Evaluation of route choice set generation methods for Metropolitan retail distribution trips
conference paper
Understanding of freight truck travel behavior in urban transportation network is essential for regional
economy, and choice set generation is the basis for the analysis. This paper uses the planned trip data
and realized GPS trajectory data of trucks to evaluate the performance of choice set generation
algorithms. The difference between planned trips and realized trips is first analyzed based on
descriptive statistics. Then, Monte Carlo Labelling (MCL) method and Accelerated Monte Carlo
(AMC) method are both used to generate route choice sets, and their performance are evaluated and
compared by indicators of coverage rate and consistence index, after controlling the influence of OD
pair characteristics. The results show that truck drivers follow the time scheduling made by the planner
strictly, but there are a lot of uncertainty on travel time during the travel. For the model results, MCL
method and AMC method all have good performance at reproducing observed trips with high
behavioral consistency, but the computation time of AMC is much faster than MCL.
economy, and choice set generation is the basis for the analysis. This paper uses the planned trip data
and realized GPS trajectory data of trucks to evaluate the performance of choice set generation
algorithms. The difference between planned trips and realized trips is first analyzed based on
descriptive statistics. Then, Monte Carlo Labelling (MCL) method and Accelerated Monte Carlo
(AMC) method are both used to generate route choice sets, and their performance are evaluated and
compared by indicators of coverage rate and consistence index, after controlling the influence of OD
pair characteristics. The results show that truck drivers follow the time scheduling made by the planner
strictly, but there are a lot of uncertainty on travel time during the travel. For the model results, MCL
method and AMC method all have good performance at reproducing observed trips with high
behavioral consistency, but the computation time of AMC is much faster than MCL.
Topics
TNO Identifier
981741
Source title
23rd International Conference of Hong Kong Society for Transportation Studies
Pages
1-8
Files
To receive the publication files, please send an e-mail request to TNO Repository.