Banca de QUALIFICAÇÃO: PEDRO OSVALDO ALENCAR REGIS

Uma banca de QUALIFICAÇÃO de MESTRADO foi cadastrada pelo programa.
DISCENTE : PEDRO OSVALDO ALENCAR REGIS
DATA : 13/09/2019
HORA: 09:00
LOCAL: Sala 231- CTEC
TÍTULO:

A variable neighborhood search algorithm for the water distribution network sizing problem.

PALAVRAS-CHAVES:

Metaheuristic. Optimization. Mixed integer nonlinear programming. Water distribution network.

PÁGINAS: 85
GRANDE ÁREA: Engenharias
ÁREA: Engenharia de Produção
RESUMO:

Water is undoubtedly fundamental to life on earth. For humans, clearly, water has always played a key role in its presence throughout its history. Records of evidence or use and management of water in human lives around 9000 BC. About 4,000 years was built or the first water distribution system in India. The relationship between disease and poor sanitation is well described in the literature. Water, sanitation and hygiene have the potential to prevent at least 9.1% of the global disease burden and 6.3% of all global deaths. In 2015, 4.7 billion people made use of piped water, while 2.1 billion do not, therefore, there are many people who still need piped water. For universal sanitation in Brazil, water and sewage the cost is $ 303 billion over 20 years. A water distribution network comprises approximately 50% to 75% of the value of the water supply system work, it is the more expensive part of the work. Using optimization techniques can result in a cost savings of 20 to 30%. This problem is known and is proven in the literature to be NP-hard. So because it’s NP-hard, exact algorithms for midsize networks aren’t viable for use, so heuristics and metaheuristics are used to try to get good solutions to the problem, not the best one possible for the problem. Although the problem have been studying, since before the existence of computers, the existing solutions are not satisfactory by a number of factors: the metaheuristics used do not perform well for real size networks, most of the metaheuristics used are unnecessarily complicated, and yet when testing these metaheuristics this is done in a small number of test networks that are unrepresentative of reality and because they are few are not enough in number to make rigorous performance comparisons of what was developed. Due to the problems listed, the objective of this work is to develop a variable neighborhood search algorithm (VNS) for the problem of optimized sizing of water distribution networks. This implementation of VNS will make use of the principle Less is More Approach (LIMA), which approach is supported by the maximum performance of metaheuristic maintaining its simplicity. VNS is a metaheuristic that enables the LIMA. The developed algorithm will be tested in a wide set of networks, more than 200 networks, obtained from the literature review, while most studies rarely test the algorithms developed in more than 5 networks. As results at the end of this work is expected: the development of a VNS algorithm for the problem of optimized water distribution network design, new and possibly better results for the tested networks, it is still expected to have tested the algorithm developed in the largest number of existing networks as possible to better prove the performance of the developed algorithm.

MEMBROS DA BANCA:
Presidente - 2993322 - EVERTON SANTI
Interno - 1142787 - JOSE ALFREDO FERREIRA COSTA
Externo ao Programa - 1845280 - SERGIO QUEIROZ DE MEDEIROS