
                 semi-hypothetical case study in California and analysed
                 to prove the advantage of the joint system and
                 applicability of genetic programming for the purposes
                 of this study. The fitness value was found to have
                 improved by 33percent after 83 iterations in the
                 baseline case. It was demonstrated that due to the
                 assumption that the volume data of current inflows and
                 demands were affected by their volume data one and two
                 time periods before thus forecasting information might
                 be indirectly incorporated into the functions by the
                 incorporation of these variables into the functions.
                 The complex functions generated by the model can be
                 easily calculated using computer programs. The capital
                 cost consisted of 1/3 of the total cost with an
                 equilibrium point at around 500 million dollars per
                 month when it was allocated to each month. But the
                 water demands were too high to be fully met (70percent
                 met), leading to large budget carry overs. In terms of
                 the reservoir performance, the reservoir storage was
                 drawn down before every inflow peak. If the budget was
                 not enough for this expensive way of desalinating
                 water, it had to depend more on releasing water from
                 the reservoir, whose inflows fluctuated in all time
                 periods. As the capacity of desalination plant
                 increases, demand plays a more important role in
                 deciding how much water to be released from the
                 reservoir. The scale expansion of the existing seawater
                 desalination plant could be a very effective but costly
                 way to solve water scarcity problems in coastal city
                 water supply cases, while increasing the reservoir
                 capacity is the most efficient way to reducing water
                 shortages. And the fitness value kept increasing by
                 83.05percent when the reservoir capacity went up from
                 3000 million m$^{3}$ in the baseline case to 5000
                 million m$^{3}$. But still future work needs to be done
                 to incorporate more scenarios to prove the advantage of
                 the joint operation model together with the GP
                 iterative tool.",
  notes =        "oai:repository.ust.hk:1783.1-81383",
}

Genetic Programming entries for Yi Yang

Citations