Assessment of Empty Trips Models: an Empirical Investigation

Assessment of Empty Trips Models: an Empirical Investigation


C Gonzalez-Calderon, I Sanchez-Diaz, J Holguin-Veras, Rensselaer Polytechnic Institute; E Thorson, The City College of New York, US


The paper conducts a comprehensive analyses of alternative empty trip models to assess which models perform the best. The paper uses five OD surveys conducted in Colombia.


One of the most unique - and more frequently overlooked- aspects of freight transportation is the high number of empty trips that it generates. The number of empty trips is so high that if air is considered a commodity, it would be the commodity most frequently transported. According to the Vehicle and Inventory Use Survey (U.S. Census Bureau, 2004), empty travel accounts for: 56% of the miles traveled by straight truck not pulling a trailer; 58% of the miles traveled by straight truck pulling a trailer, and 33% of the miles traveled by truck tractor (power-unit) pulling trailer(s). As a percentage of the number of trips made, empty travel typically account for about 20% of truck traffic in urban areas (Strauss-Wieder et al., 1989), and about 30-40% in intercity freight (Holguín-Veras and Thorson, 2003a). These numbers clearly indicate their numerical importance. More significant from the modeling point of view is that empty trip flows do not follow the pattern followed by the loaded trips. As a result, trying to compensate for the empty trips by expanding the matrices of loaded trips lead to major errors in the estimation of directional traffic (Holguín-Veras and Thorson, 2003b). Similarly, not accounting for the empties, lead to major errors as well.

The only way to properly account for the empty traffic is to use complementary empty trip models (Noortman and van Es, 1978; Hautzinger, 1984; Holguín-Veras and Thorson, 2003a; Holguín-Veras and Thorson, 2003b; Holguín-Veras et al., 2008). These models try to estimate the flows of empties from commodity flow matrix with the use of simplified tour models. These models have been successfully incorporated in state of the art models in Sweden, Colombia, New York, among others.

The main objective of this research is to contribute to the study of empty trips models via a systematic study of six national freight origin-destination (OD) matrices collected by the Colombi's Ministry of Transportation during the 2000 to 2005 time period. This Freight Origin-Destination Survey (FODS) provides a comprehensive picture of national freight flows in Colombia, and is the only available source of data for the highway modes that carry about 70% of the tonnage of freight transported, which is about the same mode split reported by the Commodity Flow Survey (CFS) in the United States. However, while the CFS is a shipper based survey, the FODS is roadside survey that targets the carriers. The FODS is a sizable data collection program with more than 130,000 surveys every year. It provides useful information for freight transportation planning, though it has not been fully exploited for research purposes.

At the finest level of geographic detail, the survey data are geocoded at the municipal level, which leads to OD matrices with 1,100x1,100 cells. The analyses in this paper for empty trips are based on an aggregation to 28, 36 and 69 zones because in this format the coverage is more complete (electronic versions of the original files at a finer level of detail have only been found for four years). The zoning system used includes internal zones (departments) and external zones (other countries).

Five models (NVE, HVT 1, HVT 2, HVT 3 and HVT 4) were analyzed for the different aggregation of the zones. The results show that the percentage of empty trips is quite stable as it is determined by the asymmetry of the commodity flow matrices that is a very stable feature of the economic system. The results also show the superiority of the empty trip models that use a first order trip chain representation. In all cases, the models HVT lead to lower estimation errors that are in between 11% to 27% lower than the ones produced by the Noortman and Van Es' model. Moreover, the statistical analyses of the parameters of the empty trip models considered (i.e., NVE, HVT) found that they are not time dependent, i.e., they are stable over time.


Association for European Transport