Improved heuristics for the n-job single-machine weighted tardiness problem

Abstract For the n -job single-machine weighted tardiness problem a priority rule is known, that holds for neighboring jobs in an optimal solution. This rule is exploited as an improvement step for four construction heuristics. Computational results for a large number of test problems up to size 80 and various values of problem parameters show improved results, while the computing times on a standard personal computer remain modest. Scope and purpose In many practical situations it is important to quickly schedule jobs on one machine. There are heuristics known to determine such schedules for the weighted tardiness criterion. The improvement rule produces better schedules at modest extra computer time, i.e., the improved solutions are obtained in about one second on a standard personal computer. The computational results give also insight in which heuristic is best and fastest dependent on parameters of the problem instance to be solved.