Performances of Simulated Human Planner and Computer in Solving Sequence-Dependent Setup Time Problems

In a sequence-dependent setup time problem, the total setup time of jobs is minimized by taking into account how the setup time for a job on a machine depends on the previous job. The paper compares the solving of a sequence-dependent setup time problem by a human and a computer. The nearest neighbour heuristic simulates the human planner and a heuristic improvement procedure based on the 2-OPT method represents solving the problem using a computer. In numerical experiments, the two methods are compared in the cases of five different types of setup time matrices. These setup time matrix types are random setup times, symmetric setup times, long setup times, product group-based setup times, and tool-based setup times. The results generally show how a human operator can solve the setup time problem well in the cases of tool-based and product group-based setup times. Non-symmetric setup times, especially the case with random long setups, are hard for a human planner.