Weekly scheduling in the service industry: an application to mail processing and distribution centers

This paper addresses the midterm planning problem that arises at various service facilities staffed by full-time and part-time employees. In response to updated demand forecasts, expected leave, training assignments, and other contingencies, weekly adjustments are often required to better match available personnel with demand over the planning horizon. Unlike manufacturing where uniform 8-hour shifts are the rule, service organizations may experience several busy periods during the day that do not fit a standard shift. In such cases, supervisors must adjust employee schedules by assigning overtime, increasing the number of part-time hours, and calling in temporary workers. The situation is complicated by union contracts, labor rules, and company policies. To find solutions that can be implemented in a real-world environment, a two-phase approach was developed. In the first phase, the adjustment problem is formulated as a large-scale integer program and solved with a commercial code for those cases in which demand increases are limited to no more than 10% above the baseline. For the more general case, a new target-based heuristic was designed with the goal of finding good feasible solutions. In the second phase, the shift schedules are post-processed to provide daily assignments for each worker. An analysis of the problem arising in the first phase is presented for an application involving weekly scheduling at a mail processing and distribution center. The results indicate that high quality solutions can be obtained within a matter of minutes.