Test a wide variety of policies and infrastructure improvements, from pricing strategies to freight-specific facilities. Cube Cargo is the Cube functional library for freight forecasting, offering specific methodologies for studying freight demand using a commodity-based approach. Cube Cargo operates seamlessly with all Cube modules, including Cube Voyager and Cube Analyst for Matrix Estimation.
Image 1 - Cube Voyager – Cube Cargo complete integration
With Cube Cargo, you can add freight forecasting by leveraging your existing passenger data and models.
Many areas are recognizing the need to have a better understanding of freight movements, how they will change over time, and how they can take actions to plan for and to improve the flow of goods while mitigating negative financial, environmental and travel impacts. Key elements in freight planning are matrices of commodity and truck flows.
Cube Cargo forecasts:
Cube Cargo models three distinct freight segments:
Choose your commodity groups according to the level and quality of your data. Accurately model the movement of such diverse goods as milk, oil, grain, paper, and machine tools.
The diagram represents the typical hierarchy of regions for a Cube Cargo model. The study area is the smallest region in the diagram. The inner area is the largest enclosing region for which socioeconomic data is available. All of the places, not in the inner area, which can serve as the origin or destination of freight trips observed in the inner area, form the outer area.
Image 2 - Regional hierarchy
Cube Cargo uses a two-level zoning system:
Cube Cargo is the Cube module that allows comprehensive freight forecasting for urban, regional, state and national planning, to answer policy questions like:
Cube Cargo is comprised of six programs, each of them with a specific model implemented:
Image 3 - Cargo programs
The resulting matrices are assigned to the network to provide link-level truck flows by truck type. Assignments are done using Cube Voyager under Cube Base. The assignment models provide additional information such as vehicle miles or kilometers travelled, and average speed and volume by road type. This information can then be used to estimate environmental impacts.
Image 4 - Cube Cargo modelling steps in Application Manager
The generation model forecasts the number of tons, by commodity group, produced and consumed for each coarse-level zone. The productions are segmented into internal productions, which are to be transported to an internal zone, and exports, which are sent to external zones. Similarly, the consumptions are classified as internal or as imports.
Productions and consumptions are estimated using multivariate linear regression models estimated using local data.
The distribution models allocate the forecasted productions by commodity group from their zone of origin to their zone of consumption. The productions and consumptions are split into short and long-haul trips. Both trip types are distributed using gravity models using different generalized cost functions. Short trips are distributed using distance; long-haul trips are distributed using a composite cost of travel time, distance, and cost.
The modal choice models are applied on the long-haul trip matrices only, using multinomial logit choice functions. Short-haul trips are assumed to travel by road. Long-haul trips are split into truck, rail, inland waterway and modal combinations (combined transport).
Generalized cost functions are defined using local data for each combination of commodity group and mode and distance class. The modal choice functions incorporate time, distance and cost.
Transport logistics nodes (TLN) are places such as major goods yards, multimodal terminals, railway stations, and ports, where trip chaining occurs.
The Transport Logistics Node model examines the matrices created by the long-haul modal choice model and partitions them into direct transport and transport chain matrices.
The goods in the direct transport matrices will be transported directly from their initial origin to their final destination. The goods in the transport chain matrices are divided into two segments: from origin to the TLN and from the TLN to the destination. Of these two sections, one will be classified as long-haul and the other will be classified as short-haul. At this stage of the model, Cube Cargo has estimated the commodity flow matrices by product type and mode.
For each combination of mode and commodity group, the matrices are converted using gravity formulations to the fine level zone system.
This transition is made in order to produce truck vehicle matrices at a zone level sufficiently fine to provide estimations of link-level truck flows.
The vehicle model estimates the number of vehicle trips per day given the mode and commodity group matrices (tonnage) from the previous model steps.
The model iterates over all origins across all of the various matrices, by commodity class, and applies two models which separately model direct trips and touring vehicle trips
The results are combined to provide matrices of vehicle truck volumes by truck type for assignment.
In urban areas, there is a significant amount of local delivery and non-goods related truck traffic. This includes transport of relatively small amounts of goods and the transport of services.
The service traffic model generates local truck matrices for these purposes using linear regression generation models and gravity models for distribution.
The outputs from the Cube Cargo model are the freight matrices at coarse and fine level, and the vehicles matrices at fine level. Several analysis can be achieved by combining these matrices with the travel time and distance matrices, disaggregated by commodity type
The demand is then directly assigned to the roadway network using Cube Voyager, but a multimodal approach can also be implemented, for current or forecasting scenarios.
Image 5 - Example of analysis of the outputs obtained with Cube Cargo