Multimodal pricing and the optimal design of bus services: new elements and extensions

Abstract

This thesis analyses the pricing and design of urban transport systems; in particular the optimal design and efficient operation of bus services and the pricing of urban transport. Five main topics are addressed: (i) the influence of considering non-motorised travel alternatives (walking and cycling) in the estimation of optimal bus fares, (ii) the choice of a fare collection system and bus boarding policy, (iii) the influence of passengers’ crowding on bus operations and optimal supply levels, (iv) the optimal investment in road infrastructure for buses, which is attached to a target bus running speed and (v) the characterisation of bus congestion and its impact on bus operation and service design. Total cost minimisation and social welfare maximisation models are developed, which are complemented by the empirical estimation of bus travel times. As bus patronage increases, it is efficient to invest money in speeding up boarding and alighting times. Once on-board cash payment has been ruled out, allowing boarding at all doors is more important as a tool to reduce both users and operator costs than technological improvements on fare collection. The consideration of crowding externalities (in respect of both seating and standing) imposes a higher optimal bus fare, and consequently, a reduction of the optimal bus subsidy. Optimal bus frequency is quite sensitive to the assumptions regarding crowding costs, impact of buses on traffic congestion and congestion level in mixed-traffic roads. The existence of a crowding externality implies that buses should have as many seats as possible, up to a minimum area that must be left free of seats. Bus congestion in the form of queuing delays behind bus stops is estimated using simulation. The delay function depends on the bus frequency, bus size, number of berths and dwell time. Therefore, models that use flow measures (including frequency only or frequency plus traffic flow) as the only explanatory variables for bus congestion are incomplete. Disregarding bus congestion in the design of the service would yield greater frequencies than optimal when congestion is noticeable, i.e. for high demand. Finally, the optimal investment in road infrastructure for buses grows with the logarithm of demand; this result depends on the existence of a positive and linear relationship between investment in infrastructure and desired running speed.