Two alternative methods for conducting a preliminary analysis of the problem of locating and sizing park-ride lots for an urban bus transit system are examined. These alternative methods are tested using a large and complex street network representing the northern half of the City of Seattle, Washington. Four design criteria are identified and an overall measure of system performance is devised. The problem is to determine the size and location of three park-ride lots that will maximize the performance of the system. The first method utilizes a computing system called LOCATOR II in an interactive graphic mode. In an experimental setting, five students use LOCATOR II to locate three park-ride lots and their solutions are found to exhibit a very high average level of performance. The second method does not involve a direct interaction with the computing system but does use LOCATOR II in a batch mode. Several students participate in three rounds of the batch mode experiment and the performance of their designs is somewhat poorer than the designs found in the interactive graphics experiment. Overall, the experimental results show that relatively inexperienced persons can find a high performance design for a reasonably complex problem quickly and inexpensively. These designs can provide a good starting point for the detailed consideration of many of the other aspects of the problem of providing a high performance bus transit service in an urban area.
Read full abstract