Freight Modelling

In the existing freight models, the choices of logistics facility locations and the selection of logistics facilities for routing the shipments are often treated without distinction although these two decisions are distinct and affected by different set of factors in reality. In this paper, we develop models of logistics facility choice that match truck trip ends with logistics facilities using a large urban freight survey data from the Tokyo Metropolitan Area. The models can be used to estimate truck traffic flows associated with transshipments. We categorize the urban portion of logistics chains into five types of movements and develop a total of 30 models to separately analyze six commodity groups. The results indicate that the proposed approach can capture the mechanism of the logistics facility selection by movement and commodity types. The tests for the reproducibility of the models warrant the future use of the models for urban freight demand analysis.

" Inequality " is a common observation about us, as members of society. In this article, we unify physics with economics by showing that the distribution of wealth is related proportionally to the movement of all the streams of a live society. The hierarchical distribution of wealth on the earth happens naturally. Hierarchy is unavoidable, with staying power, and difficult to efface. We illustrate this with two architectures, river basins and the movement of freight. The physical flow architecture that emerges is hierarchical on the surface of the earth and in everything that flows inside the live human bodies, the movement of humans and their belongings, and the engines that drive the movement. The nonuniform distribution of wealth becomes more accentuated as the economy becomes more developed, i.e., as its flow architecture becomes more complex for the purpose of covering smaller and smaller interstices of the overall (fixed) territory. It takes a relatively modest complexity

In following with previous papers on international trade technology, this paper will focus on the issues related to bulk trade of commodities and exchanges such as the Baltic Exchange for contract bidding of freight carriage. Ocean is by volume the largest mover of commodities and finished goods around the world in dry, liquid or containerized cargo. Navigable rivers as well as intra-costal waterways serve as the feeder means to major ports that serve the continental trade across oceans also known as blue waters such as the Mississippi river which runs through the heartland of grain production to ports in the Gulf of Mexico for export. Other important assets to the waterborne connections are the canals and the straits that become the limiting factors to the size of ships needing to get thru either because water depth or passage girth. A good example of this is the intra-coastal canal system that runs around the fan that forms the US Gulf of Mexico except Florida, and the eastern seaboard.

As centres of production and consumption, cities rely heavily on the mobility of freight for the provision of goods and services to residents, visitors, firms and organisations. Volumes of freight mobility are increasing and courier, express and parcel (CEP) services are growing rapidly with ongoing urbanisation and changes in consumption and shopping habits and delivery structures. Further change can be expected in light of the ongoing restructuring of logistics and supply chains and the rise of the smart city and vehicle automation. Yet the parcels, distribution centres, vehicles and pipelines that make up the systems of freight delivery often remain invisible in geographical studies of transport and mobilities. Similarly, policies to reduce the negative impacts of road freight transport are seldom focused at the city scale, and urban mobility is rarely prioritised in urban planning. In this session, we seek to address these gaps, through in-depth explorations of the social-spatialities of urban goods mobility. Topics of interest include, but are not limited to:

In this study we present an on-street parking model for downtowns in urban centers that incorporates the often-neglected parking demand of delivery trucks. The behavior of truck deliveries is distinctly different from commuter parking: trucks do not cruise for parking spaces, and demand for goods delivery is driven by customers and is practically inelastic to the delivery costs. We generalize the downtown on-street parking model from Arnott and Inci (2006) to study the relationship between passenger vehicles' parking and truck delivery behaviors, and provide tools for policy makers to optimize the trade-offs in parking space allocation, pricing, and aggregate network congestion. The social optimum can be obtained by solving a nonlinear optimization problem. The parking model is able to replicate the commuter-only scenario as a special case. It is shown that ignoring truck delivery behavior can significantly overestimate travel speeds and cruising stock. We applied the model to a case study of downtown Toronto and found that compared to a baseline scenario representative of Toronto in 2015, increasing parking fees from CAD $4/hour to nearly CAD $7.85/hour and assigning 4.1% of parking spaces to truck deliveries would eliminate cruising and truck double-parking, resulting in a social surplus gain of over CAD $14,304/hr/mi^2. In a first-best allocation scenario where total parking spaces can also change, we found that increasing total parking spaces by 18%, having 3.5% truck delivery allocation, and reducing parking fees to CAD $2.47/hr would eliminate cruising and double-parking while increasing social surplus by CAD $24,883/hr/mi^2. These model findings are along the same level of effect as demonstrated in the literature.

The article proposes the application of the System Dynamics theory to study the behavior of the costs of the roadway transportation of an important part of the Brazilian soybean crop production, from Mato Grosso (MT) as far the Port of Santos, in São Paulo (SP) state with an average of 2,000 km distance. This market is regulated basically by the offer and demand of the grains along the year, and its relation with the existing regional cargo firms that own a certain number of trucks. Many factors have influence on the freight price, freight rate and freight perception, and some of them are: inflation, profit, revenues, costs, volume of grains, capacity of the trucks, number of available trucks, number of needed trucks, the gap between the needed and the available number of trucks, and others. The problem is to contribute for the freight market behavior understanding analyzing the dynamical hypothesis using causal loops of and simulation of a model in Vensim ® to present some interesting results and conclusions. The time horizon is 20 years, the reference model is obtained from 7.5 years of data observation and the behavior of the price simulation follows with certain specific characteristics, the reference model presented. The profitability seems to compensate the investment and if there are companies facing financial troubles, it is not due to the aspects analyzed in the paper, and can be related to the management styles or other factors like the new transport made by individuals on the " spot " market.

One potential solution to this dilemma, the focus of this study, is to segregate truck traffic from passenger traffic by implementing exclusive truck facilities in key economic corridors. Exclusive truck facilities can be dedicated lanes on existing roadways or exclusive truck highways. The benefits of exclusive truck facilities to industry are obvious. Mobility can be improved, productivity gains for the logistics sector can be realized through the safe use of larger vehicle configurations, and safety improvements are possible by segregating trucks ...

Truck freight dominates the freight transportation. Motorways are especially important in serving truck freight demand due to more reliable travel times and favorable driving conditions, and result in different mode and route choice behaviors. Thus, it is important to understand the characteristics of truck commodity flows on motorways, which is the objective of this study. In the absence of commodity flow data in Turkey, the major source of freight data is the roadside axle load surveys conducted annually by the Turkish General Directorate of Highways (TGDH). Unfortunately, these surveys do not include observations on/around motorways, except for the year 2005, when 6299 trucks were surveyed at 19 locations on three motorway corridors in Turkey (Northern corridor connecting Edirne-İstanbul-Ankara, Aegean Corridor of İzmir-Aydın and southern corridors connecting Adana-Şanlıurfa). The following data was collected for each surveyed truck: Origin-Destination (O-D), commodity type, truck type, payload and location information. s. Therefore, truck circulation characteristics on different motorways corridors were studied to observe corridor based differences in commodity types, truck types, payloads, loading conditions and O-D patterns. The three motorway corridors serve many O-D pairs in the national level. Differences among the corridors are expected due to differences in regional economics and industries, as well existence of ports and border gates. The motorways are developed as corridors and are not total connected, yet, which may restrict the national flows on some of them.

In order to accommodate all concerning materials and still keep this paper length-readable it is made available in two parts. Freight rates in nodal format based on origin-destination published by carriers; leave out minor ports in the network. Using a minimal approach of 50 nodal points to cover the entire world, carriers can reduce the number of individual rates to (50)²= 2,500 for each service type that can easily be kept and updated in a data base thus, avoiding storage and updates of millions of rates. It fallows that with this recipe, carriers can use algorithms for covering the ports in the network not included in the nodes. The balancing act relies in developing an algorithm that works across all service types within the 50 nodes. For instance, if 4 service types are considered it would render 4*2,500= 10,000 rates with all its attributes. Even though this method can work for carriers, how does a freight broker keep track of all the possible combinations of routes and rates? How does a manufacturer find out about a particular freight rate before committing to a production-delivery contract? The list goes on to include traders, bankers, governments and individuals who need to know freight rates as accurately as possible in any part of the world in the age of the Internet with instantaneous response for a management board in session?

The availability of large sensor-generated transportation datasets has increased rapidly in many parts of the world, but it has not yet been fully leveraged by transportation modellers in the USA, Australia and elsewhere. Further, the low quantity and quality of data available for freight transportation planning model development requires modellers to extract as much information as possible from every dataset. On a different note, the general public and decision makers have been increasingly demanding of well-crafted visual depictions of data and model results, especially for GIS-enabled data, but such demand requires outputs of data processing that are not standard in the transportation/freight modelling practice. While GPS data from vehicles has the potential to provide a wealth of data for route choice behaviours, cross-validation of traditional vehicle classification counts, and vehicle movement animations, these areas have yet to be fully explored. In the case of freight modelling, the absence of sufficient data to support modelling makes it even more pressing to extract all possible information from datasets for freight-carrying vehicles. This means moving beyond the use of truck GPS data for simple travel time and speed measurements but rather developing methods to extract highly detailed truck activity data depicting route selection, trip chaining, and stop clusters, for instance. Before these detailed analyses for passenger or freight GPS data are possible, it is necessary to first reconstruct the vehicle routes based on their GPS data. This paper presents an algorithm for route reconstruction leveraging solely open source software, and presents a test case based on bus GPS data for a European city, as currently no truck GPS data is freely available for research purposes. The steps necessary for applying this algorithm to proprietary vehicle GPS data are also presented. OBJECTIVES AND MOTIVATIONS The last decade has seen a rapid increase in the availability of large transportation datasets generated by a variety of sensors such as Global Position Systems (GPS), cell phones, and in-pavement and roadside sensors. This robust data, however, has not been fully leveraged for development of either passenger or freight transportation planning models. Rather, these valuable data sets have seen limited use in transportation planning models as sources of network travel time information or more generally for national level performance measurement programs. In parallel, the quantity and quality of data available for freight transportation planning model development pales in comparison to what is routinely collected for developing passenger transportation models. This is due in part to privacy concerns across most industry sectors, which limits the type and level of detail of data made available. Also, unlike passenger travel, freight movements are the result of complex global supply chains and economic conditions which widen the scope of data needed to accurately model freight movements. Thus, it becomes more pressing to extract all possible information from the datasets that are currently available for freight-carrying vehicles. This means moving beyond the use of advanced datasets for simple travel time and speed measurements and instead developing methods to extract highly detailed truck activity data depicting route selection, trip chaining, stop clusters, and other trip characteristics.

In recent years the required fidelity of statewide freight forecasting models in the United States has grown significantly, with several models having analysis zones smaller than counties, those being the most disaggregate spatial levels for which public freight data sources are typically available in the USA. As freight models become more disaggregate, public agencies have begun to show interest in the seamless integration between this new generation of freight models with the typically more spatially disaggregate personal transportation models. However, these two types of models are still often constructed with very disparate aggregation levels, with personal transportation models having zones that are between 30 and 60 times more detailed than statewide freight models. This inconsistency is exemplified in California by the recently developed California Statewide Freight Forecasting Model (CSFFM) developed by the University of California, Irvine Institute of Transportation Studies (ITS-Irvine) and the equally novel California Statewide Travel Demand Model (CSTDM) developed by Cambridge Systematics Inc., both commissioned by the California Department of Transportation (Caltrans). The matrix disaggregation algorithm presented in this paper builds on the foundation of these models by proposing a matrix disaggregation procedure of the CSFFM that presents the following set of important characteristics:  Maintains consistency with CSFFM: matrices match when aggregated to CSFFM level;  Fast development (i.e. did involve complex and time consuming model estimation) ; and  Simple requirements: Requires only the computation of a skim matrix on CSTDM's network. The method present on this paper can be described as a factoring method, although largely more complex than the methods currently used for disaggregating freight demand in metropolitan model developments. In a nutshell, the algorithm presented in this paper involves the estimation of a direct demand model at the aggregate – CSFFM Freight Analysis Zone (FAZ) – level using only independent variables readily available at the disaggregate – CSTDM – zoning level as inputs, and the final truck matrices estimated by CSFFM as dependent variables. The CSFFM outputs can then be applied to CSTDM's zoning system with the resulting estimates being appropriately scaled to match the input CSFFM matrix (with one scaling factor per cell of the CSFFM truck trip table). In turn, the algorithm also includes a novel implementation of the stepwise linear regression method and designed to work under this model's specific conditions. Lastly, this procedure recognizes that there does not appear to be any currently available data to support behaviour-based modelling of freight generation and distribution at finer spatial levels such as the zoning system used in the CSTDM. Furthermore, no other procedure to integrate models with different spatial aggregation by seamlessly disaggregating matrices from a coarser aggregation level to a finer level was found in the literature, which makes this procedure a unique tool for application in other states and regions that operate passenger models at different aggregation levels from their freight counterpart.

Railroad transportation is one of the most efficient means used to move the Brazilian grains production to export ports. However, grain production has recently been growing at a faster rate than the railway infrastructure, leading to several operational restrictions for this mode of transportation. Since new investments take long to be completed, railroad operators seek to improve operational efficiency in order to ensure better utilization of existing resources. In this context, this paper examines some operational strategies that could be used to achieve increased capacity for a grain rail transportation system operating in a closed loop, by means of a simulation-optimization model. The proposed model consists of a traditional stochastic simulation model in which all decisions (assets distributions and queue management) were replaced by prioritization rules (very simple heuristics). As there were tested as many as 10 possible prioritization rules for each one of the eight stochastic processes replaced, it became a combinatorial problem to find the best set of prioritization rules, which led us to apply a procedure analogous to the Hooke & Jeeves algorithm (Hooke & Jeeves 1961). We provide the details of the whole modelling process from the developing of the stochastic simulation model used to characterize the system, including some very promising results. However, an especial attention is given to the prioritization rules that were incorporated into the model (as well as how they were matched together) and to the Hooke & Jeeves analogy used to search for the best set of prioritization rules for the system analysed.

Talk given at International Workshop on Activity-Travel Behavior Analysis and Multi-state Supernetwork Modeling hosted by HKSTS in 2014. Work in collaboration with Iris You and Stephen Ritchie, published in 2016 under doi 10.1080/23249935.2016.1189723 in Transportmetrica A special issue.