This study is an operational research approach towards the effective design of advanced ground transportation systems. To this effect, a stochastic model was constructed over a network with M stations connected arbitrarily by routes and the relationships between parameters were simulated by this model.
The model is based on two important postulates. 1) the assumption of complete randomness of passenger demand (homogeneous or non-homogeneous Poisson process) 2) the adoption of a dynamic vehicle dispatching policy. “Dynamic” is a term used here in contrast to the term “fixed”. Fixed dispatching policy is a policy which schedules vehicles according to predetermined time-table. Dynamic dispatching policy is a strategy to dispatch vehicles without any predetermined time-table, dynamically responding to the continuous change of passenger demand.
Although there are many conceivable kinds of dynamic dispatching policies, two of them, gc-when-fill policy and go-when-fill-with-time-constraint policy were studied.
First, the dynamic dispatching policy which plays a central role in this study was examined, and the probability density function of vehicle departure interval, average passenger waiting time, and average vehicle loading factor were derived. Next, the stochastic behaviours of vehicle flows in the network such as distribution of vehicles as the function of time in the network were examined for the vehicles flows dispatched by the dynamic dispatching policy. Lastly, it was studied how to control the distribution of vehicles over the network to realize a smooth operation of the systems.