Tour-based Freight Origin-destination Synthesis



Tour-based Freight Origin-destination Synthesis

Authors

C Gonzalez-Calderon,J Holguin-Veras, X Ban, Rensselaer Polytechnic Institute, US

Description

It is proposed an entropy maximization approach to develop tour-based urban freight travel demand models to perform freight Origin-Destination Synthesis including trip chains.

Abstract

For transport planning and management tasks purposes, the modeler needs appropriate basic data for the estimation of future freight transportation demand and supply. One of the most important elements in such a process is the use of freight origin-destination (OD) matrices to represent the travel pattern of truck trips. These freight OD matrices could be estimated from direct samples or from secondary data sources (Holguin-Veras, 2000). OD matrices obtained from the field (from large scale survey such as home, warehouse or roadside interviews) tend to be costly and labor intensive. Moreover, in the case of freight OD matrices the private firms are often unwilling to provide information. Moreover, it has to be take into account that freight flows vary over time, and hence would require repeated surveys. A very attractive option in this regard is the use of traffic counts to estimate OD matrices (also known as OD synthesis) because of the availability and low cost of this observable data. Accordingly, OD synthesis produces an estimate of the trip flow OD matrix that matches secondary data, e.g. link traffic counts. Therefore OD synthesis has the potential of minimizing data collection costs and speeding up the process of model calibration and updating (Van Zuylen and Willumsen, 1980), obtaining freight demand models of good quality at a low cost. One of the unique features of urban commercial vehicle movements is trip chaining behavior. Treated similarly as passenger vehicles, trucks were usually assumed to behave independently trying to minimize the transportation costs. In that case, freight vehicles are assumed to make independent trips rather than tours composed of linked trips (Wang, 2008). Nowadays, it is known that trucks make a significant number of trip chains (tours) and the focus of an investigation of freight movements must be focus on that direction to get more accurate results. The aforementioned formulations about freight OD synthesis have some limitations in their ability to depict freight movements, and they do not model commercial vehicle trip-chains. Based on that, it is necessary to produce a new freight modeling approach considering previous techniques developed for freight demand models to obtain OD matrices from traffic counts, considering trip chains. To bridge the gap, it is proposed to adopt entropy maximization to develop tour-based urban freight travel demand models, given the aggregate information available in the network. This method applied to freight transportation states that when no enough information is available to model the trip chaining behavior of trucks, their individual flows that travel along any tour are assumed to be equally probable. Moreover, the most probable tours can be generated in the greatest number of ways under the known constraints (Wang, 2008; Wang and Holguin-Veras, 2009). The tour-based entropy maximization approach requires a pre-specification of tours potentially visited by trucks and the associated impedances. In this context, a behavioral-based tour choice model will be developed to generate a sufficient and effective set of tours as the input to the entropy maximization formulations in order to perform freight ODS including trip chains. The method will estimate the distribution patterns of commercial vehicle tour flows given link traffic counts or any other easily collected secondary data. For solving the problem, we defined three states for an urban freight system: 1) Micro state: individual commercial vehicle journey starting and ending at a home base (tour flow) by following tour m; 2) Meso state: where tm is the number of commercial vehicle journeys (tour flows) following tour m; and 3) Macro state: Oi is the total number of trips produced by node i (trip production); Dj is the total number of trips attracted to node j (trip attraction), C is the total tour impedance in the commercial network, and Va is the observed truck traffic count in link a. From these states, we can use the entropy formulation as an equivalent minimization program. The objective function indicates that the objective of this problem is to find the most likely ways to distribute tours. The first group of constraints is the trip production/attraction constraint. The second constraint is the total impedance constraint. The third constraint is the observed traffic counts constraint. The last groups of constraints are the nonnegative constraints implying tour flows equal to or greater than zero. The optimal solution found shows that the number of tour flows following a tour is an exponential function of the Lagrange multipliers associated with the trip productions/attractions of nodes along that tour, the tour impedance, and the observed traffic counts in links.

Publisher

Association for European Transport