Nicolau Dionísio Fares Gualda

Integrated scheduling of buses and drivers

In Brazil, bus services, which have to meet extraordinary levels of demand and traffic congestion, are generally planned by a city transport authority, and franchised out to several private companies for operation. The nature of the scheduling problem is different from most of those described at previous workshops, because vehicle and driver schedules are developed simultaneously, but on a line by line basis. The present paper describes work carried out on representative Brazilian city schedules using enhancements of the Leeds scheduling systems reported at previous workshops. Although the precise nature of the problem varies among the cities studied, the approaches adopted in this research are similar. The Leeds scheduling software, adapted for the Brazilian situation, is used in a novel way to produce efficient combined bus and driver schedules, with buses being switched between lines to improve efficiency. Already savings have been indicated. The paper includes a detailed exposition of the processes used in one of the cities, serving to emphasize the practical problems which must be addressed in any implementation.

Modelagem de redes logísticas com custos de inventário calculados a partir da cobertura de estoque

A consideracao dos custos de inventario tem se tornado cada vez mais importante na analise dos trade-offs e na tomada de decisoes sobre redes logisticas. O artigo propoe uma metodologia para tratamento do custo de inventario a partir da cobertura de estoque esperada em cada instalacao, incorporada na modelagem matematica da localizacao de fabricas e/ou centros de distribuicao em redes logisticas com varios elos, representando uma nova versao do modelo apresentado em Hamad e Gualda (2008). As principais contribuicoes desta metodologia em relacao a outros metodos encontrados na literatura sao a facilidade de aplicacao do metodo, a inclusao de restricoes ligadas a capacidade de armazenagem e a consideracao do custo total de inventario, e nao apenas do custo relacionado aos produtos modelados.

Fleet maintenance scheduling with an ant colony system approach

The paper presents highlights of a doctoral research addressed to model and solve the problem of preventive maintenance scheduling of fleets of vehicles. The problem is formulated and several Ant Colony based approaches were proposed and tested considering different instances of the maintenance scheduling problem, including applications related to the preventive maintenance of an aircraft fleet belonging to the Brazilian Air Force. The most successful approach has shown to be the one based on ACS with a specific local search procedure and inspired on the application of ACO to the timetabling problem. Details on the problem formulation, on the proposed approaches, and on the performed tests and applications are presented.

Modeling of logistic networks with seasonality: influence of carrying cost and ICMS credit

This paper presents an approach to solve the location problem in a global network with special focus on stocks and tax management. The methodology was applied successfully in large transnational agribusiness companies operating in Brazil. The results show that storage capacity constraints and inventory carrying costs, two components not usually included in location models, have a great correlation with the definition of the number of Distribution Centers. The methodology, from the installation of new facilities in different States or countries in MERCOSUR, improved the cash flow as much as US

Global Sourcing Approach to Improve Cash Flow of Agribusiness Companies in Brazil

22 million and showed the negative effect of the ICMS legislation that drives to a logistics cost increase to improve the cash flow. It is demonstrated that the model must cover at least four echelons in the chain to solve the network problem of companies with credit of ICMS.

Long Haul Trucks Dynamic Assignment with Penalties

We developed a multiechelon location-inventory modeling approach and applied it to redesign the global network of four large transnational agricultural companies operating in Brazil. The results show that tax planning is the most significant factor in minimizing the operating cash outflow and that inventory carrying costs play a more important role than transfer freight, handling, and fixed costs. Distribution centers and production plants in countries within the Mercosur are excellent targets for cash outflow reduction and service level improvement. Furthermore, contrary to common sense, reducing operating costs can sometimes increase cash outflow in companies with large volumes of tax credit in Brazil. Our methodology generated processes that helped one company reduce its logistic cash outflow by 49 percent and its logistics costs by

Um algoritmo de geração de arcos para o problema de programação de veículos

10 million.

Model for Facilities or Vendors Location in a Global Scale Considering Several Echelons in the Chain

Long haul transportation system operational planning implies to solve a capacitated dynamic network optimization problem, aiming to perform the freight movements in an efficient and effective way, utilizing the available transportation capacity. This work employs an approximate adaptive dynamic programming to solve this kind of problem, introducing a network modeling to manage demands not attended, with the payment of penalties, or the utilization of third parties transportation firms to avoid the payment of such penalties. The proposed methodology employs a dynamic programming model applied to each stage of the problem instead of a network global optimization procedure. A heuristic is used to estimate the value of future state functions transferred from a current stage to future ones. This technique greatly reduces the quantity of involved variables, allowing the utilization of more realistic mathematical models on a longer planning horizon. Results from a successful application of the model with the utilization of the Jensen Network Solver are presented.

Vehicle Scheduling with Network Flow Models

Este trabalho retrata a continuidade de uma pesquisa de doutorado voltada para a utilizacao de modelos de fluxo em redes na resolucao de problemas de programacao de veiculos (de onibus, em particular). Nesta etapa do trabalho foi aplicada a tecnica de Geracao de Arcos ao modelo de pseudo-designacao para se obter uma representacao eficiente de problemas reais, apoiada na utilizacao do algoritmo out-of- kilter para otimizacao de fluxo em redes. Com esta abordagem foi possivel reduzir o numero de arcos da rede, reduzir o tempo de resolucao e ainda incluir restricoes praticas, tornando o modelo mais adequado a realidade. A metodologia apresentada foi testada em problemas reais das cidades de Reading - Reino Unido e Sorocaba Brasil, e os resultados foram comparados com aqueles produzidos pelo sistema heuristico BOOST, desenvolvido pelo grupo de programacao de veiculos e tripulacao da Universidade de Leeds no Reino Unido.