UC Irvine

Among the main freight modeling approaches, commodity-based models stand out in their ability to incorporate all travel modes and capture the economic mechanisms driving freight movements. However, challenges still exist on the effective use of public freight data and the ability to accurately reflect the supply chain relationships between commodities. In this research, a commodity-based framework for freight demand forecasting using a Structural Regression Model (SRM) is explored, and applied to the original California Statewide Freight Forecasting Model (CSFFM) using the Freight Analysis Framework Version 4 (FAF4) data.

The framework developed in this study contains four innovative components: (1) mathematical approach for determining freight economic centroids; (2) the aggregation of commodities using the Fuzzy C-means clustering algorithm; (3) employing weighted travel distance by commodity group (CG) instead of highway skim to provide a more representative travel distance across multiple modes; and (4) the forecasting of freight demand using SRM method to comprehensively consider the direct effect, indirect effect and latent variables. The SRM is adopted in both the total generation model and domestic direct demand model. The application results are further compared with the original CSFFM forecasts in 2012 to illustrate the advantages of the proposed framework.