Ports are primary generators of truck traffic in the United States. Seaport operations will require operational and infrastructure changes to maintain the growth of international cargo operations. Truck trip generation models will provide transportation planners and public agencies with valuable information necessary for prioritizing funds for roadway upgrade projects and port infrastructure modifications. A new methodology is presented that combines backpropagation neural networks (BPN) and time series to forecast inbound and outbound heavy truck movements at seaports. The new method uses vessel freight data to identify which parameters are relevant for use as model input in predicting truck traffic at seaports. The method is successfully applied to five ports in Florida--Miami, Tampa, Palm Beach, Jacksonville, and Everglades--thus demonstrating its transferability. Details are provided for the Port of Everglades. The commodities at this port are classified into tons, barrels, and containers. It was found that the primary factors affecting truck traffic are imported containers, imported tonnage, imported barrels, exported containers, exported tonnage, and the particular weekday of operation. Separate BPN models were developed for inbound and outbound truck traffic at the ports. The new method forecasts that the Port of Everglades will have a 33% increase in average daily inbound heavy trucks and a 30% increase in average daily outbound heavy trucks by 2005 (2000 is the base year). The accuracy of the inbound and the outbound truck models is 93% and 92%, respectively.
Samenvatting