Scheduling Process at GM
...s spend at the facility to complete their required tests. In addition, to improve the throughput, the staff must schedule tests effectively and efficiently. To reduce the time vehicles spend at the Cold-Weather Test Facility, GM’s management decide to prepare a daily master schedule were the schedulers assigns test to vehicles and arrange enough drivers to perform the tests. Management’s objective in scheduling is to maximize the overall efficiency of vehicle testing (the percentage of time vehicles are being tested). Sometime ago, scheduling procedures at GM were done manually through trial and error often with suboptimal results. Furthermore, schedulers took hours to prepare each day’s schedule and therefore based schedules on forecasted temperatures and estimated vehicle availability. These estimates were often inaccurate. For example, in the 2000-2001 test season, forecast for temperatures of -30° and -20° were accurate only 30 and 50 percent of the time. The resulting inefficient schedules contributed to long vehicle turnaround times and vehicles failing to complete required test during the test season. As GM management strove to improve the throughput of the durability test operations, if focused on improving scheduling. Its primary goal was to optimize daily scheduling quickly so that schedulers could issue the master schedule at the start of each day’s operations (rather than on the prior day) when they had accurate and up-to-the minute information on temperature, vehicle availability, and drivers. Initially engineers try to develop a program using CPLEX, a commercial mathematical programming solver. However, scheduling using this type of program was requiring an impractical amount of computation time. Engineers decide to develop a customized program that uses a weighted-sampling priority-rule method. With this program, at each stage, the scheme includes all vehicles ready for their next test assignment in the decision set as candidates. Engineers use a formula considering some variables to identify which vehicle needs additional tests. For the implementation of this system management developed a decision-support tool that consists of the front-end user interface and the scheduling engine, which were linked to the corporate test-tracking database. This project began on July 2001 and after four months the first version was ready. On 2001-2002 test season the program...