| Schedule synchronization may significantly reduce transfer delays at terminals where various routes interconnect. Since vehicle arrivals are stochastic, slack time allowances in vehicle schedules may be desirable to reduce the probability of missed connections, even at the cost of some additional waiting times for vehicles, drivers, and some non-transferring passengers. Total system cost functions are formulated here to assess the effectiveness of coordinated operations under various demand and traffic conditions. Numerical integration is used to compute transfer delays due to probabilistic vehicle arrival distributions. Afterwards, a newly developed headway clustering algorithm is used to jointly optimize headways and slack times for all coordinated routes. The results show that as demand decreases, increased coordination becomes desirable. They also show, for routes with widely differing characteristics, the value of coordination with integer-ratio headways rather than a single common headway. The results from sensitivity analysis also show that coordination is not worth attempting, even in low demand and high headway cases, if arrival randomness becomes excessive. The variance of vehicle arrivals turns out to have different effects on the optimal headways for the uncoordinated and coordinated operations. For uncoordinated operation, the independently optimal headway increases as arrival variance increases, since wait times are increasingly influenced by variances as well as mean headways. For coordinated operation, within some level of arrival variance, the additional uncertainty provides economic justification for a smaller headway. However, schedule coordination is not worth attempting when standard deviations of vehicle arrivals become large fractions of the headways. Then, the optimal headways increase with arrival variances, as for uncoordinated operation.; For real-time dispatching control, this study optimizes the holding time for each ready vehicle based on predicted arrival delays of late vehicles and other factors such as expected transfer volumes and vehicle operating costs. Holding times for each ready vehicle are also optimized with the proposed numerical approach. That approach evaluates the dispatching decision at frequent intervals for ready vehicles by evaluating a dispatching objective function. That function is computed by numerically integrating relevant probability distributions. The numerical results can provide general dispatching guidelines. However, the dispatching algorithms are efficient enough to be used in real- time for each particular decision.; Pre-planned slack times and real-time dispatching, either separately or jointly, improve the transit systems analyzed here. The joint use of these two options achieves the lowest average costs. The results also show that, even in the absence of the hardware and software for real-time dispatching, optimized slack times can significantly improve timed transfers. |
