<div>Bike sharing system (BSS) is growing worldwide. Although bike sharing is viewed as a sustainable transportation mode, it still has environmental footprints from its operation (e.g., bike rebalancing using automobiles) and upstream impacts (e.g., bike and docking station manufacturing). Thus, evaluating the environmental impacts of a BSS from the life cycle perspective is vital to inform decision making for the system design and operation. In this study, we conducted a comparative life cycle assessment (LCA) of station-based and dock-less BSS in the U.S. The results show that dock-less BSS has a greenhouse gas (GHG) emissions factor of 118 g CO2-eq/bike-km in the base scenario, which is 82% higher than the station-based system. Bike rebalancing is the main source of GHG emissions, accounting for 36% and 73% of the station-based and dock-less systems, respectively. However, station-based BSS has 54% higher total normalized environmental impacts (TNEI), compared to dock-less BSS. The dock manufacturing dominants the TNEI (61%) of station-based BSS and the bike manufacturing contributes 52% of TNEI in dock-less BSS. BSS can also bring environmental benefits through substituting different transportation modes. Car trip replacement rate is the most important factor. The results suggest four key approaches to improve BSS environmental performance: 1) optimizing the bike distribution and rebalancing route or repositioning bikes using more sustainable approaches, 2) incentivizing more private car users to switch to using BSSs, 3) prolonging lifespans of docking infrastructure to significantly reduce the TNEI of station-based systems, and 4) increasing the bike utilization efficiency to improve the environmental performance of dock-less systems.</div><div>To improve the design of current BSS from the life cycle perspective, we first proposed a simulation framework to find the minimal fleet size and their layout of the system. Then we did a tradeoff analysis between bike fleet size and the rebalancing frequency to investigate the GHG emission if we rebalance once, twice and three times a day. The optimal BSS design and operation strategies that can minimize system GHG emission are identified for a dock-less system in Xiamen, China. The results show that at most 15% and 13% of the existing fleet size is required to serve all the trip demand on weekday and weekend, if we have a well-designed bike layout. The tradeoff analysis shows that the GHG emission may increase if we continue to reduce the fleet size through more frequent rebalancing work. Rebalancing once a day during the night is the optimal strategy in the base scenario. We also tested the impacts of other key factors (e.g., rebalancing vehicle fleet size, vehicle capacity and multiple depots) on results. The analysis results showed that using fewer vehicles with larger capacity could help to further reduce the GHG emission of rebalancing work. Besides, setting 3 depots in the system can help to reduce 30% of the GHG emission compared with 1-depot case, which benefits from the decrease of the commuting trip distance between depot and the serve region.</div>

Este trabalho trata do problema de roteirização de veículos para o abastecimento de linhas de produção, o qual pode ser entendido como uma particularização do problema clássico de roteirização de veículos (VRP Vehicle Routing Problem). Neste problema, peças estão armazenadas em um estoque central, chamado de supermercado, de onde são transferidas para pontos de uso localizados ao longo da linha de produção. O ritmo de fabricação na linha de produção é suposto constante, o que torna periódica a necessidade de reposição das embalagens com peças. Uma frota de rebocadores transporta as embalagens, dispostas sobre plataformas com rodas puxadas pelo mesmo e configurando um comboio. O objetivo do problema é roteirizar a frota de rebocadores, maximizando sua utilização e garantindo o atendimento da demanda gerada pela linha de produção. O problema é comum a muitas empresas de manufatura de bens de consumo e possui impacto direto nos custos operacionais. A literatura sobre o tema é escassa e as soluções empregadas na indústria habitualmente se baseiam na experiência prática de operadores ou responsáveis pela movimentação de materiais. Este trabalho propõe uma heurística para obtenção de uma solução para o problema, baseada em métodos de inserção. A heurística proposta foi aplicada a um caso na indústria automobilística e a comparação entre a solução obtida e aquela formulada por operadores demonstrou ganho no número de rotas. / This work studies the routing of vehicles for material delivery to assembly lines, which consists of a generalization of the classic Vehicle Routing Problem (VRP). In this problem, parts are stored in central depot called supermarket and from where they are distributed to points of use placed along the production line. Production rate in the assembly line is considered constant, which means that parts are delivered to points of use periodically. A fleet of tow cars transfers the boxes or containers of parts using wheeled towed carts. The objective of this problem is to route the fleet of the tow cars maximizing their utilization and fulfilling the assembly line demand for parts. This problem is common to several companies and has direct impacts in material handling costs. The theme is poorly explored in routing studies and many companies use operator experience to configure tow cars routes. This work proposes a heuristic based on insertion methods to find a solution for the problem. The heuristic was applied to a real problem and resulted in the reduction of the number of routes when compared to former operator solution.

Two-echelon vehicle routing problems which contain synchronization between vehicles can be deeply impacted by time uncertainty, because one vehicle's delay can propagate to other vehicles. In this paper, we evaluate the deterministic solution of such a problem based on simulated travel time scenarios. The information obtained by simulation is incorporated in the optimization procedure iteratively. Computational results show that the degree of synchronization in an instance is directly correlated with the potential improvements by reoptimization. We present findings on the number of travel time scenarios required to obtain a representative picture of the stochastic solutions. In addition, we demonstrate that time dependent travel times can be aggregated on a city-wide level and linearized as a function of free flow times without major loss of reliability.

O transporte de cargas desempenha um papel fundamental nos sistemas logísticos uma vez que possibilita o fluxo de produtos entre os elementos dos canais de distribuição, além de representar um elevado percentual de participação nos custos logísticos totais. Desse modo, a eficiência dos serviços de transporte depende diretamente de um adequado processo de planejamento em nível estratégico, tático e operacional. O transporte rodoviário, em particular, apresenta em nível operacional, problemas de elevada complexidade como a roteirização e o carregamento de veículos. Durante os últimos 50 anos estes problemas vêm sendo objeto de intensos estudos e desenvolvimentos, porém de forma separada, conduzindo a resultados difíceis de serem implementados nas situações práticas, devido à grande interdependência existente entre roteirização e carregamento de veículos. A resolução integrada destes dois problemas é recente na literatura, o que abre um campo bastante promissor para novos desenvolvimentos. Assim, esta tese tem como objetivo propor uma abordagem integrada de resolução para o problema de roteirização e carregamento de veículos onde, além das restrições de carregamento tridimensionais, as restrições de tempo de ciclo e distribuição de peso nos eixos são consideradas. Este modelo adota uma abordagem aproximativa que combina o uso de heurísticas para construção dos carregamentos com a metaheurística de Busca Tabu para melhoria dos roteiros de entrega. Para execução dos testes computacionais é proposto um conjunto de 12 instâncias que buscam simular situações práticas encontradas nas operações de entrega feitas por empresas de transporte e prestadores de serviços logísticos. A partir desse conjunto de instâncias são construídos cenários que avaliam o uso de diferentes tipos de veículos e os respectivos custos operacionais. Testes complementares realizados com um conjunto de 27 instâncias disponíveis na literatura, demonstram a capacidade do modelo proposto de atingir boas soluções em um tempo de processamento aceitável. / Freight transportation plays a fundamental role in logistics systems, once it enables the product flow between distribution channels elements, besides to represents a high percentage of participation on the total logistics costs. In this way, transport service efficiency depends directly of a suitable process of planning in strategic, tactical and operational level. The road transportation, in particular, shows in operational level high complexity problems as routing and loading of vehicles. During the last 50 years, these problems has been object of deep studies and developments, but in a separate manner, leading to results that are difficult to implement in practical situation due to the interdependence between routing and loading of vehicles. The integrated resolution of these two problems is recent in the literature, what opens a quite promising field for new developments. Thus, this thesis has as objective to propose an integrated resolution approach for the vehicle routing and loading where, besides the three-dimensional loading constraints, constraints, the time cicle and the distribution load over de axles are considered. This model adopts an approximated approach that combines the use of heuristics for loading construction with Tabu Search metaheuristics to improve the delivery routes. For computational tests execution is proposed a 12 instances set that tries to simulate practical situation found in delivery operations made by freight firms and thirty party logistics. From this instance set, it is built scenarios that evaluate the use of different vehicle types and the respectives operational costs. Complementary tests made with a 27 instances set available in literature show the capacity of proposed model to get good solutions in an acceptable computational time.

Este trabalho apresenta uma variação do problema de roteamento de veículos que permite o uso de veículos com múltiplos compartimentos. A necessidade de veículos com múltiplos compartimentos surge com frequência em aplicações práticas quando uma série de produtos, que possuem diferentes qualidades ou tipo, precisam ser transportados mas não podem ser misturados. Este problema é chamado na literatura de roteamento de veículos com múltiplos compartimentos (PRVMC). Nós propomos uma heurística busca tabu implementada em uma busca local iterada para resolver este problema. Experimentos foram feitos para avaliar a performance da busca tabu iterada e os resultados obtidos foram comparados com os resultados disponíveis na literatura. O algoritimo proposto é capaz de encontrar soluções melhores e em menos tempo de processamento que as heurísticas existentes. / We study a variant of the vehicle routing problem that allows vehicles with multiple compartments. The need for multiple compartments frequently arises in practical applications when there are several products of different quality or type, that must be kept or handled separately. The resulting problem is called the multi-compartment vehicle routing problem (MCVRP). We propose a tabu search heuristic and embed it into an iterated local search to solve the MCVRP. In several experiments we analyze the performance of the iterated tabu search and compare it with results from the literature. We find that it consistently produces solutions that are better than existing heuristic algorithms.

L’interconnexion croissante dans les systèmes de transports a conduit à la modélisation de nouvelles contraintes, dites contraintes de synchronisation, dans les problèmes de tournées de véhicules. Dans cette thèse, nous nous intéressons à deux cas dans lesquels ce type de problématiques apparaît. Dans un premier temps, nous proposons une méthode heuristique pour un problème à deux échelons rencontré pour la distribution de marchandises en ville. Dans un second temps, nous étudions l’intégration d’un cross-dock dans des tournées de collectes et livraisons. Une première contribution à ce sujet concerne le problème de tournées de véhicules avec cross-docking, et une seconde contribution intègre, en plus, des contraintes de ressources au cross-dock dans le problème de routage. Une méthode pour un problème de chargement 3D, étudié lors d’un stage doctoral en entreprise, est également présentée. / Transportation systems are more and more interconnected, this has lead to a new kind of constraints, called synchronization constraints, in vehicle routing problems. In this thesis, we study two cases in which this type of constraints arises. First, we propose a heuristic method for a two-echelon problem arising in City Logistics. Second, we study the integration of a cross-dockin pickup and delivery vehicle routing problems. To that end we propose a matheuristic for the vehicule routing problem with cross-docking, and we propose an extension of this problem that integrates specific resource synchonization constraints arising at the cross-dock. A method for a 3D loading problem is also presented.

Apesar de serem utilizados com sucesso em problemas de roteirização clássicos como o do caixeiro-viajante e o de roteirização de veículos com janelas de tempo, os algoritmos genéticos não apresentavam bons resultados nos problemas de roteirização de veículos sem janelas de tempo. Utilizando-se de uma tendência recente de hibridização de algoritmos genéticos, Prins (2004) elaborou um algoritmo para o problema de roteirização de veículos sem janelas de tempo, monoperíodo, e que obrigatoriamente atenda a todos os clientes cujos resultados, quando aplicado a instâncias de Christofides et al. (1979) e de Golden et al. (1998), são comparáveis aos melhores códigos elaborados com base na busca tabu. Diferentemente da maioria dos algoritmos genéticos apresentados para solução de problemas de roteirização de veículos, no método desenvolvido por Prins (2004) o cromossomo é composto apenas pelos pontos a serem atendidos, não contendo delimitadores de rotas. Estas são definidas a partir de um método de particionamento do cromossomo. Este trabalho implementa o algoritmo descrito por Prins (2004) e propõe a este melhorias em diversas de suas etapas, como inicialização, operação de crossover, operação de mutação, reinicialização e particionamento do cromossomo. As alterações implantadas são aplicadas às instâncias de Christofides et al. (1979) e comparadas com o algoritmo inicial em termos de qualidade de solução e tempo de processamento. Finalmente, é elaborado um algoritmo genético que contempla as alterações que obtiveram resultados positivos. / In the Vehicle Routing Problem (VRP) we seek for a set of minimum-cost vehicle routes for a fleet of identical vehicles, each starting and ending at a depot, such that each customer is visited exactly once and the total demand of any route does not exceed the vehicle capacity. Several families of heuristics have been proposed for the VRP. They can be broadly classified into two main classes: classical heuristics developed between 1960 and 1990, and, more recently, metaheuristics. Among them, tabu search plays an key role, being acknowledged by most authors as the most successful approach for the VRP. In the literature some successful implementations of metaheuristic Genetic Algorithm (GA) can be found for classic routing problems such as the traveling salesman and vehicle routing problems with time windows. However, until recently, the same did not apply for the VRP. In this thesis we develop a genetic algorithm without trip delimiters, and hybridized with a local search procedure, for the solving the VRP, which is based on the work of Prins (2004). At any time, a chromosome can be converted into an optimal VRP solution (subject to chromosome sequence) by means of a splitting procedure, in which the chromosome sequence, representing a giant tour, is partitioned into feasible routes in terms of vehicle capacities. Starting with the procedure originally proposed Prins (2004), we then introduce new improvements in terms of the different components of the GA, aiming to obtain improved solutions. These include how we determine the initial population, different partitioning approaches, alternative reproduction (crossover) processes, a granular tabu mechanism similar to the one proposed by Toth and Vigo (2003 and, finally, in changes in the reinitialization process, aiming to reestablish diversity. Computational experiments are presented, based on the 14 classical Christofides instances for the VRP. The results show that the proposed improved versions of the GA allow us to obtain better solutions when compared to the original approach by Prins (2004).

Agricultural producers consider utilizing multiple machines to reduce field completion times for improving effective field capacity. Using a number of smaller machines rather than a single big machine also has benefits such as sustainability via less compaction risk, redundancy in the event of an equipment failure, and more flexibility in machinery management. However, machinery management is complicated due to logistics issues. In this work, the allocation and ordering of field paths among a number of available machines have been transformed into a solvable Vehicle Routing Problem (VRP). A basic heuristic algorithm (a modified form of the Clarke-Wright algorithm) and a meta-heuristic algorithm, Tabu Search, were employed to solve the VRP. The solution considered optimization of field completion time as well as improving the field efficiency. Both techniques were evaluated through computer simulations with 2, 3, 5, or 10 vehicles working simultaneously to complete the same operation. Furthermore, the parameters of the VRP were changed into a dynamic, multi-depot representation to enable the re-route of vehicles while the operation is ongoing. The results proved both the Clarke-Wright and Tabu Search algorithms always generated feasible solutions. The Tabu Search solutions outperformed the solutions provided by the Clarke-Wright algorithm. As the number of the vehicles increased, or the field shape became more complex, the Tabu Search generated better results in terms of reducing the field completion times. With 10 vehicles working together in a real-world field, the benefit provided by the Tabu Search over the Modified Clarke-Wright solution was 32% reduction in completion time. In addition, changes in the parameters of the VRP resulted in a Dynamic, Multi-Depot VRP (DMDVRP) to reset the routes allocated to each vehicle even as the operation was in progress. In all the scenarios tested, the DMDVRP was able to produce new optimized routes, but the impact of these routes varied for each scenario. The ability of this optimization procedure to reduce field work times were verified through real-world experiments using three tractors during a rotary mowing operation. The time to complete the field work was reduced by 17.3% and the total operating time for all tractors was reduced by 11.5%. The task of a single large machine was also simulated as a task for 2 or 3 smaller machines through computer simulations. Results revealed up to 11% reduction in completion time using three smaller machines. This time reduction improved the effective field capacity.

When a package is shipped, the customer often requires the delivery to be made within a particular time window or by a deadline. However, meeting such time requirements is difficult, and delivery companies may not always know ahead of time which customers will need a delivery. In this thesis, we present models and solution approaches for two stochastic last-mile delivery problems in which customers have delivery time constraints and customer presence is known in advance only according to a probability distribution. Our solutions can help reduce the operational costs of delivery while improving customer service. The first problem is the probabilistic traveling salesman problem with time windows (PTSPTW). In the PTSPTW, customers have both a time window and a probability of needing a delivery on any given day. The objective is to find a pre-planned route with an expected minimum cost. We present computational results that characterize the PTSPTW solutions. We provide insights for practitioners on when solving the PTSPTW is beneficial compared to solving the deterministic analogue of the problem. The second problem is the same-day delivery problem (SDDP). The SDDP is a dynamic and stochastic pick-up and delivery problem. In the SDDP, customers make delivery requests throughout the day and vehicles are dispatched from a warehouse or brick and mortar store to serve the requests. Associated with each request is a request deadline or time window. In order to make better-informed decisions, our solution approach incorporates information about future requests into routing decisions by using a sample scenario planning approach with a consensus function. We also introduce an analytical result that identifies when it is beneficial for vehicles to wait at the depot. We present a wide range of computational experiments that demonstrate the value of our approaches.

Data-driven decision making has become more popular in today’s businesses including logistics and vehicle routing. Leveraging historical data, companies can achieve goals such as customer satisfaction management, scalable and efficient operation, and higher overall revenue. In the management of customer satisfaction, logistics companies use consistent assignment of their drivers to customers over time. Creating this consistency takes time and depends on the history experienced between the company and the customer. While pursuing this goal, companies trade off the cost of capacity with consistency because demand is unknown on a daily basis. We propose concepts and methods that enable a parcel delivery company to balance the trade-off between cost and customer satisfaction. We use clustering methods that use cumulative historical service data to generate better consistency using the information entropy measure. Parcel delivery companies route many vehicles to serve customer requests on a daily basis. While clustering was important to the development of early routing algorithms, modern solution methods rely on metaheuristics, which are not easily deployable and often do not have open source code bases. We propose a two-stage, shape-based clustering approach that efficiently obtains a clustering of delivery request locations. Our solution technique is based on creating clusters that form certain shapes with respect to the depot. We obtain a routing solution by ordering all locations in every cluster separately. Our results are competitive with a state-of-the-art vehicle routing solver in terms of quality. Moreover, the results show that the algorithm is more scalable and is robust to problem parameters in terms of runtime. Fish trawling can be considered as a vehicle routing problem where the main objective is to maximize the amount of fish (revenue) facing uncertainty on catch. This uncertainty creates an embedded prediction problem before deciding where to harvest. Using previous catch data to train prediction models, we solve the routing problem a fish trawler faces using dynamically updated routing decisions allowing for spatiotemporal correlation in the random catch. We investigate the relationship between the quality of predictions and the quality of revenue generated as a result.