Weekly scheduling for freight rail engineers & trainmen

Standard practice in the freight rail industry has been used to provide only short-term crew schedules for a day or two at a time. This essentially means that the engineers and other trainmen function on a “just-in-time” basis with as little as two hours notice to report to work. This approach offers the greatest flexibility for management but plays havoc with the work-life balance of the crew. The purpose of this paper is to show that it is possible to construct robust weekly schedules that satisfy the full range of legal regulations and company policies without greatly enlarging the size of the workforce. This is done with a 3-phase algorithm that relies on the logic of column generation and local improvement procedures. Additional features include the option to generate cyclic schedules and a parametric approach to account for random trip times. To demonstrate the effectiveness of the methodology, computational experiments and statistical analysis are conducted using data sets from a Class I railroad. The robustness of the derived schedules is confirmed through simulation with varied parameter settings by comparing each of the baseline schedules produced by the 3-phase algorithm with 100 random instances.