The development of measures of service availability

Transit system failures impact system operations and economics in many ways: maintenance requirements, spare parts inventory requirements, reserve fleet requirements, maintenance fleet requirements, and service interruptions. The concept of service and ability treats this latter effect--the service interrupting effects of system failures, as viewed by the passengers. The transit industry seeks to control service interruptions to some level which is acceptable to the passengers. Hence, service availability measures are necessary to guide these control actions and gage their effectiveness. Service availabity has always been a concern in the design and operations of a transit system. However, the direct passenger service orientation was not explicitly utilized. Rather, the emphasis was on equipment performance--specifically equipment reliability and restorability measures. As new transit systems evolved, the experience of existing systems was utilized to establish desired reliability and restorability goals. Implicit in this approach was the assumption that improvements in these two parameters was tantamount to an improvement in service availability. Because of the similarity of applications and technology, this assumption was valid and the quantification of actual passenger-oriented service availablity was unnecessary. AGT systems, however, evolved to meet different service requirements: frequent service, short headways, high equipment utilization, and little tolerance to failure. Furthermore, both applications and competing technologies varied widely. Establishing performance requirements and measuring achieved performance levels requires the explicit treatment of passenger service levels--specifically expected delay patterns. Many models have been proposed to relate passenger delay potential and system failure characteristics. However, these are unsuitable for general use for different technologies and/or different transit system applications. An important conclusion is that no single measure or model will suffice in this role. Rather, each application and technology will require the derivation of specific control measures and values in response to specific passenger delay criteria.