Abstract
During the past 35 years, the major railroad companies and the Association of American Railroads all have invested heavily in the development of computerized car-for-car scheduling systems designed to help keep carload rail traffic moving on schedule. Unfortunately, these systems have not worked as intended, because railroad operations do not really lend themselves to handling cars on an individual basis out in the field. The experience at Consolidated Rail Corporation (Conrail) has borne that out. Beyond that, rail markets have shifted steadily to the large, annual use-rate customers that move heavy carload shipments, and those heavy carload markets are less sensitive to transit-time unreliability than the small carload markets that were more typical in the early postwar era. Shipper cost models developed at Transmode and Science Applications International Corporation have been used to analyze the traffic moving in a number of typical carload markets on Conrail during 1998. That analysis shows why unreliable service, if planned for by shippers and receivers, will not unduly increase “nontransport logistics costs.” Thus, it can be concluded from both experience and theory that the railroads should give up trying to use car scheduling to control carload service quality in real time. They should concentrate, instead, on improving car cycle times and reconfiguring shipper supply chains to take advantage of the lower costs associated with large carload shipments.