Agent-based optimisation of public transport supply and pricing: impacts of activity scheduling decisions and simulation randomness

The optimal setting of public transport pricing and supply levels has been traditionally analysed with analytical models that combine the objectives of users, service providers and decision makers in optimisation problems. In this paper, public transport fare and headway are jointly optimised using an activity-based simulation framework. Unlike traditional analytical models that find single optimal values for headway, fare and other optimisation variables, we obtain a range of values for the optimal fare and headway, due to the randomness in user behaviour that is inherent to an agent-based approach. Waiting times and implications of an active bus capacity constraint are obtained on an agent-by-agent basis. The maximisation of operator profit or social welfare result in different combinations of the most likely optimal headway and fare. We show that the gap between welfare and profit optimal solutions is smaller when users can adjust their departure time according to their activities, timetabling and convenience of the public transport service.