Orientador: Flávio Keidi Miyazawa / Dissertação (mestrado) - Universidade Estadual de Campinas, Instituto de Computação / Made available in DSpace on 2018-08-25T12:28:53Z (GMT). No. of bitstreams: 1 Lopes_MauroCardoso_M.pdf: 3797752 bytes, checksum: c82bee131ad99d747e42150908135190 (MD5) Previous issue date: 2014 / Resumo: Investigamos uma variante do problema de localização e roteamento com relação de muitos-para-muitos concentradores que consiste em particionar o conjunto de vértices de um grafo em ciclos contendo exatamente um concentrador cada e determinar um ciclo adicional interligando todos os concentradores. Qualquer vértice do grafo pode ser um concentrador; faz parte do problema determinar quais vértices devem ser concentradores. Esse problema tem aplicações práticas relevantes em áreas como transporte urbano e redes de computadores. Desenvolvemos uma heurística baseada em busca local com operações de inserção, remoção e troca de vértices. Soluções iniciais são geradas de maneira aleatória, e suas vizinhanças são exploradas a fim de obter melhores soluções. Além disso, elaboramos um algoritmo exato com estrutura de branch-and-cut para a formulação em Programação Linear Inteira proposta. Restrições de capacidade e eliminação de caminhos são adicionadas como planos de corte, com algoritmos de separação baseados em árvores de corte mínimo e nas componentes conexas de um grafo suporte. Diversos experimentos computacionais mostram a capacidade de resolução do algoritmo exato para instâncias pequenas e da heurística para instâncias pequenas e médias. São comparados também os desempenhos para outras variantes do problema / Abstract: We investigate a variant of the many-to-many hub location-routing problem which consists in partitioning the set of vertices of a graph into cycles containing exactly one hub each and determining an extra cycle interconnecting all hubs. Any vertex of the graph can be a hub; it is part of the problem to determine which vertices should be hubs. This problem has relevant practical applications in areas such as urban transportation and computer networks. A local search based heuristic that considers add/remove and swap operations is developed. Initial solutions can be generated at random, and their neighborhoods are explored in order to get better solutions. Also a branch-and-cut approach that solves an integer formulation is investigated. Capacity and path elimination constraints are added in a cutting plane way, so the separation algorithms are based on the computation of min-cut trees and in the connected components of a support graph. Many computational experiments over several instances adapted from literature show the problem-solving capability of the exact algorithm for small instances and of the heuristic for small to medium-sized instances. We also compare the performance of other variants of the problem / Mestrado / Ciência da Computação / Mestre em Ciência da Computação

Orientador: Fernando José Von Zuben / Dissertação (mestrado) - Universidade Estadual de Campinas, Faculdade de Engenharia Elétrica e de Computação / Made available in DSpace on 2018-08-25T22:55:48Z (GMT). No. of bitstreams: 1 Raimundo_MarcosMedeiros_M.pdf: 764820 bytes, checksum: 80ad4c20c482ad09b06c3e07d1b2c240 (MD5) Previous issue date: 2014 / Resumo: O gerenciamento de cadeias de suprimento no mundo corporativo é de grande relevância prática e uma de suas versões é conhecida como problema de roteamento e estoque. Este trabalho propõe uma formulação linear-inteira genérica e flexível para este problema de otimização, assim como uma metodologia de solução. Nesta nova formulação proposta, algumas peculiaridades da rede de suprimentos podem ser especificadas como parâmetros de entrada, permitindo assim que o usuário seja capaz de realizar modificações na estrutura, na hierarquia e no elenco de restrições da cadeia de suprimentos, sem precisar refazer a formulação matemática associada. Com isso, é possível resolver uma grande diversidade de configurações do problema, sem a necessidade de adaptações junto à metodologia de solução. A natureza genérica e flexível da formulação linear-inteira se deve às seguintes propriedades, todas elas passíveis de serem definidas como parâmetros de entrada: (1) Todo nó da rede pode produzir ou consumir produtos; (2) Todo nó da rede pode enviar e receber produtos; (3) Decorrente das propriedades (1) e (2), a hierarquia de entrega fica generalizada, com o produto podendo passar por vários nós antes de ser consumido; (4) Restrições presentes na formulação garantem consistência, por exemplo, entre quantidade de produto entregue pelos fornecedores e recebida pelos consumidores; (5) Restrições presentes na formulação estão associadas a especificações que podem ser ativadas, como intervalo de tempo entre entregas. Os resultados experimentais contemplam soluções para múltiplas configurações do problema, todas representáveis pela formulação proposta e, portanto, todas resolvidas pela mesma metodologia de solução. Essas múltiplas configurações trabalhadas nos experimentos evidenciam os benefícios do emprego de uma formulação estendida para o problema de roteamento e estoque. Além disso, visando comparação com propostas alternativas disponíveis na literatura, tomou-se uma configuração específica e bem-estabelecida do problema, para a qual existe uma formulação própria e uma metodologia de solução dedicada. Neste experimento comparativo, chegou-se às mesmas soluções e, em algumas parametrizações, até a soluções de melhor qualidade / Abstract: Managing supply chains in the corporate world is of great practical relevance and one of its versions is named inventory routing problem. This work proposes a more generic and flexible linear-integer formulation for this optimization problem, together with a solution methodology. In the novel formulation proposed here, some peculiarities of the supply network can be specified as input parameters, thus allowing the user to make modifications to the structure, the hierarchy and the set of constraints in the supply chain, without having to rebuild the associated mathematical formulation. Therefore, it is possible to solve a wide variety of configurations of the problem without the need for adjustments in the solution methodology. The generic and flexible nature of the linear-integer formulation is due to the following properties, all of them being definable as input parameters: (1) Every node of the network can produce or consume products; (2) Every node of the network can send and receive products; (3) Due to properties (1) and (2), the hierarchy of delivery is generalized, with the product being able to pass through several nodes before being consumed; (4) Some restrictions of the formulation ensure consistency, for example, between the amount of product delivered by the suppliers and received by the consumers; (5) Some restrictions of the formulation are associated with specifications that can be activated, as the time interval between deliveries. The experimental results include solutions for multiple configurations of the problem, all representable by the proposed formulation and, as a consequence, all able to be solved by the same solution methodology. Those multiple configurations considered in the experiments highlight the benefits of employing an extended formulation for the inventory routing problem. Aiming at comparing to alternative proposals available in the literature, it was considered a specific and well-established configuration of the problem, for which there are a proper formulation and a dedicated solution methodology. In this comparative experiment, we came to the same solutions and, in some parameterizations, even better solutions / Mestrado / Engenharia de Computação / Mestre em Engenharia Elétrica

Orientador: Eduardo Candido Xavier / Tese (doutorado) - Universidade Estadual de Campinas, Instituto de Computação / Made available in DSpace on 2018-08-24T00:15:42Z (GMT). No. of bitstreams: 1 Silveira_JeffersonLuizMoisesda_D.pdf: 2236708 bytes, checksum: 8e569408c2f068347058e36031689c3a (MD5) Previous issue date: 2013 / Resumo: Neste trabalho estamos interessados em problemas de empacotamento e roteamento. Assumindo a hipótese de que P ? NP, sabemos que não existem algoritmos eficientes para resolver tais problemas. Além de algoritmos exatos, duas das abordagens para resolver tais problemas são Algoritmos Aproximados e Heurísticas. Nesta tese mostramos algoritmos baseados nestas três abordagens para ambos os problemas, de empacotamento e roteamento. Os dois primeiros problemas atacados foram generalizações de problemas clássicos de empacotamento: O problema da mochila bidimensional e o problema de empacotamento em faixas. Estes foram generalizados adicionando restrições na forma de carregamento e descarregamento dos itens no recipiente (restrições estas, que aparecem no contexto de problemas de roteamento). O terceiro problema é uma combinação de problemas de empacotamento e roteamento. Neste caso, atacamos uma generalização do clássico Pickup and Delivery Problem. Propomos os primeiros resultados de aproximação para algumas versões dos problemas de empacotamento supracitados. Além disto, apresentamos algumas abordagens práticas para o terceiro problema. As heurísticas foram avaliadas através de experimentos computacionais comparando os seus resultados com algoritmos exatos / Abstract: In this work we are interested in packing and routing problems. Assuming P ? NP, we have that there are no efficient algorithms to deal with such problems. Besides exact algorithms, two approaches to solve such problems are Approximation Algorithms and Heuristics. In this thesis we show algorithms using these three approaches for both packing and routing problems. The first two addressed problems are generalizations of classical packing problems: The Two Dimensional Knapsack problem and the Strip Packing problem. These problems were generalized by adding constraints on the way the items can be inserted/removed into/from the bin (These constraints appear in the context of routing problems). The third problem is combination of packing and routing problems. It is a generalization of the classical Pickup and Delivery problem. We propose the first approximation results for some packing problems. Besides that, we present some practical algorithms for the third problem. The heuristics were assessed through computational experiments by comparing their results with exact algorithms / Doutorado / Ciência da Computação / Doutor em Ciência da Computação

This thesis demonstrates practical using of vehicle routing problem with time windows (VRPTW) and its solution by heuristic method. There are described teoretical principles of integer models, mathematical definitions of VRP with one or more vehicles, VRPTW and some heuristics for VRP. The practical part is solution of VRP by heuristic nearest neighbor. Product distribution is planed according to the firm settings in Prague. I compare existing situation and computed solution that show benefits of using described methods in conclusion.

This diploma thesis deals with Ant Colony algorithms and their usage for solving Travelling Salesman Problems and Vehicle Routing Problems. These algorithms are metaheuristics offering new approach to solving NP-hard problems. Work begins with a description of the forementioned tasks including ways to tackle them. Next chapter analyses Ant Colony metaheuristic and its possible usage and variations. The most important part of the thesis is practical and is represented by application Ant Colony Optimization Framework. It is easily extensible application written in Java that is able to solve introduced problems. In conclusion this work presents analysis of solutions on test data.

This thesis focuses on a description of the split delivery vehicle routing problem (SDVRP), in which the restriction that each customer has to be visited exactly once is not assumed, contrary to the classical vehicle routing problem, and split deliveries are allowed. Considering the fact that the split delivery vehicle routing problem in NP-hard, a number of heuristic algorithms proposed in the literature are presented. Computational experiments are reported and the results show that the largest benefits of split deliveries are obtained in case of instances with fairly specific characteristics and also several drawbacks of implemented Tabu Search algorithm (SPLITABU) are point out.

There are many distribution tasks, that vary in a number of vehicles, time windows, dividend or undivided delivery or if it is static problems or dynamic problems. In this essay I focus just on few of them. At first I put my mind to simple statistic distribution task with one vehicle. Later I extend it with time windows, when the point can be served by one car and by many cars. In the essay it will be solved the real task of distribution and delivery transportation of infusion in the hospitals. Because of the difficulty of solution I have to use the heuristic methods - Method of nearest neighbour, Savings method and Insert method. All these methods are modified by capacity requirements, time windows and also observence of the rules, which the drivers have to keep during a distribution. The aim is to minimize the distance travelled by the vehicles. And at least the company could be recommended the best solution.

This thesis deals with the optimization of distribution of products to consumers based on a real case study of a particular company from Opava. For this purpose, a mathematical optimization model is used to illustrate the vehicle routing problem. The study will also offer an explanation on the relevancy of heuristic methods, mainly with respect to their application in solving real life situations analogous to the one surveyed. On the basis of chosen heuristic methods (i.e. the nearest neighbour algorithm and the savings algorithm) and having taken into account the restricting conditions of the company, four algorithms were designed. These four algorithms are programmed in Visual Basic for Applications MS Excel 2007. They are aimed at solving the real problems with the distribution of ordered products that the particular company must deal with. The thesis compares the results provided by an employee of this company, and the results presented by the designed algorithms.

The diploma thesis deals with the use of heuristic methods for solving messenger problems. The first part focuses on the proposed vehicle routing problems, especially on travelling salesman problems and the messenger problems. For individual problems are theoretically discussed mathematical models for variants with one or more vehicles and the expansion of the use of time windows and capacity constraints. The second part introduces heuristic algorithms for the method of nearest neighbour, savings method, insertion method and the exchange method on the travelling salesman problems. Then all of the mentioned algorithms are modified for use in three variants of messenger problem. First, the disposition of one vehicle, the second allows more messengers in the same starting point and the last option is more messengers at different starting points. The last part describes computational experiments and comparison of results provided by different methods. The diploma thesis includes application for solving messenger problems using savings methods programmed in Visual Basic for Application in MS Excel.

At the basic there is a connection between operation research and automotive logistics. The supply chain in this type of industry is hard to coordinate and flexibility is considered as a main key indicator. Companies increase their competitiveness through the optimization of supply chain and try to take control of every flow or a process. This idea goes throughout the departments. The main concept, which is about a connection of whole supply chain with information flow means, that there isn't need of stock and every input is highly available on time and in requested quantity and quality. This thesis describes inbound logistics and introduces some of the methods used for planning vehicle routes. In our situation which is based on car maker data set was chosen a group of a suppliers for the analysis. We used vehicle routing problem with some modifications, heuristics based on the covering problem. Founded solution was compared with the present state.