Exploring multiple discreteness in regional freight transport : a behavioural model for grain consolidators in Argentina

Abstract

Freight transport modal choice has traditionally been modelled as a discrete choice problem. This is true in most cases, especially at an operational level. Nevertheless, there are some cases where multiple-discreteness can be considered. Such might be the case for tactical planning at a regional level. This thesis develops a mode-port choice model for grain consolidators in Argentina in order to investigate multiple discreteness in freight regional choices. Traditional discrete choice models (Nested Logit and Fractional split) and a Multiple Discrete Continuous Extreme Value model (MDCEV) were estimated. The models were calibrated with Revealed and Stated Preference data (RP and SP). The processing of consignment bills to create a RP dataset is a contribution for the usage of alternative data sources for choice modelling. SP data was obtained with a provincial and national level survey suggesting that multiple-discreteness exists in the mode-port choice. First models suggest that Fractional Split models have a better forecasting performance than the MDCEV, but they lose in interpretation of the choice situation and econometrical assumptions. The parameters included in the model were travel time, freight cost, free alongside ship (FAS) price and lead time. The results of the implementation of the models for rail infrastructure renovation analysis suggest that the overall benefits (improvements in travel time and lead time) would be higher than the costs.