The scheduling of rail at union pacific railroad

  • Authors:
  • Kathleen Murphy;Elizabeth Ralston;David Friedlander;Rodney Swab;Paul Steege

  • Affiliations:
  • Brightware Incorporated, Novato, California;Brightware Incorporated, Novato, California;Brightware Incorporated, Novato, California;Union Pacific Railroad, Omaha, Nebraska;Union Pacific Railroad, Omaha, Nebraska

  • Venue:
  • AAAI'97/IAAI'97 Proceedings of the fourteenth national conference on artificial intelligence and ninth conference on Innovative applications of artificial intelligence
  • Year:
  • 1997

Quantified Score

Hi-index 0.00

Visualization

Abstract

The Union Pacific Railroad (UPRR) has over 31,000 miles of track covering a 24 state region. Planning and scheduling the production, packaging, delivery, and pickup of rail, involved in the maintenance of this network, is a very complex task. Manually scheduling only a subset of the resources required has historically taken several days to accomplish. Moreover, the inability to fully schedule all resources can lead to inefficient resource utilization. This paper describes the Rail Train Scheduler (RTS), designed and developed to capture the expertise of the UPRR scheduler, generate production schedules of all the resources involved, and provide a decision support tool for determining the best mix of resources required. RTS is an expert system that uses constraint satisfaction and domain specific heuristics to produce good, low cost schedules. It has been deployed since January, 1996. UPRR anticipates a savings of about $500,000 per year from the use of RTS.