Transit assignment has become an important component of transportation modelling as planners seek to evaluate the passenger demand and fare revenues for improvements to existing and new transit schemes. Transit assignment attempts to solve a complicated problem, with complexities arising from both the transit network definition as well as the human decision-making process. Acknowledged shortcomings in commonly used assignment approaches have been highlighted in the literature. The application of universal traveller decision rules in transit assignment are not necessarily compatible with traveller preference and choice behaviour, especially with the advent of smartphone-based, real-time transit operations information. This paper investigates the potential for transit line patronage distortion in South African metropolitan areas when using standard frequency-based transit assignment based on optimal strategies. These distortions arise from the dense transit networks in metropolitan areas, the high number of competing lines, as well as the behavioural assumptions made in the process. The potential distortions in assigned transit line patronage volumes make link-based passenger volume validation difficult to achieve. Moreover, the use of these models to forecast the demand for new transit schemes could result in misleading estimates of demand and fare revenue. This paper suggests that more advanced transit assignment techniques are required to more accurately reflect traveller choice and demand in dense transit networks. It is proposed that a logit-based trip allocation approach in transit assignment is necessary to achieve satisfactory assignment validation.
Papers presented at the 36th Southern African Transport Conference, CSIR International Convention Centre, Pretoria, South Africa on 10-13 July 2017.