International Choice Modelling Conference, International Choice Modelling Conference 2017

Font Size: 
Commodity-based heavy vehicle model for Greater Sydney
Richard B. Ellison, Collins Teye, David A. Hensher

Last modified: 28 March 2017

Abstract


1. Introduction

Almost without exception, the design, planning and management of the road network is determined by travel demand largely derived from passenger travel models. The neglect of freight modelling or development of analysis tools in the past was typically justified on the assumption that freight constitutes a very small fraction of the daily road traffic. The difficulty and the cost of collecting freight data has also contributed in large measure to the general absence of freight model systems that are sophisticated and behaviourally appropriate.  However, with the growing acknowledgment of the importance of freight to both the local and national economies and also the disproportionate impacts of trucks on congestion, pollution, accidents and other road hazards, there is stronger call for a better understanding of the freight system (Hensher and Figliozzi 2007). Network planners or managers are also keen to understand freight movements and their impacts on road capacity so as to better manage congestion and plan for the future. To achieve this, we need innovative freight models (Hensher and Figliozzi 2007) capable of capturing all the key behavioural responses and the interaction of actors within the freight system. Freight models are critical to assessing national, regional and local road capacities, economic development initiatives, and for informing the transport planning process.

Freight is however difficult to model due to several factors, among them the non-availability of data on commodities, shipments, demand and production cycles; the lack of understanding about the actors and how they interact on the supply and logistics corridors, and the broad economic influences on local freight movements (Hensher  and Figliozzi  2007).  These limitations mean that in the short to medium term modellers may not have the resources needed to develop a freight model system with the level of detail and richness similar to the current state of the art in passenger modelling (e.g., activity-based models) to answer policy questions of interest. The current practice in freight modelling is therefore based on the efficacy of building models using existing data sources to answer as many important policy questions as possible.

Drawing on existing commodity-based freight models that have incorporated the generation and attraction of commodities into freight models (Wisetjindawat et al., 2006; Holguin-Veras and Patil (2008) and models that incorporate interaction between agents in the supply chain (Hensher and Pucket, 2005; Pucket et al, 2007), this paper presents a novel approach, implemented for the Sydney metropolitan area,  of combining diverse pieces of data from various sources on commodity flows, vehicle usage, and distribution of freight by time of day, trip length distribution by commodity type and other relevant information, to construct a model framework of linked logit models suitable for explaining the main interactions between the key stakeholders (shipper-carrier-consumer) in the freight system, and how changes in land-use and transport network conditions affect commodity productions and distributions.  To achieve this, the Greater Sydney Metropolitan Area (SGMA) is divided into commodity production and consumption areas where commodities can be seen as being produced and transported to their consumption areas for consumption. The key models in the framework include a commodity generation chain model (CGCM) model at the national level, a commodity generation model (CGM) at the zonal level of the study area, a commodity distribution model (CDM), avehicle class model (VCM), and a time of day (TOD) distribution model. The framework also involves other sub-routines including an empty trips model (ETM), conversion of commodities to vehicle trips, and preparing matrices of trips for assignment.  The demand sub-models form the cargo flow model (CFM) and its interactions with the assignment models is illustrated in Figure 1. An important aspect of this model is the CGCM, which will briefly be described below.

Figure 1: Supply-demand model framework

2. Methodology

The CGCM is modelled at the national level and uses commodity flows between states together with land-use and other commodity attributes to capture the chain reactions triggered by the production or consumption of one commodity on other commodities. The main output of this model is the total commodity by type produced/consumed in each state and the evaluated factors governing the generation of these commodities. The outputs are then transformed into logit models to predict the level of production and consumption of each commodity for each freight analysis zone in Greater Sydney as shown in Equation (1):

The quantity of commodity by type k generated by FAZ can be estimated as:

where is the estimated production (or consumption) of commodity type in zone , is the total quantity of commodity type produced in the Greater Sydney Metropolitan Area, is the set of explanatory variables representing land-use and other transport variables describing production or consumption of each commodity type in each zone, and and are estimated parameters of the model.

An illustrative example of key features of this model is shown in Figure 2 showing direct influence of employment in Agriculture (Agr) and indirect influence of employment in manufacturing and population on the production of food (Fod). Food production (Fod) is also influenced by food consumption (CFd) and is also indirectly influenced by cereal production (Crl). This then triggers a chain reaction, with food production influencing beverage consumption (CBv) and live animals consumption (CLv), and CBv in turn influences the production of beverages (Bvr). Detailed discussion of the working of this model and how it was used in the development of the commodity based model will be presented in the paper.

Figure 2: Illustrative chain reactions of commodity productions and consumptions

3. Conclusions

This paper presents a commodity-based model capturing the dynamics in commodity production and consumption, and how changes in the production or consumption of one commodity triggers a chain reaction in the production and/or consumption of other commodities.  The results are then used in a series of linked logit models to explain freight generation and movements and distribution patterns in Greater Sydney. The models are implemented in a fuller model system (called MetroScan-TI) that incorporates a full range of individual decisions, firm location decisions, passenger travel decisions, service vehicle travel decisions  that together provide fully endogenous inputs for applying the commodity and freight models described in this paper.

References

Hensher, D.A., Puckett, S.M.: Refocusing the modeling of freight distribution: development of an economic based framework to evaluate supply chain behavior in response to congestion charging. Transportation 32(6), 573–602 (2005)

Hensher, D., Figliozzi, M.A.: Behavioural insights into the modeling of freight transportation and distribution systems. Transp. Res. B 41(9), 921–923 (2007)

Holguín-Veras, J. & Patil, G.R.: A Multicommodity Integrated Freight Origin-destination Synthesis Model. Netw Spat Econ 8: 309 (2008)

Puckett, S.M., Hensher, D.A., Rose, J.M., Collins, A.: Design and development of a stated choice experiment for interdependent agents: accounting for interactions between buyers and sellers of urban freight services. Transportation 34, 429–451 (2007).

Sivakumar, A., Bhat, C.: Fractional split-distribution model for statewide commodity-flow analysis. Transp. Res. Rec. 1790, 80–88 (2002)

Wisetjindawat, W., Sano, K.: A behavioral modeling in micro-simulation for urban freight transportation. J. East. Asia Soc Transp. Stud. 5, 2193–2208 (2003)

Wisetjindawat, W., Sano, K., Matsumoto, S.: Commodity distribution model incorporating spatial interactions for urban freight movement. Transp. Res. Rec. 1966, 41–50 (2006)


Conference registration is required in order to view papers.