Modelling shared vehicle system design and operation using discrete-event simulation technique.

摘要

As an alternative transportation paradigm, shared vehicle systems (SVS) have been implemented in recent years in Europe, North America, Japan, and other countries of the world. SVS systems consist of a fleet of vehicles that are used several times each day by different users. As compared with private automobile, the SVS systems offer a number of advantages. They reduce the number of vehicles, hence parking demand, required to meet total travel demand. Additionally, energy and environmental benefits materialize when low-polluting e.g., electric vehicles (EV) are used.; To date, a number of SVS systems have been implemented and evaluated in North America and around the world. The evaluation findings of these projects revealed the effects of SVS systems on user travel behaviors (i.e., mode choice, commuting habit). These findings and the state-of-the-art of modelling efforts suggest research need to enhance the design and operations of SVS systems with a complete set of design variables. Therefore, the objective of this study is to develop an improved simulation model for the design and operations of SVS systems focused on multiple station and station car. For achieving this objective, discrete-event simulation technique within the application of queuing theory to network framework is adopted. Within this framework, a powerful simulation model has been developed and is implemented in Microsoft Visual C++ environment using CSIM 19. Based on an iterative approach, an efficient and effective design configuration is identified which satisfies a set of measures of performance including user waiting time and number of vehicle relocations.; Analyses showed that SVS systems performance is highly sensitive to vehicle-to-trip ratio and parking-to-vehicle ratio. User waiting time and number of vehicle relocations is found as direct function of vehicle-to-trip and parking-to-vehicle ratios. The SVS systems capacity was found to increase with vehicle-to-trip ratio. Systems with higher demand proved to be served with lower vehicle-to-trip ratio and more economically. The test case study shows that electric vehicles are suitable for serving travel demand requirement of SVS systems with respect to existing battery range.; Analysis with Bayesian probabilistic technique in estimating travel time showed that in the absence of incident related congestion and delay, the SVS systems operation is insensitive to the accuracy gain in travel time for customer wait time and number of relocations. Results showed that the system performance was not significantly affected, at 5% or 10% level, until travel times were 1.75 times the base travel times. (Abstract shortened by UMI.)