Characterising and Modelling Urban Freight in Developing Economies

Abstract

Urban freight systems in developing countries present significant challenges due to their complexity. Authorities often have inadequate institutional structures, making it difficult to identify and implement relevant initiatives. This thesis aims to characterise the systems in developing economies and model freight demand using innovative approaches by considering new attributes, dimensions and alternatives. As a first modelling step, freight (trip) generation was improved by considering spatial and locational determinants, as freight activities are strongly related to spatial and locational characteristics of establishments. Spatial models were developed using a combined spatial autoregressive model (SAR) and geographically weighted regression (GWR) or multiscale GWR (MGWR) (GWR/MGWR-SAR model). This model accounted for non-linearity, spatial heterogeneity and spatial dependency and demonstrated significant improvements (R2 0.29-0.71, RMSE reduced by 71% and AIC value by 56%). Shipment size decisions related to the choice of truck type were strongly timedependent, with commodity type, activities at the trip end, truck body type and industry sector affecting the preferences. Freight demand, including shipment size choices, was influenced by economic fluctuations, with shipment size declining after an economic slowdown. In freight demand modelling, it is imperative to consider economic conditions, especially those in developing countries, which are often susceptible to strong economic fluctuations. The models were applied in ex ante testing of a policy restricting large trucks from entering a city centre, as commonly considered in many developing countries. In tests, the truck restriction was accompanied by single-tier and two-tier distribution systems. The results showed that the two-tier system had a slight advantage over the single-tier system regarding operational expenditure and emission levels. Truck restriction was generally counterproductive, even when accompanied by distribution systems with greater speed and efficiency. We conclude that the models enhance the accurate prediction of freight demand patterns. The ex ante evaluation of policy alternatives supports the decision-making process for urban freight systems of large cities in developing economies. The models allow considering relevant practical, local contextual conditions.