TY - JOUR
T1 - Optimal Sampling Plan for Freight Demand Synthesis with Mode Choice
T2 - A Case study of Bangladesh
AU - Kalahasthi, Lokesh Kumar
AU - Sutar, Pranav
AU - Yushimito, Wilfredo F.
AU - Holguín-Veras, José
N1 - Publisher Copyright:
© National Academy of Sciences: Transportation Research Board 2023.
PY - 2023/12
Y1 - 2023/12
N2 - This paper uses a comprehensive experimental design to investigate the influence of various traffic count sampling plans for Bangladesh on the performance of the Freight Origin-Destination Synthesis model with Mode Choice (FODS-MC) developed by Kalahasthi et al. FODS-MC estimates a national-level freight demand model including trip distribution, mode choice, empty truck trips, and empty rail trips, where one of the key inputs is the freight truck and rail, traffic counts. The traffic count sample comprises three types of road links (national, regional, and zilla) and one category for the rail link across the country. A Box–Behnken Design (BBD) with a response surface for each of four FODS-MC parameters (distribution, mode choice, truck empty trips, and rail empty trips) is constructed. The results showed that the response surfaces are nonlinear planes for all parameters. There is no single optimal sampling plan for FODS-MC as each model parameter demands different distribution across the truck and rail links. The random and stratified samples perform almost similarly if less than 20% of the sample is collected. Minimizing the loss functions between the estimated and true parameters shows that a random sample between 20% and 25% of the truck and rail links estimates the best freight demand model. Overall, this research develops a framework to assist public practitioners in the optimum usage of the limited time and resources in collecting the traffic count data that could estimate the freight demand and mode choice models effectively.
AB - This paper uses a comprehensive experimental design to investigate the influence of various traffic count sampling plans for Bangladesh on the performance of the Freight Origin-Destination Synthesis model with Mode Choice (FODS-MC) developed by Kalahasthi et al. FODS-MC estimates a national-level freight demand model including trip distribution, mode choice, empty truck trips, and empty rail trips, where one of the key inputs is the freight truck and rail, traffic counts. The traffic count sample comprises three types of road links (national, regional, and zilla) and one category for the rail link across the country. A Box–Behnken Design (BBD) with a response surface for each of four FODS-MC parameters (distribution, mode choice, truck empty trips, and rail empty trips) is constructed. The results showed that the response surfaces are nonlinear planes for all parameters. There is no single optimal sampling plan for FODS-MC as each model parameter demands different distribution across the truck and rail links. The random and stratified samples perform almost similarly if less than 20% of the sample is collected. Minimizing the loss functions between the estimated and true parameters shows that a random sample between 20% and 25% of the truck and rail links estimates the best freight demand model. Overall, this research develops a framework to assist public practitioners in the optimum usage of the limited time and resources in collecting the traffic count data that could estimate the freight demand and mode choice models effectively.
KW - freight model/modeling
KW - freight systems
KW - general
UR - http://www.scopus.com/inward/record.url?scp=85162692979&partnerID=8YFLogxK
U2 - 10.1177/03611981231169527
DO - 10.1177/03611981231169527
M3 - Article
AN - SCOPUS:85162692979
SN - 0361-1981
VL - 2677
SP - 718
EP - 730
JO - Transportation Research Record
JF - Transportation Research Record
IS - 12
ER -