Calibrating Freight Value‐to‐Tonnage Conversion Factors for Input–Output–Based Integrated Land Use Transport Models Using an Iterative Feedback Approach

A review of the existing literature reveals a lack of systematic methodologies for estimating and calibrating value‐to‐tonnage conversion factors (VTTCFs), which are essential for freight demand modeling within integrated land use transportation models based on input–output theories and techniques (IO‐based ILUTMs). To address this gap, this study proposes an iterative feedback–based approach for the estimation and calibration of freight VTTCFs. Firstly, a production, exchange, and consumption allocation system (PECAS) model is developed to estimate commodity‐specific origin–destination (OD) matrices in monetary units. These economic flows are initially converted into freight OD matrices in physical units (i.e., tonnage) using preliminary VTTCFs (i.e., the median price of imported/exported commodities from the Customs in value/ton). Subsequently, a multimodal transportation model is developed to simulate freight volumes on highway, railway, and waterway networks. An iterative feedback calibration procedure is then proposed and implemented to adjust VTTCFs as well as mode‐specific and link‐specific alternative specific constants (ASCs), subject to multiple consistency conditions. Specifically, the modeled freight volumes are required to closely match the observed data in terms of (1) total freight volume shipped within the modeling area (ton/day), (2) freight volume shipped by transport mode (ton/day), and (3) freight volume on specific network links (ton/day). In addition, the calibrated VTTCFs are constrained to remain within commodity‐specific price ranges statistically defined using the price data of imported/exported goods from the Customs. Case study results indicate that the absolute percentage error (APE) of the estimated total freight volume shipped in tonnage of the entire modeled area is 3.33%. The corresponding APEs for highway, railway, and waterway modes are 3.59%, 1.44%, and 0.47%, respectively. At the link level, the mean absolute percentage errors (MAPEs) for highway, railway, and waterway segments with observed data are 6.88%, 16.19%, and 10.68%, respectively. These results demonstrate that the proposed approach can effectively reconcile discrepancies between the estimated economic flows in the monetary units from the land‐use module and the observed freight flows in tonnage from a corresponding multimodal transportation module of an IO‐based ILUTM. Accordingly, this study offers a practical and systematic tool for estimating/calibrating freight VTTCFs for IO‐based ILUTMs.